WO2021061949A1 - Fabricating active-bridge-coupled gpu chiplets - Google Patents

Fabricating active-bridge-coupled gpu chiplets Download PDF

Info

Publication number
WO2021061949A1
WO2021061949A1 PCT/US2020/052444 US2020052444W WO2021061949A1 WO 2021061949 A1 WO2021061949 A1 WO 2021061949A1 US 2020052444 W US2020052444 W US 2020052444W WO 2021061949 A1 WO2021061949 A1 WO 2021061949A1
Authority
WO
WIPO (PCT)
Prior art keywords
gpu
chiplet
chiplets
carrier wafer
bridge
Prior art date
Application number
PCT/US2020/052444
Other languages
French (fr)
Inventor
Skyler J. Saleh
Ruijin WU
Milind S. Bhagavat
Rahul Agarwal
Original Assignee
Advanced Micro Devices, Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Advanced Micro Devices, Inc. filed Critical Advanced Micro Devices, Inc.
Priority to EP20867435.8A priority Critical patent/EP4035019A4/en
Priority to KR1020227011449A priority patent/KR20220072838A/en
Priority to JP2022517359A priority patent/JP2022549787A/en
Priority to CN202080067204.0A priority patent/CN114467166A/en
Publication of WO2021061949A1 publication Critical patent/WO2021061949A1/en

Links

Classifications

    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/02—Manufacture or treatment of semiconductor devices or of parts thereof
    • H01L21/04—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer
    • H01L21/50—Assembly of semiconductor devices using processes or apparatus not provided for in a single one of the subgroups H01L21/06 - H01L21/326, e.g. sealing of a cap to a base of a container
    • H01L21/56—Encapsulations, e.g. encapsulation layers, coatings
    • H01L21/568—Temporary substrate used as encapsulation process aid
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F8/00—Arrangements for software engineering
    • G06F8/40—Transformation of program code
    • G06F8/41—Compilation
    • G06F8/45—Exploiting coarse grain parallelism in compilation, i.e. parallelism between groups of instructions
    • G06F8/451—Code distribution
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/02—Manufacture or treatment of semiconductor devices or of parts thereof
    • H01L21/04—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer
    • H01L21/18—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer the devices having semiconductor bodies comprising elements of Group IV of the Periodic Table or AIIIBV compounds with or without impurities, e.g. doping materials
    • H01L21/30—Treatment of semiconductor bodies using processes or apparatus not provided for in groups H01L21/20 - H01L21/26
    • H01L21/302—Treatment of semiconductor bodies using processes or apparatus not provided for in groups H01L21/20 - H01L21/26 to change their surface-physical characteristics or shape, e.g. etching, polishing, cutting
    • H01L21/304—Mechanical treatment, e.g. grinding, polishing, cutting
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/02—Manufacture or treatment of semiconductor devices or of parts thereof
    • H01L21/04—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer
    • H01L21/48—Manufacture or treatment of parts, e.g. containers, prior to assembly of the devices, using processes not provided for in a single one of the subgroups H01L21/06 - H01L21/326
    • H01L21/4814—Conductive parts
    • H01L21/4846—Leads on or in insulating or insulated substrates, e.g. metallisation
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/02—Manufacture or treatment of semiconductor devices or of parts thereof
    • H01L21/04—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer
    • H01L21/50—Assembly of semiconductor devices using processes or apparatus not provided for in a single one of the subgroups H01L21/06 - H01L21/326, e.g. sealing of a cap to a base of a container
    • H01L21/52—Mounting semiconductor bodies in containers
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/02—Manufacture or treatment of semiconductor devices or of parts thereof
    • H01L21/04—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer
    • H01L21/50—Assembly of semiconductor devices using processes or apparatus not provided for in a single one of the subgroups H01L21/06 - H01L21/326, e.g. sealing of a cap to a base of a container
    • H01L21/56—Encapsulations, e.g. encapsulation layers, coatings
    • H01L21/563—Encapsulation of active face of flip-chip device, e.g. underfilling or underencapsulation of flip-chip, encapsulation preform on chip or mounting substrate
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/67—Apparatus specially adapted for handling semiconductor or electric solid state devices during manufacture or treatment thereof; Apparatus specially adapted for handling wafers during manufacture or treatment of semiconductor or electric solid state devices or components ; Apparatus not specifically provided for elsewhere
    • H01L21/683—Apparatus specially adapted for handling semiconductor or electric solid state devices during manufacture or treatment thereof; Apparatus specially adapted for handling wafers during manufacture or treatment of semiconductor or electric solid state devices or components ; Apparatus not specifically provided for elsewhere for supporting or gripping
    • H01L21/6835—Apparatus specially adapted for handling semiconductor or electric solid state devices during manufacture or treatment thereof; Apparatus specially adapted for handling wafers during manufacture or treatment of semiconductor or electric solid state devices or components ; Apparatus not specifically provided for elsewhere for supporting or gripping using temporarily an auxiliary support
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L23/00—Details of semiconductor or other solid state devices
    • H01L23/48—Arrangements for conducting electric current to or from the solid state body in operation, e.g. leads, terminal arrangements ; Selection of materials therefor
    • H01L23/481—Internal lead connections, e.g. via connections, feedthrough structures
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L23/00—Details of semiconductor or other solid state devices
    • H01L23/52—Arrangements for conducting electric current within the device in operation from one component to another, i.e. interconnections, e.g. wires, lead frames
    • H01L23/538—Arrangements for conducting electric current within the device in operation from one component to another, i.e. interconnections, e.g. wires, lead frames the interconnection structure between a plurality of semiconductor chips being formed on, or in, insulating substrates
    • H01L23/5381—Crossover interconnections, e.g. bridge stepovers
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L23/00—Details of semiconductor or other solid state devices
    • H01L23/52—Arrangements for conducting electric current within the device in operation from one component to another, i.e. interconnections, e.g. wires, lead frames
    • H01L23/538—Arrangements for conducting electric current within the device in operation from one component to another, i.e. interconnections, e.g. wires, lead frames the interconnection structure between a plurality of semiconductor chips being formed on, or in, insulating substrates
    • H01L23/5386—Geometry or layout of the interconnection structure
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L23/00—Details of semiconductor or other solid state devices
    • H01L23/52—Arrangements for conducting electric current within the device in operation from one component to another, i.e. interconnections, e.g. wires, lead frames
    • H01L23/538—Arrangements for conducting electric current within the device in operation from one component to another, i.e. interconnections, e.g. wires, lead frames the interconnection structure between a plurality of semiconductor chips being formed on, or in, insulating substrates
    • H01L23/5389—Arrangements for conducting electric current within the device in operation from one component to another, i.e. interconnections, e.g. wires, lead frames the interconnection structure between a plurality of semiconductor chips being formed on, or in, insulating substrates the chips being integrally enclosed by the interconnect and support structures
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L24/00—Arrangements for connecting or disconnecting semiconductor or solid-state bodies; Methods or apparatus related thereto
    • H01L24/01—Means for bonding being attached to, or being formed on, the surface to be connected, e.g. chip-to-package, die-attach, "first-level" interconnects; Manufacturing methods related thereto
    • H01L24/02—Bonding areas ; Manufacturing methods related thereto
    • H01L24/03—Manufacturing methods
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L24/00—Arrangements for connecting or disconnecting semiconductor or solid-state bodies; Methods or apparatus related thereto
    • H01L24/01—Means for bonding being attached to, or being formed on, the surface to be connected, e.g. chip-to-package, die-attach, "first-level" interconnects; Manufacturing methods related thereto
    • H01L24/18—High density interconnect [HDI] connectors; Manufacturing methods related thereto
    • H01L24/19—Manufacturing methods of high density interconnect preforms
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L24/00—Arrangements for connecting or disconnecting semiconductor or solid-state bodies; Methods or apparatus related thereto
    • H01L24/93—Batch processes
    • H01L24/95—Batch processes at chip-level, i.e. with connecting carried out on a plurality of singulated devices, i.e. on diced chips
    • H01L24/96—Batch processes at chip-level, i.e. with connecting carried out on a plurality of singulated devices, i.e. on diced chips the devices being encapsulated in a common layer, e.g. neo-wafer or pseudo-wafer, said common layer being separable into individual assemblies after connecting
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L24/00—Arrangements for connecting or disconnecting semiconductor or solid-state bodies; Methods or apparatus related thereto
    • H01L24/93—Batch processes
    • H01L24/95—Batch processes at chip-level, i.e. with connecting carried out on a plurality of singulated devices, i.e. on diced chips
    • H01L24/97—Batch processes at chip-level, i.e. with connecting carried out on a plurality of singulated devices, i.e. on diced chips the devices being connected to a common substrate, e.g. interposer, said common substrate being separable into individual assemblies after connecting
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L25/00—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof
    • H01L25/03—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof all the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/00, or in a single subclass of H10K, H10N, e.g. assemblies of rectifier diodes
    • H01L25/04—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof all the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/00, or in a single subclass of H10K, H10N, e.g. assemblies of rectifier diodes the devices not having separate containers
    • H01L25/065—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof all the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/00, or in a single subclass of H10K, H10N, e.g. assemblies of rectifier diodes the devices not having separate containers the devices being of a type provided for in group H01L27/00
    • H01L25/0652—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof all the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/00, or in a single subclass of H10K, H10N, e.g. assemblies of rectifier diodes the devices not having separate containers the devices being of a type provided for in group H01L27/00 the devices being arranged next and on each other, i.e. mixed assemblies
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L25/00—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof
    • H01L25/03—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof all the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/00, or in a single subclass of H10K, H10N, e.g. assemblies of rectifier diodes
    • H01L25/04—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof all the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/00, or in a single subclass of H10K, H10N, e.g. assemblies of rectifier diodes the devices not having separate containers
    • H01L25/065—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof all the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/00, or in a single subclass of H10K, H10N, e.g. assemblies of rectifier diodes the devices not having separate containers the devices being of a type provided for in group H01L27/00
    • H01L25/0657—Stacked arrangements of devices
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L25/00—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof
    • H01L25/18—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof the devices being of types provided for in two or more different subgroups of the same main group of groups H01L27/00 - H01L33/00, or in a single subclass of H10K, H10N
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L25/00—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof
    • H01L25/50—Multistep manufacturing processes of assemblies consisting of devices, each device being of a type provided for in group H01L27/00 or H01L29/00
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/02—Manufacture or treatment of semiconductor devices or of parts thereof
    • H01L21/04—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer
    • H01L21/50—Assembly of semiconductor devices using processes or apparatus not provided for in a single one of the subgroups H01L21/06 - H01L21/326, e.g. sealing of a cap to a base of a container
    • H01L21/60—Attaching or detaching leads or other conductive members, to be used for carrying current to or from the device in operation
    • H01L2021/60007—Attaching or detaching leads or other conductive members, to be used for carrying current to or from the device in operation involving a soldering or an alloying process
    • H01L2021/60022—Attaching or detaching leads or other conductive members, to be used for carrying current to or from the device in operation involving a soldering or an alloying process using bump connectors, e.g. for flip chip mounting
    • H01L2021/6006—Attaching or detaching leads or other conductive members, to be used for carrying current to or from the device in operation involving a soldering or an alloying process using bump connectors, e.g. for flip chip mounting with temporary supporting member not part of an apparatus, e.g. removable coating, film or substrate
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2221/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof covered by H01L21/00
    • H01L2221/67—Apparatus for handling semiconductor or electric solid state devices during manufacture or treatment thereof; Apparatus for handling wafers during manufacture or treatment of semiconductor or electric solid state devices or components; Apparatus not specifically provided for elsewhere
    • H01L2221/683—Apparatus for handling semiconductor or electric solid state devices during manufacture or treatment thereof; Apparatus for handling wafers during manufacture or treatment of semiconductor or electric solid state devices or components; Apparatus not specifically provided for elsewhere for supporting or gripping
    • H01L2221/68304—Apparatus for handling semiconductor or electric solid state devices during manufacture or treatment thereof; Apparatus for handling wafers during manufacture or treatment of semiconductor or electric solid state devices or components; Apparatus not specifically provided for elsewhere for supporting or gripping using temporarily an auxiliary support
    • H01L2221/68372—Apparatus for handling semiconductor or electric solid state devices during manufacture or treatment thereof; Apparatus for handling wafers during manufacture or treatment of semiconductor or electric solid state devices or components; Apparatus not specifically provided for elsewhere for supporting or gripping using temporarily an auxiliary support used to support a device or wafer when forming electrical connections thereto
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2224/00—Indexing scheme for arrangements for connecting or disconnecting semiconductor or solid-state bodies and methods related thereto as covered by H01L24/00
    • H01L2224/01—Means for bonding being attached to, or being formed on, the surface to be connected, e.g. chip-to-package, die-attach, "first-level" interconnects; Manufacturing methods related thereto
    • H01L2224/02—Bonding areas; Manufacturing methods related thereto
    • H01L2224/023—Redistribution layers [RDL] for bonding areas
    • H01L2224/0231—Manufacturing methods of the redistribution layers
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2224/00—Indexing scheme for arrangements for connecting or disconnecting semiconductor or solid-state bodies and methods related thereto as covered by H01L24/00
    • H01L2224/01—Means for bonding being attached to, or being formed on, the surface to be connected, e.g. chip-to-package, die-attach, "first-level" interconnects; Manufacturing methods related thereto
    • H01L2224/02—Bonding areas; Manufacturing methods related thereto
    • H01L2224/04—Structure, shape, material or disposition of the bonding areas prior to the connecting process
    • H01L2224/04105—Bonding areas formed on an encapsulation of the semiconductor or solid-state body, e.g. bonding areas on chip-scale packages
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2224/00—Indexing scheme for arrangements for connecting or disconnecting semiconductor or solid-state bodies and methods related thereto as covered by H01L24/00
    • H01L2224/01—Means for bonding being attached to, or being formed on, the surface to be connected, e.g. chip-to-package, die-attach, "first-level" interconnects; Manufacturing methods related thereto
    • H01L2224/02—Bonding areas; Manufacturing methods related thereto
    • H01L2224/07—Structure, shape, material or disposition of the bonding areas after the connecting process
    • H01L2224/08—Structure, shape, material or disposition of the bonding areas after the connecting process of an individual bonding area
    • H01L2224/081—Disposition
    • H01L2224/0812—Disposition the bonding area connecting directly to another bonding area, i.e. connectorless bonding, e.g. bumpless bonding
    • H01L2224/08135—Disposition the bonding area connecting directly to another bonding area, i.e. connectorless bonding, e.g. bumpless bonding the bonding area connecting between different semiconductor or solid-state bodies, i.e. chip-to-chip
    • H01L2224/08145—Disposition the bonding area connecting directly to another bonding area, i.e. connectorless bonding, e.g. bumpless bonding the bonding area connecting between different semiconductor or solid-state bodies, i.e. chip-to-chip the bodies being stacked
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2224/00—Indexing scheme for arrangements for connecting or disconnecting semiconductor or solid-state bodies and methods related thereto as covered by H01L24/00
    • H01L2224/01—Means for bonding being attached to, or being formed on, the surface to be connected, e.g. chip-to-package, die-attach, "first-level" interconnects; Manufacturing methods related thereto
    • H01L2224/10—Bump connectors; Manufacturing methods related thereto
    • H01L2224/11—Manufacturing methods
    • H01L2224/11001—Involving a temporary auxiliary member not forming part of the manufacturing apparatus, e.g. removable or sacrificial coating, film or substrate
    • H01L2224/11002—Involving a temporary auxiliary member not forming part of the manufacturing apparatus, e.g. removable or sacrificial coating, film or substrate for supporting the semiconductor or solid-state body
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2224/00—Indexing scheme for arrangements for connecting or disconnecting semiconductor or solid-state bodies and methods related thereto as covered by H01L24/00
    • H01L2224/01—Means for bonding being attached to, or being formed on, the surface to be connected, e.g. chip-to-package, die-attach, "first-level" interconnects; Manufacturing methods related thereto
    • H01L2224/10—Bump connectors; Manufacturing methods related thereto
    • H01L2224/12—Structure, shape, material or disposition of the bump connectors prior to the connecting process
    • H01L2224/12105—Bump connectors formed on an encapsulation of the semiconductor or solid-state body, e.g. bumps on chip-scale packages
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2224/00—Indexing scheme for arrangements for connecting or disconnecting semiconductor or solid-state bodies and methods related thereto as covered by H01L24/00
    • H01L2224/01—Means for bonding being attached to, or being formed on, the surface to be connected, e.g. chip-to-package, die-attach, "first-level" interconnects; Manufacturing methods related thereto
    • H01L2224/26—Layer connectors, e.g. plate connectors, solder or adhesive layers; Manufacturing methods related thereto
    • H01L2224/31—Structure, shape, material or disposition of the layer connectors after the connecting process
    • H01L2224/32—Structure, shape, material or disposition of the layer connectors after the connecting process of an individual layer connector
    • H01L2224/321—Disposition
    • H01L2224/32151—Disposition the layer connector connecting between a semiconductor or solid-state body and an item not being a semiconductor or solid-state body, e.g. chip-to-substrate, chip-to-passive
    • H01L2224/32221—Disposition the layer connector connecting between a semiconductor or solid-state body and an item not being a semiconductor or solid-state body, e.g. chip-to-substrate, chip-to-passive the body and the item being stacked
    • H01L2224/32225—Disposition the layer connector connecting between a semiconductor or solid-state body and an item not being a semiconductor or solid-state body, e.g. chip-to-substrate, chip-to-passive the body and the item being stacked the item being non-metallic, e.g. insulating substrate with or without metallisation
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2224/00—Indexing scheme for arrangements for connecting or disconnecting semiconductor or solid-state bodies and methods related thereto as covered by H01L24/00
    • H01L2224/73—Means for bonding being of different types provided for in two or more of groups H01L2224/10, H01L2224/18, H01L2224/26, H01L2224/34, H01L2224/42, H01L2224/50, H01L2224/63, H01L2224/71
    • H01L2224/732—Location after the connecting process
    • H01L2224/73251—Location after the connecting process on different surfaces
    • H01L2224/73267—Layer and HDI connectors
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2224/00—Indexing scheme for arrangements for connecting or disconnecting semiconductor or solid-state bodies and methods related thereto as covered by H01L24/00
    • H01L2224/80—Methods for connecting semiconductor or other solid state bodies using means for bonding being attached to, or being formed on, the surface to be connected
    • H01L2224/80001—Methods for connecting semiconductor or other solid state bodies using means for bonding being attached to, or being formed on, the surface to be connected by connecting a bonding area directly to another bonding area, i.e. connectorless bonding, e.g. bumpless bonding
    • H01L2224/80003—Methods for connecting semiconductor or other solid state bodies using means for bonding being attached to, or being formed on, the surface to be connected by connecting a bonding area directly to another bonding area, i.e. connectorless bonding, e.g. bumpless bonding involving a temporary auxiliary member not forming part of the bonding apparatus
    • H01L2224/80006—Methods for connecting semiconductor or other solid state bodies using means for bonding being attached to, or being formed on, the surface to be connected by connecting a bonding area directly to another bonding area, i.e. connectorless bonding, e.g. bumpless bonding involving a temporary auxiliary member not forming part of the bonding apparatus being a temporary or sacrificial substrate
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2224/00—Indexing scheme for arrangements for connecting or disconnecting semiconductor or solid-state bodies and methods related thereto as covered by H01L24/00
    • H01L2224/93—Batch processes
    • H01L2224/95—Batch processes at chip-level, i.e. with connecting carried out on a plurality of singulated devices, i.e. on diced chips
    • H01L2224/97—Batch processes at chip-level, i.e. with connecting carried out on a plurality of singulated devices, i.e. on diced chips the devices being connected to a common substrate, e.g. interposer, said common substrate being separable into individual assemblies after connecting
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2225/00—Details relating to assemblies covered by the group H01L25/00 but not provided for in its subgroups
    • H01L2225/03—All the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/648 and H10K99/00
    • H01L2225/04—All the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/648 and H10K99/00 the devices not having separate containers
    • H01L2225/065—All the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/648 and H10K99/00 the devices not having separate containers the devices being of a type provided for in group H01L27/00
    • H01L2225/06503—Stacked arrangements of devices
    • H01L2225/06541—Conductive via connections through the device, e.g. vertical interconnects, through silicon via [TSV]
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2225/00—Details relating to assemblies covered by the group H01L25/00 but not provided for in its subgroups
    • H01L2225/03—All the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/648 and H10K99/00
    • H01L2225/04—All the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/648 and H10K99/00 the devices not having separate containers
    • H01L2225/065—All the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/648 and H10K99/00 the devices not having separate containers the devices being of a type provided for in group H01L27/00
    • H01L2225/06503—Stacked arrangements of devices
    • H01L2225/06548—Conductive via connections through the substrate, container, or encapsulation
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2924/00—Indexing scheme for arrangements or methods for connecting or disconnecting semiconductor or solid-state bodies as covered by H01L24/00
    • H01L2924/15—Details of package parts other than the semiconductor or other solid state devices to be connected
    • H01L2924/181—Encapsulation
    • H01L2924/1815—Shape
    • H01L2924/1816—Exposing the passive side of the semiconductor or solid-state body
    • H01L2924/18162—Exposing the passive side of the semiconductor or solid-state body of a chip with build-up interconnect

Definitions

  • Computing devices such as mobile phones, personal digital assistants (PDAs), digital cameras, portable players, gaming, and other devices requires the integration of more performance and features into increasingly smaller spaces.
  • PDAs personal digital assistants
  • Some conventional multi-chip modules include two or more semiconductor chips mounted side-by-side on a carrier substrate or in some cases on an interposer (so-called “2.5D”) that is, in-turn, mounted on a carrier substrate.
  • 2.5D interposer
  • FIG. 1 is a block diagram illustrating a processing system employing active bridge chiplets for coupling GPU chiplets in accordance with some embodiments.
  • FIG. 2 is a block diagram illustrating a sectional view of GPU chiplets and active bridge chiplets in accordance with some embodiments.
  • FIG. 3 is a block diagram illustrating another sectional view of GPU chiplets and active bridge chiplets in accordance with some embodiments.
  • FIGS. 4A-4G are block diagrams illustrating a face-to-face process flow for fabricating active-bridge-coupled GPU chiplets in accordance with some embodiments.
  • FIGS. 5A-5D are block diagrams illustrating a face-to-back process flow for fabricating active-bridge-coupled GPU chiplets in accordance with some embodiments. DETAILED DESCRIPTION
  • FIGS. 1-5D illustrate systems and methods of manufacturing active-bridge-coupled GPU chiplets.
  • various architectures already have at least one level of cache (e.g., L3 or other last level cache (LLC)) that is coherent across the entire conventional GPU die.
  • LLC last level cache
  • the chiplet-based GPU architecture positions those physical resources (e.g., LLC) on different dies and communicably couples those physical resources such that the LLC level is unified and remains cache coherent across all GPU chiplets.
  • the L3 cache level is coherent.
  • a method of manufacture includes a face-to-face process in which a first GPU chiplet and a second GPU chiplet are bonded to a temporary carrier wafer. A face surface of an active bridge chiplet is bonded to a face surface of the first and second GPU chiplets before mounting the GPU chiplets to a carrier substrate. In other embodiments, a method of manufacture includes a face-to-back process in which a face surface of an active bridge chiplet is bonded to a back surface of the first and second GPU chiplets.
  • FIG. 1 is a block diagram illustrating a processing system 100 employing an active bridge chiplet for coupling GPU chiplets in accordance with some embodiments.
  • the system 100 includes a central processing unit (CPU) 102 for executing instructions and an array 104 of one or more GPU chiplets, such as the three illustrated GPU chiplets 106-1 , 106-2, and through 106-N (collectively, GPU chiplets 106).
  • the term “chiplet” refers to any device including, but is not limited to, the following characteristics: 1) a chiplet includes an active silicon die containing at least a portion of the computational logic used to solve a full problem (i.e.
  • the computational workload is distributed across multiples of these active silicon dies); 2) chiplets are packaged together as a monolithic unit on the same substrate; and 3) the programming model preserves the concept that the combination of these separate computational dies (i.e., the GPU chiplets) are a single monolithic unit (i.e., each chiplet is not exposed as a separate device to an application that uses the chiplets for processing computational workloads).
  • the CPU 102 is connected via a bus 108 to a system memory 110, such as a dynamic random access memory (DRAM).
  • system memory 110 can also be implemented using other types of memory including static random access memory (SRAM), nonvolatile RAM, and the like.
  • the CPU 102 communicates with the system memory 110 and also the GPU chiplet 106-1 over bus 108 that is implemented as a peripheral component interconnect (PCI) bus, PCI-E bus, or other type of bus.
  • PCI peripheral component interconnect
  • PCI-E PCI-E bus
  • some embodiments of the system 100 includes the GPU chiplet 106-1 communicating with the CPU 102 over a direct connection or via dedicated buses, bridges, switches, routers, and the like.
  • the CPU 102 includes a number of processes, such as executing one or more application(s) 112 to generate graphic commands and a user mode driver 116 (and/or other drivers, such as a kernel mode driver).
  • the one or more applications 112 include applications that utilizes the functionality of the GPU chiplets 106, such as applications that generate work in the system 100 or an operating system (OS).
  • an application 112 includes one or more graphics instructions that instruct the GPU chiplets 106 to render a graphical user interface (GUI) and/or a graphics scene.
  • GUI graphical user interface
  • the graphics instructions include instructions that define a set of one or more graphics primitives to be rendered by the GPU chiplets 106.
  • the application 112 utilizes a graphics application programming interface (API) 114 to invoke a user mode driver 116 (or a similar GPU driver).
  • User mode driver 116 issues one or more commands to the array 104 of one or more GPU chiplets for rendering one or more graphics primitives into displayable graphics images.
  • the user mode driver 116 formulates one or more graphics commands that specify one or more operations for GPU chiplets to perform for rendering graphics.
  • the user mode driver 116 is a part of the application 112 running on the CPU 102.
  • the user mode driver 116 is part of a gaming application running on the CPU 102.
  • a kernel mode driver (not shown), alone or in combination with the user mode driver 116, formulates the one or more graphics commands as part of an operating system running on the CPU 102.
  • an active bridge chiplet 118 communicably couples the GPU chiplets 106 (i.e., GPU chiplets 106-1 through 106- N) to each other.
  • the active bridge chiplet 118 includes an active silicon bridge that serves as a high- bandwidth die-to-die interconnect between GPU chiplet dies.
  • the active bridge chiplet 118 operates as a memory crossbar with a shared, unified last level cache (LLC) to provide inter-chiplet communications and to route cross chiplet synchronization signals.
  • LLC last level cache
  • Caches are naturally an active component (i.e. , require electrical power for operations), so the memory crossbar (e.g., the active bridge chiplet 118) is active for holding those cache memories.
  • Cache sizing is therefore configurable, as a function of the physical size of the active bridge chiplet 118, for different applications along with different chiplet configurations, and the base chiplet(s) to which the active bridge chiplet 118 (e.g., the GPU chiplets 106) is communicably coupled do not pay the cost (e.g., costs related to physical space, power constraints, and the like) of this external cache on the active bridge chiplet 118.
  • the cost e.g., costs related to physical space, power constraints, and the like
  • the CPU 102 is communicably coupled to a single GPU chiplet (i.e., GPU chiplet 106-1) through the bus 108.
  • a single GPU chiplet i.e., GPU chiplet 106-1
  • CPU-to-GPU transactions or communications from the CPU 102 to the array 104 of chiplets 106 is received at the GPU chiplet 106-1.
  • any inter-chiplet communications are routed through the active bridge chiplet 118 as appropriate to access memory channels on other GPU chiplets 106.
  • the GPU chiplet-based system 100 includes GPU chiplets 106 that are addressable as a single, monolithic GPU from a software developer’s perspective (e.g., the CPU 102 and any associated applications / drivers are unaware of the chiplet-based architecture), and therefore avoids requiring any chiplet-specific considerations on the part of a programmer or developer.
  • a semiconductor chip including the array 104 of FIG. 1 is constructed using a face-to- face process flow or a face-to-back process flow.
  • layout 111 illustrates a top down view of an arrangement of the active bridge 118 providing an interconnect for four or more GPU chiplets in accordance with some embodiments.
  • GPU chiplets are arranged in pairs, to form two “columns” of GPU chiplets with the active bridge 118 placed between the columns.
  • GPU chiplet 106-2 is placed lateral to GPU chiplet 106- 1
  • GPU 106-3 is placed below the GPU chiplet 106-1
  • GPU 106-4 is placed lateral to GPU chiplet 106-3 and below GPU chiplet 106-4.
  • the active bridge 118 is placed between the lateral pairs of GPU chiplets.
  • FIG. 2 is a block diagram illustrating a sectional view of active- bridge-coupled GPU chiplets 200 in accordance with some embodiments.
  • the view provides a sectional view of GPU chiplets 106-1 , 106-2, and active bridge chiplet 118 of FIG. 1 along a cross section along a line illustrated in FIG. 1 as line “A”.
  • each GPU chiplet 106 is constructed without any through silicon vias (TSVs).
  • TSVs through silicon vias
  • the GPU chiplets 106 are communicably coupled by way of the active bridge chiplet 118.
  • the active bridge chiplet 118 is an interconnect chip constructed of silicon, germanium or other semiconductor materials and is a bulk semiconductor, semiconductor on insulator or other designs.
  • the active bridge chiplet 118 includes a plurality of internal conductor traces (not shown), which in different embodiments is on a single level or multiple levels as desired.
  • the traces interface electrically with, for example, conductor structures of the PHY regions of the GPU chiplets 106 (e.g., memory PHY 212 of FIG. 2) by way of conducting pathways.
  • the active bridge chiplet 118 is an active bridge die that communicably couples and routes communications between the GPU chiplets 106, thereby forming an active routing network.
  • a carrier wafer 202 is bound to the GPU chiplets 106-1 and
  • TSVs 204 pass through the active bridge chiplet to the GPU chiplets 106 but the graphics core die(s) themselves are not constructed with any TSVs. Instead, in order to pass signal data, through dielectric vias (TDVs) 206 tunnel through a gap fill dielectric layer 208.
  • the gap fill dielectric layer 208 (or other gap fill material) occupies areas where the bridge chiplet die and graphics core die(s) are not present (e.g., areas with vertical discrepancy between the GPU chiplets 106 and the active bridge chiplet 118).
  • the TDVs 206 connect input/output (I/O) power of the GPU chiplets 106 down to the solder interconnects 210, which in different embodiments is a solder bump, micro bump, and the like.
  • the gap fill dielectric layer 208 bring both planes of the bumps on both the GPU chiplets 106 and the active bridge chiplet 118 (e.g., bump 212) into the same plane.
  • the active-bridge- coupled GPU chiplets 200 are constructed using a face-to-face process flow. That is, the active-bridge-coupled GPU chiplets 200 are oriented such that a face surface F of the GPU chiplets 106 are facing a face surface F of the active bridge chiplet.
  • the face surface F (also known as and interchangeably referred to as the “active surface” or the “front surface”) refers to a first surface of a semiconductor die upon which active circuitry 214 (e.g., functional elements, wirings, and the like) are positioned.
  • the back surface B (also known as and interchangeably referred to as the “bottom surface”) refers to a second surface opposite that of the face surface F on the semiconductor die.
  • the components as illustrated in FIG. 2 interface electrically with other electrical structures, such as circuit boards or other structures, by way of interconnect structures 210 and 212 (e.g., solder balls and the like).
  • interconnect structures 210 and 212 e.g., solder balls and the like.
  • the active-bridge-coupled GPU chiplets 200 are mounted on another device, such as circuit board 216.
  • interconnect structures such as pins, land grid array structures, other interconnects, and the like are used without departing from the scope of this disclosure.
  • FIG. 3 is a block diagram of illustrating another sectional view of active-bridge- coupled GPU chiplets 300 in accordance with some embodiments.
  • the view provides a sectional view of GPU chiplets 106-1 , 106-2, and active bridge chiplet 118 of FIG. 1 along line A.
  • the GPU chiplets 106 are communicably coupled by way of the active bridge chiplet 118.
  • the active bridge chiplet 118 is an interconnect chip constructed of silicon, germanium or other semiconductor materials and is a bulk semiconductor, semiconductor on insulator or other designs.
  • the active bridge chiplet 118 includes a plurality of internal conductor traces (not shown), which in different embodiments is on a single level or multiple levels as desired.
  • the traces interface electrically with, for example, conductor structures of the PHY regions of the GPU chiplets 106 (e.g., memory PHY 212 of FIG. 2) by way of conducting pathways.
  • the active bridge chiplet 118 is an active bridge die that communicably couples and routes communications between the GPU chiplets 106, thereby forming an active routing network.
  • each GPU chiplet 106 includes through silicon vias (TSVs) 304.
  • TSVs 304 pass through the GPU chiplets 106 but the active bridge chiplet 118 itself is not constructed with any TSVs.
  • the active-bridge-coupled GPU chiplets also do not include any TDVs as the TSVs 304 connect input/output (I/O) power of the active bridge chiplet down to the solder interconnects 306, which in different embodiments is a solder bump, micro bump, and the like.
  • Interconnect structures 308 electrically couple to the GPU chiplets 106.
  • a layer of dummy silicon 310 occupies areas where the bridge chiplet die and graphics core die(s) are not present (e.g., areas with vertical discrepancy between the GPU chiplets 106 and the active bridge chiplet 118). In this manner, the layer of dummy silicon 310 bring both interconnect bumps associated with communicably and electrically coupling the GPU chiplets 106 and the active bridge chiplet 118 into the same plane and to form a monolithic chip.
  • the active-bridge- coupled GPU chiplets 300 are constructed using a face-to-back process flow.
  • the active-bridge-coupled GPU chiplets 200 are oriented such that a face surface F of the active bridge 118 is facing a back surface B of the GPU chiplets 106.
  • the face surface F also known as and interchangeably referred to as the “active surface” or the “front surface” refers to a first surface of a semiconductor die upon which active circuitry 312 and 314 (e.g., functional elements, wirings, and the like) are positioned.
  • the back surface B also known as and interchangeably referred to as the “bottom surface” refers to a second surface opposite that of the face surface F on the semiconductor dies.
  • the components as illustrated in FIG. 3 interface electrically with other electrical structure, such as circuit boards, substrates, or other structures, by way of interconnect structures 306 and 308 (e.g., solder balls and the like).
  • interconnect structures 306 and 308 e.g., solder balls and the like.
  • the active-bridge-coupled GPU chiplets 300 are mounted on another device, such as circuit board 316.
  • interconnect structures such as pins, land grid array structures, other interconnects, and the like are used without departing from the scope of this disclosure.
  • the active bridge chiplet 118 such as described above with respect to FIGSs. 1-3, thus provides communications between routing fabric of two or more dies and provides coherent L3 memory access with uniform memory access behavior (or mostly uniform memory access behavior).
  • uniform memory access behavior or mostly uniform memory access behavior.
  • Those skilled in the art will recognize that the performance of a processing system generally scales linearly based on the number of GPU chiplets utilized by nature of physical duplication (e.g., as the number of GPU chiplets increases, so does the number of memory PHYs, workgroup processors (WGPs) 202, and the like).
  • FIGS. 4A-4G illustrated are block diagrams of a face-to-face process flow for fabricating active-bridge-coupled GPU chiplets (e.g., active-bridge- coupled GPU chiplets 200 of FIG. 2).
  • active-bridge-coupled GPU chiplets e.g., active-bridge- coupled GPU chiplets 200 of FIG. 2.
  • the active-bridge-coupled GPU chiplets described herein are fabricated as single units and in other embodiments are fabricated en masse in a wafer-like structure (e.g., a reconstituted wafer) in what amounts to a wafer level process.
  • FIG. 4A illustrated is a sectional view depicting a temporary carrier wafer 402.
  • the temporary carrier wafer 402 is constructed of glass, silicon, other types of carrier wafer materials, and the like.
  • a plurality of known-good-dies (KGDs) of graphics core dies (GCDs) 404 e.g., GPU chiplets 106 as previously described with respect to FIGS. 1-3) are bonded to the temporary carrier wafer 402.
  • each of the graphics core dies 404 e.g., GPU chiplets 106) is constructed without any through silicon vias (TSVs) and is also interchangeably referred to herein as a “TSV-free GCD” or a “TSV-free GPU chiplet”.
  • TSVs through silicon vias
  • FIG. 4A includes temporarily bonding a face surface F of the graphics core dies 404 to the temporary carrier wafer 402.
  • temporarily bonding the graphics core dies 404 includes using an adhesive, such as a light or heat activated adhesive, a two-sided tape, or other type of joining technique than can be subsequently undone.
  • the graphics core dies 404 of FIG. 4A are thinned and one or more gap fill dielectric layers 406 are deposited on top of the temporary carrier wafer 402.
  • the gap filling of FIG. 4B includes applying the one or more gap fill dielectric layers 406 using spin coating and baking techniques, other dielectric layer deposition techniques, and the like. Further, in various embodiments, the gap filling of FIG. 4B includes applying one or more gap fill dielectric layers 406 such as to form a planar surface by filling in areas with vertical discrepancy between the temporary carrier wafer 402 and the back surface B of the graphics core dies 404.
  • the surface of the one or more gap fill dielectric layers 406 is subjected to a grinding process to expose the substrate portion of the back surface B of the graphics core dies 404.
  • the combination of the graphics core dies 404 and the one or more gap fill dielectric layers 406 make up a reconstituted unit (or wafer if performed on a wafer- level basis) that is separate-able from the temporary carrier wafer 402.
  • the separating of FIG. 4C includes applying a process appropriate for the technique originally used to join the temporary carrier wafer 402 to the graphics core dies 404.
  • the separating of FIG. 4C includes thermal release tapes or adhesives, light- or heat-activated adhesive releases, and the like.
  • the combination of the graphics core dies 404 and the one or more gap fill dielectric layers 406 is bonded to a second carrier wafer 408 with the back surface B of the graphics core dies 404 bonding to the second carrier wafer 408 for mechanical support. In this manner, the front surface F of the graphics core dies 404 are exposed.
  • a known-good-die of an active bridge die 410 (e.g., active bridge chiplet 118 as previously described with respect to FIGS. 1-3) is bonded to the second carrier wafer 408.
  • the active bridge die 410 includes TSVs 204, as previously described with respect to FIG. 2.
  • the operations of FIG. 4D includes bonding a face surface F of the active bridge die 410 to a face surface F of the graphics core dies 404 (hence the “face-to-face process” term referred to herein).
  • the active bridge die 410 is thinned to expose the TSVs 204 and an additional one or more gap fill dielectric layers 406 are deposited over the graphics core dies 404 and existing gap fill dielectric layers 406 of FIGS. 4A-4D.
  • the gap filling of FIG. 4E includes applying the one or more gap fill dielectric layers 406 using spin coating and baking techniques, other dielectric layer deposition techniques, and the like.
  • the gap filling of FIG. 4B includes applying one or more gap fill dielectric layers 406 such as to form a planar surface by filling in areas with vertical discrepancy between the graphics core dies 404 and the back surface B of the active bridge die 410.
  • the surface of the one or more gap fill dielectric layers 406 is subjected to a grinding process to expose the substrate portion of the back surface B of the active bridge die 410.
  • one or more through dielectric vias 412 are etched in the one or more gap fill dielectric layers 406 that tunnel through the one or more gap fill dielectric layers 406 down to the graphics core dies 404.
  • the one or more gap fill dielectric layers 406 are suitably masked and lithographically patterned, such as by way of photolithography to establish voids / openings to the graphics core dies 404.
  • the void(s) resulting from etching through the one or more gap fill dielectric layers 406 are filled with Copper or other material with high thermal and/or electrical conductivity to form conductive pillars (e.g., TDVs 412) that communicably couple to the graphics core dies 404.
  • a redistribution layer (RDL) structure 414 is fabricated at the bottom surface of the active bridge die 410 and the one or more gap fill dielectric layers 406.
  • the RDL structure 414 includes one or more conductor structures 416 that are coupled to the TDVs 412 through the gap fill dielectric layers 406 and one or more conductor structures 418 coupled to the TSVs 204 of the active bridge die 410.
  • the RDL structure 414 is fabricated with design rules for small spaces associated with I/O mappings of the graphics core dies 404.
  • the one or more conductor structures 416 and 418 are bump pads formed by masking and plating processes to establish laterally extending conductors.
  • the bumping process is completed by applying conductor bumps 420 to the one or more conductor structures 416 and 418 (e.g., bump pads) using known solder plating, pick and place, or printing and reflow techniques.
  • the active-bridge-coupled GPU chiplets are complete and ready to be mounted to a substrate, such as the circuit board 216 illustrated in FIG. 2.
  • FIGS. 5A-5D illustrated are block diagrams of a face-to-back process flow for fabricating active-bridge-coupled GPU chiplets (e.g., active-bridge- coupled GPU chiplets 300 of FIG. 3).
  • active-bridge-coupled GPU chiplets e.g., active-bridge- coupled GPU chiplets 300 of FIG. 3
  • active-bridge-coupled GPU chiplets described herein are fabricated as single units and in other embodiments are fabricated en masse in a wafer-like structure (e.g., a reconstituted wafer) in what amounts to a wafer level process.
  • FIG. 5A illustrated is a sectional view depicting a temporary carrier wafer 502.
  • the temporary carrier wafer 502 is constructed of glass, silicon, other types of carrier wafer materials, and the like.
  • a plurality of known-good-dies (KGDs) of graphics core dies (GCDs) 504 are bonded to the temporary carrier wafer 502.
  • GCDs graphics core dies
  • each GPU chiplet 106 includes through silicon vias (TSVs) 506.
  • TSVs through silicon vias
  • the bonding of FIG. 5A includes temporarily bonding a face surface F of the graphics core dies 504 to the temporary carrier wafer 502.
  • temporarily bonding the graphics core dies 504 includes using an adhesive, such as a light or heat activated adhesive, a two-sided tape, or other type of joining technique than can be subsequently undone.
  • the graphics core dies 504 of FIG. 5A are thinned to expose the TSVs 506 such that the TSVs 506 pass through the GPU chiplets 106 from a face surface F to a back surface B.
  • one or more gap fill dielectric layers 508 are deposited on top of the temporary carrier wafer 502.
  • the gap filling of FIG. 5B includes applying the one or more gap fill dielectric layers 508 using spin coating and baking techniques, other dielectric layer deposition techniques, and the like. Further, in various embodiments, the gap filling of FIG.
  • 5B includes applying one or more gap fill dielectric layers 508 such as to form a planar surface by filling in areas with vertical discrepancy between the temporary carrier wafer 502 and the back surface B of the graphics core dies 504.
  • the applying of the one or more gap fill dielectric layers 508 covers up the back surface B of the graphics core dies 504
  • the surface of the one or more gap fill dielectric layers 508 is subjected to a grinding process to expose the substrate portion of the back surface B of the graphics core dies 504.
  • the combination of the graphics core dies 504 and the one or more gap fill dielectric layers 508 make up a reconstituted unit (or wafer if performed on a wafer-level basis) that is separate-able from the temporary carrier wafer 502.
  • a known-good-die of an active bridge die 510 (e.g., active bridge chiplet 118 as previously described with respect to FIGS. 1-3) is bonded to the graphics core dies 504.
  • the active bridge die 510 is constructed without any TSVs and is also interchangeably referred to herein as a “TSV-free active bridge die” or a “TSV- free active bridge chiplet”.
  • the operations of FIG. 5C includes bonding a face surface F of the active bridge die 510 to a back surface B of the graphics core dies 504 (hence the “face-to-back process” term referred to herein).
  • the operations of FIG. 5C include depositing one or more layers of dummy silicon 512 on top of the graphics core dies 504 and existing one or more gap fill dielectric layers 508.
  • the one or more layers of dummy silicon 512 provide structural integrity for the reconstituted unit after binding to the second carrier wafer 514.
  • the one or more layers of dummy silicon 512 improve thermal performance by conducting heat away the graphics core dies 504 during operation.
  • an additional one or more layers of gap fill dielectric are deposited instead of the one or more layers of dummy silicon 512.
  • the separating of FIG. 5D includes applying a process appropriate for the technique originally used to join the temporary carrier wafer 502 to the graphics core dies 504.
  • the separating of FIG. 5D includes thermal release tapes or adhesives, light- or heat-activated adhesive releases, and the like.
  • the combination of the graphics core dies 504, the one or more gap fill dielectric layers 508, and the one or more layers of dummy silicon 512 is bonded to a second carrier wafer 514 with the back surface B of the active bridge die 510 bonding to the second carrier wafer 514 for mechanical support.
  • the front surface F of the graphics core dies 504 are exposed.
  • the bumping process is completed by applying conductor bumps 516 to the face surface F of the graphics core dies 504 using known solder plating, pick and place, or printing and reflow techniques.
  • the active-bridge-coupled GPU chiplets are complete and ready to be mounted to a substrate, such as the circuit board 316 illustrated in FIG. 3.
  • FIGS. 1-5D are described here in the specific context of a rectangular active bridge chiplet die 118 spanning across the middle of two or three GPU chiplets, various other configurations, die shapes, and in other embodiments different geometries are utilized without departing from the scope of this disclosure.
  • GPU chiplets are fabricated to include active bridge chiplets at one or more corners of a square GPU chiplet such that multiple GPU chiplets are tiled together in a chiplet array.
  • GPU chiplets are fabricated to include active bridge chiplets spanning an entire side of a GPU chiplet such that multiple GPU chiplets are strung together in a long row/column configuration with an intervening active bridge chiplet.
  • an active bridge chiplet deploys monolithic GPU functionality using a set of interconnected GPU chiplets in a manner that makes the GPU chiplet implementation appear as a traditional monolithic GPU from a programmer model/developer perspective.
  • the scalable data fabric of one GPU chiplet is able to access the lower level cache(s) on the active bridge chiplet in nearly the same time as to access the lower level cache on its same chiplet, and thus allows the GPU chiplets to maintain cache coherency without requiring additional inter- chiplet coherency protocols.
  • This low-latency, inter-chiplet cache coherency in turn enables the chiplet-based system to operate as a monolithic GPU from the software developer’s perspective, and thus avoids chiplet-specific considerations on the part of a programmer or developer.
  • a method includes: bonding a first GPU chiplet and a second GPU chiplet to a temporary carrier wafer; bonding a face surface of an active bridge chiplet to a face surface of the first and second GPU chiplets, wherein the active bridge chiplet includes a level of cache memory that is shared by the first and second GPU chiplets; and mounting the first and second GPU chiplets to a carrier substrate.
  • bonding the first GPU chiplet and the second GPU chiplet to the temporary carrier wafer includes bonding the face surface of the first and second GPU chiplets to the temporary carrier wafer.
  • the method includes: depositing one or more gap fill dielectric layers on top of the temporary carrier wafer.
  • depositing one or more gap fill dielectric layers on top of the temporary carrier wafer includes forming a planar surface by filling in areas with a vertical discrepancy between a surface of the temporary carrier wafer and a back surface of the first and second GPU chiplets.
  • the method includes: performing carrier flipping by separating the face surface of the first and second GPU chiplets from the temporary carrier wafer; and bonding a back surface of the first and second GPU chiplets to a second carrier wafer.
  • the method includes: etching one or more voids through the one or more gap fill dielectric layers; and filling the one or more voids with a conductive material to form a set of conductive pillars that communicably couple to at least one of the first and second GPU chiplets.
  • the set of conductive pillars include through dielectric vias.
  • the method includes: fabricating a redistribution layer structure at a back surface of the active bridge chiplet.
  • fabricating the redistribution layer structure further includes positioning a conductor structure on top of each of the set of conductive pillars.
  • the method includes: thinning, subsequent to bonding the face surface of the active bridge chiplet to the face surface of the first and second GPU chiplets, the active bridge chiplet to expose a set of through silicon vias (TSVs) extending from the face surface of the active bridge chiplet to a back surface opposite the face surface of the active bridge chiplet.
  • mounting the first and second GPU chiplets includes mounting active-bridge-coupled GPU chiplets including the first and second GPU chiplets on a circuit board.
  • a method of forming an active-bridge-coupled GPU chiplet unit includes: bonding a first GPU chiplet and a second GPU chiplet to a temporary carrier wafer; bonding a face surface of an active bridge chiplet to a back surface of the first and second GPU chiplets, wherein the active bridge chiplet includes a level of cache memory that is shared by the first and second GPU chiplets; and mounting the active-bridge-coupled GPU chiplet unit including the first and second GPU chiplets to a carrier substrate.
  • the method includes: thinning the first and second GPU chiplets to expose a set of through silicon vias (TSVs) extending from a face surface to the back surface opposite the face surface of the first and second GPU chiplets.
  • TSVs through silicon vias
  • the first GPU chiplet and the second GPU chiplet to the temporary carrier wafer includes bonding a face surface of the first and second GPU chiplets to the temporary carrier wafer.
  • the method includes: depositing one or more gap fill dielectric layers on top of the temporary carrier wafer to form a planar surface by filling in areas with a vertical discrepancy between a surface of the temporary carrier wafer and a back surface of the first and second GPU chiplets.
  • the method includes performing carrier flipping by separating the face surface of the first and second GPU chiplets from the temporary carrier wafer; and bonding a back surface of the active bridge chiplet to a second carrier wafer.
  • the method includes attaching one or more layers of dummy silicon between the second carrier wafer and a back surface of first and second GPU chiplets.
  • the method includes: coupling one or more conductor structures to a face surface of the first and second GPU chiplets.
  • a processor is formed by a method of forming active- bridge-coupled GPU chiplets, the method including: bonding a first GPU chiplet and a second GPU chiplet to a temporary carrier wafer; bonding a face surface of an active bridge chiplet to a face surface of the first and second GPU chiplets, wherein the active bridge chiplet includes a level of cache memory that is shared by the first and second GPU chiplets; and mounting the first and second GPU chiplets to a carrier substrate.
  • bonding the first GPU chiplet and the second GPU chiplet to the temporary carrier wafer includes bonding the face surface of the first and second GPU chiplets to the temporary carrier wafer.
  • a computer readable storage medium may include any non-transitory storage medium, or combination of non-transitory storage media, accessible by a computer system during use to provide instructions and/or data to the computer system.
  • Such storage media can include, but is not limited to, optical media (e.g., compact disc (CD), digital versatile disc (DVD), Blu-Ray disc), magnetic media (e.g., floppy disc , magnetic tape, or magnetic hard drive), volatile memory (e.g., random access memory (RAM) or cache), non-volatile memory (e.g., read-only memory (ROM) or Flash memory), or microelectromechanical systems (MEMS)-based storage media.
  • optical media e.g., compact disc (CD), digital versatile disc (DVD), Blu-Ray disc
  • magnetic media e.g., floppy disc , magnetic tape, or magnetic hard drive
  • volatile memory e.g., random access memory (RAM) or cache
  • non-volatile memory e.g., read-only memory (ROM)
  • the computer readable storage medium may be embedded in the computing system (e.g., system RAM or ROM), fixedly attached to the computing system (e.g., a magnetic hard drive), removably attached to the computing system (e.g., an optical disc or Universal Serial Bus (USB)-based Flash memory), or coupled to the computer system via a wired or wireless network (e.g., network accessible storage (NAS)).
  • system RAM or ROM system RAM or ROM
  • USB Universal Serial Bus
  • NAS network accessible storage
  • certain aspects of the techniques described above are implemented by one or more processors of a processing system executing software.
  • the software includes one or more sets of executable instructions stored or otherwise tangibly embodied on a non-transitory computer readable storage medium.
  • the software can include the instructions and certain data that, when executed by the one or more processors, manipulate the one or more processors to perform one or more aspects of the techniques described above.
  • the non-transitory computer readable storage medium can include, for example, a magnetic or optical disk storage device, solid state storage devices such as Flash memory, a cache, random access memory (RAM) or other non-volatile memory device or devices, and the like.
  • the executable instructions stored on the non-transitory computer readable storage medium are in source code, assembly language code, object code, or other instruction format that is interpreted or otherwise executable by one or more processors. Note that not all of the activities or elements described above in the general description are required, that a portion of a specific activity or device may not be required, and that one or more further activities may be performed, or elements included, in addition to those described. Still further, the order in which activities are listed are not necessarily the order in which they are performed. Also, the concepts have been described with reference to specific embodiments. However, one of ordinary skill in the art appreciates that various modifications and changes can be made without departing from the scope of the present disclosure as set forth in the claims below. Accordingly, the specification and figures are to be regarded in an illustrative rather than a restrictive sense, and all such modifications are intended to be included within the scope of the present disclosure.

Landscapes

  • Engineering & Computer Science (AREA)
  • Microelectronics & Electronic Packaging (AREA)
  • Power Engineering (AREA)
  • Computer Hardware Design (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Condensed Matter Physics & Semiconductors (AREA)
  • Manufacturing & Machinery (AREA)
  • General Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Ceramic Engineering (AREA)
  • Software Systems (AREA)
  • Geometry (AREA)
  • Semiconductor Memories (AREA)
  • Semiconductor Integrated Circuits (AREA)

Abstract

Various multi-die arrangements and methods of manufacturing the same are disclosed. In some embodiments, a method of manufacture includes a face-to-face process in which a first GPU chiplet and a second GPU chiplet are bonded to a temporary carrier wafer. A face surface of an active bridge chiplet is bonded to a face surface of the first and second GPU chiplets before mounting the GPU chiplets to a carrier substrate. In other embodiments, a method of manufacture includes a face-to-back process in which a face surface of an active bridge chiplet is bonded to a back surface of the first and second GPU chiplets.

Description

FABRICATING ACTIVE-BRIDGE-COUPLED GPU CHIPLETS BACKGROUND
Computing devices such as mobile phones, personal digital assistants (PDAs), digital cameras, portable players, gaming, and other devices requires the integration of more performance and features into increasingly smaller spaces. As a result, the density of processor dies and number of dies integrated within a single integrated circuit (1C) package have increased. Some conventional multi-chip modules include two or more semiconductor chips mounted side-by-side on a carrier substrate or in some cases on an interposer (so-called “2.5D”) that is, in-turn, mounted on a carrier substrate.
BRIEF DESCRIPTION OF THE DRAWINGS
The present disclosure may be better understood, and its numerous features and advantages made apparent to those skilled in the art by referencing the accompanying drawings. The use of the same reference symbols in different drawings indicates similar or identical items.
FIG. 1 is a block diagram illustrating a processing system employing active bridge chiplets for coupling GPU chiplets in accordance with some embodiments.
FIG. 2 is a block diagram illustrating a sectional view of GPU chiplets and active bridge chiplets in accordance with some embodiments. FIG. 3 is a block diagram illustrating another sectional view of GPU chiplets and active bridge chiplets in accordance with some embodiments.
FIGS. 4A-4G are block diagrams illustrating a face-to-face process flow for fabricating active-bridge-coupled GPU chiplets in accordance with some embodiments. FIGS. 5A-5D are block diagrams illustrating a face-to-back process flow for fabricating active-bridge-coupled GPU chiplets in accordance with some embodiments. DETAILED DESCRIPTION
Conventional monolithic die designs are becoming increasingly expensive to manufacture. Chiplets have been used successfully in CPU architectures to reduce cost of manufacture and improve yields, as the heterogeneous computational nature of CPUs is more naturally suited to separate CPU cores into distinct units that do not require much inter-communication. In contrast, GPU work by its nature includes parallel work. However, the geometry that a GPU processes includes not only sections of fully parallel work but also work that requires synchronous ordering between different sections. Accordingly, a GPU programming model that spreads sections of work on different threads is often inefficient because the parallelism is difficult to distribute across multiple different working groups and chiplets. In particular it is difficult and expensive computationally to synchronize the memory contents of shared resources throughout the entire system to provide a coherent view of the memory to applications. Additionally, from a logical point of view, applications are written with the view that the system only has a single GPU. That is, even though a conventional GPU includes many GPU cores, applications are programmed as addressing a single device. For at least these reasons, it has been historically challenging to bring chiplet design methodology to GPU architectures.
To improve system performance by using GPU chiplets without altering a relatively simple programming model, FIGS. 1-5D illustrate systems and methods of manufacturing active-bridge-coupled GPU chiplets. Currently, various architectures already have at least one level of cache (e.g., L3 or other last level cache (LLC)) that is coherent across the entire conventional GPU die. Here, the chiplet-based GPU architecture positions those physical resources (e.g., LLC) on different dies and communicably couples those physical resources such that the LLC level is unified and remains cache coherent across all GPU chiplets. Thus, although operating in a massively parallel environment, the L3 cache level is coherent. During operations, a memory address request from the CPU to the GPU is transmitted only to a single GPU chiplet, which then communicates with the active bridge chiplet to locate the requested data. From the CPU’s point of view, it appears to be addressing a single die, monolithic GPU. This allows for use of a large-capacity, multi-chiplet GPU that appears as a single device to an application. As discussed here, in various embodiments, a method of manufacture includes a face-to-face process in which a first GPU chiplet and a second GPU chiplet are bonded to a temporary carrier wafer. A face surface of an active bridge chiplet is bonded to a face surface of the first and second GPU chiplets before mounting the GPU chiplets to a carrier substrate. In other embodiments, a method of manufacture includes a face-to-back process in which a face surface of an active bridge chiplet is bonded to a back surface of the first and second GPU chiplets.
FIG. 1 is a block diagram illustrating a processing system 100 employing an active bridge chiplet for coupling GPU chiplets in accordance with some embodiments. In the depicted example, the system 100 includes a central processing unit (CPU) 102 for executing instructions and an array 104 of one or more GPU chiplets, such as the three illustrated GPU chiplets 106-1 , 106-2, and through 106-N (collectively, GPU chiplets 106). In various embodiments, and as used herein, the term “chiplet” refers to any device including, but is not limited to, the following characteristics: 1) a chiplet includes an active silicon die containing at least a portion of the computational logic used to solve a full problem (i.e. , the computational workload is distributed across multiples of these active silicon dies); 2) chiplets are packaged together as a monolithic unit on the same substrate; and 3) the programming model preserves the concept that the combination of these separate computational dies (i.e., the GPU chiplets) are a single monolithic unit (i.e., each chiplet is not exposed as a separate device to an application that uses the chiplets for processing computational workloads).
In various embodiments, the CPU 102 is connected via a bus 108 to a system memory 110, such as a dynamic random access memory (DRAM). In various embodiments, the system memory 110 can also be implemented using other types of memory including static random access memory (SRAM), nonvolatile RAM, and the like. In the illustrated embodiment, the CPU 102 communicates with the system memory 110 and also the GPU chiplet 106-1 over bus 108 that is implemented as a peripheral component interconnect (PCI) bus, PCI-E bus, or other type of bus. However, some embodiments of the system 100 includes the GPU chiplet 106-1 communicating with the CPU 102 over a direct connection or via dedicated buses, bridges, switches, routers, and the like. As illustrated, the CPU 102 includes a number of processes, such as executing one or more application(s) 112 to generate graphic commands and a user mode driver 116 (and/or other drivers, such as a kernel mode driver). In various embodiments, the one or more applications 112 include applications that utilizes the functionality of the GPU chiplets 106, such as applications that generate work in the system 100 or an operating system (OS). In some implementations an application 112 includes one or more graphics instructions that instruct the GPU chiplets 106 to render a graphical user interface (GUI) and/or a graphics scene. For example, in some implementations, the graphics instructions include instructions that define a set of one or more graphics primitives to be rendered by the GPU chiplets 106.
In some embodiments, the application 112 utilizes a graphics application programming interface (API) 114 to invoke a user mode driver 116 (or a similar GPU driver). User mode driver 116 issues one or more commands to the array 104 of one or more GPU chiplets for rendering one or more graphics primitives into displayable graphics images. Based on the graphics instructions issued by application 112 to the user mode driver 116, the user mode driver 116 formulates one or more graphics commands that specify one or more operations for GPU chiplets to perform for rendering graphics. In some embodiments, the user mode driver 116 is a part of the application 112 running on the CPU 102. For example, in some embodiments the user mode driver 116 is part of a gaming application running on the CPU 102. Similarly, in some implementations a kernel mode driver (not shown), alone or in combination with the user mode driver 116, formulates the one or more graphics commands as part of an operating system running on the CPU 102.
In the depicted embodiment of FIG. 1 , an active bridge chiplet 118 communicably couples the GPU chiplets 106 (i.e., GPU chiplets 106-1 through 106- N) to each other. Although three GPU chiplets 106 are shown in FIG. 1 , the number of GPU chiplets in the chiplet array 104 is a matter of design choice and varies in other embodiments, such as described in more detail below. In various embodiments, such as discussed below in more detail with respect to FIG. 2, the active bridge chiplet 118 includes an active silicon bridge that serves as a high- bandwidth die-to-die interconnect between GPU chiplet dies. Additionally, the active bridge chiplet 118 operates as a memory crossbar with a shared, unified last level cache (LLC) to provide inter-chiplet communications and to route cross chiplet synchronization signals. Caches are naturally an active component (i.e. , require electrical power for operations), so the memory crossbar (e.g., the active bridge chiplet 118) is active for holding those cache memories. Cache sizing is therefore configurable, as a function of the physical size of the active bridge chiplet 118, for different applications along with different chiplet configurations, and the base chiplet(s) to which the active bridge chiplet 118 (e.g., the GPU chiplets 106) is communicably coupled do not pay the cost (e.g., costs related to physical space, power constraints, and the like) of this external cache on the active bridge chiplet 118.
As a general operational overview, the CPU 102 is communicably coupled to a single GPU chiplet (i.e., GPU chiplet 106-1) through the bus 108. CPU-to-GPU transactions or communications from the CPU 102 to the array 104 of chiplets 106 is received at the GPU chiplet 106-1. Subsequently, any inter-chiplet communications are routed through the active bridge chiplet 118 as appropriate to access memory channels on other GPU chiplets 106. In this manner, the GPU chiplet-based system 100 includes GPU chiplets 106 that are addressable as a single, monolithic GPU from a software developer’s perspective (e.g., the CPU 102 and any associated applications / drivers are unaware of the chiplet-based architecture), and therefore avoids requiring any chiplet-specific considerations on the part of a programmer or developer. As described in more detail below, in some embodiments a semiconductor chip including the array 104 of FIG. 1 is constructed using a face-to- face process flow or a face-to-back process flow.
It will be appreciated that in different embodiments the GPU chiplets 106 are placed in different arrangements so that the active bridge 118 supports more than two GPU chiplets. An example is illustrated at FIG. 1 as layout 111. In particular, layout 111 illustrates a top down view of an arrangement of the active bridge 118 providing an interconnect for four or more GPU chiplets in accordance with some embodiments. In the depicted example of layout 111 , GPU chiplets are arranged in pairs, to form two “columns” of GPU chiplets with the active bridge 118 placed between the columns. Thus, GPU chiplet 106-2 is placed lateral to GPU chiplet 106- 1 , GPU 106-3 is placed below the GPU chiplet 106-1 , and GPU 106-4 is placed lateral to GPU chiplet 106-3 and below GPU chiplet 106-4. The active bridge 118 is placed between the lateral pairs of GPU chiplets.
Additional details of the chiplet-based architecture may be understood with reference to FIG. 2, which is a block diagram illustrating a sectional view of active- bridge-coupled GPU chiplets 200 in accordance with some embodiments. The view provides a sectional view of GPU chiplets 106-1 , 106-2, and active bridge chiplet 118 of FIG. 1 along a cross section along a line illustrated in FIG. 1 as line “A”. In various embodiments, each GPU chiplet 106 is constructed without any through silicon vias (TSVs). The GPU chiplets 106 are communicably coupled by way of the active bridge chiplet 118. In various embodiments, the active bridge chiplet 118 is an interconnect chip constructed of silicon, germanium or other semiconductor materials and is a bulk semiconductor, semiconductor on insulator or other designs.
The active bridge chiplet 118 includes a plurality of internal conductor traces (not shown), which in different embodiments is on a single level or multiple levels as desired. The traces interface electrically with, for example, conductor structures of the PHY regions of the GPU chiplets 106 (e.g., memory PHY 212 of FIG. 2) by way of conducting pathways. In this manner, the active bridge chiplet 118 is an active bridge die that communicably couples and routes communications between the GPU chiplets 106, thereby forming an active routing network. As shown in FIG. 2, a carrier wafer 202 is bound to the GPU chiplets 106-1 and
106-2. In this embodiment configuration, TSVs 204 pass through the active bridge chiplet to the GPU chiplets 106 but the graphics core die(s) themselves are not constructed with any TSVs. Instead, in order to pass signal data, through dielectric vias (TDVs) 206 tunnel through a gap fill dielectric layer 208. The gap fill dielectric layer 208 (or other gap fill material) occupies areas where the bridge chiplet die and graphics core die(s) are not present (e.g., areas with vertical discrepancy between the GPU chiplets 106 and the active bridge chiplet 118). As shown, the TDVs 206 connect input/output (I/O) power of the GPU chiplets 106 down to the solder interconnects 210, which in different embodiments is a solder bump, micro bump, and the like. In this manner, the gap fill dielectric layer 208 bring both planes of the bumps on both the GPU chiplets 106 and the active bridge chiplet 118 (e.g., bump 212) into the same plane. As described in more detail below with respect to FIG. 4, the active-bridge- coupled GPU chiplets 200 are constructed using a face-to-face process flow. That is, the active-bridge-coupled GPU chiplets 200 are oriented such that a face surface F of the GPU chiplets 106 are facing a face surface F of the active bridge chiplet. Those skilled in the art will recognize that the face surface F (also known as and interchangeably referred to as the “active surface” or the “front surface”) refers to a first surface of a semiconductor die upon which active circuitry 214 (e.g., functional elements, wirings, and the like) are positioned. Similarly, the back surface B (also known as and interchangeably referred to as the “bottom surface”) refers to a second surface opposite that of the face surface F on the semiconductor die.
In various embodiments, the components as illustrated in FIG. 2 interface electrically with other electrical structures, such as circuit boards or other structures, by way of interconnect structures 210 and 212 (e.g., solder balls and the like). For example, as illustrated in FIG. 2, the active-bridge-coupled GPU chiplets 200 are mounted on another device, such as circuit board 216. However, those skilled in the art will appreciate that in other embodiments various types of interconnect structures such as pins, land grid array structures, other interconnects, and the like are used without departing from the scope of this disclosure.
FIG. 3 is a block diagram of illustrating another sectional view of active-bridge- coupled GPU chiplets 300 in accordance with some embodiments. The view provides a sectional view of GPU chiplets 106-1 , 106-2, and active bridge chiplet 118 of FIG. 1 along line A. As previously noted, the GPU chiplets 106 are communicably coupled by way of the active bridge chiplet 118. In various embodiments, the active bridge chiplet 118 is an interconnect chip constructed of silicon, germanium or other semiconductor materials and is a bulk semiconductor, semiconductor on insulator or other designs.
The active bridge chiplet 118 includes a plurality of internal conductor traces (not shown), which in different embodiments is on a single level or multiple levels as desired. The traces interface electrically with, for example, conductor structures of the PHY regions of the GPU chiplets 106 (e.g., memory PHY 212 of FIG. 2) by way of conducting pathways. In this manner, the active bridge chiplet 118 is an active bridge die that communicably couples and routes communications between the GPU chiplets 106, thereby forming an active routing network.
As shown in FIG. 3, and in a manner similar to the components of FIG. 2, a carrier wafer 302 is bound to the GPU chiplets 106-1 and 106-2. However, in contrast to the embodiment of FIG. 2, each GPU chiplet 106 includes through silicon vias (TSVs) 304. In this embodiment configuration, TSVs 304 pass through the GPU chiplets 106 but the active bridge chiplet 118 itself is not constructed with any TSVs. Additionally, the active-bridge-coupled GPU chiplets also do not include any TDVs as the TSVs 304 connect input/output (I/O) power of the active bridge chiplet down to the solder interconnects 306, which in different embodiments is a solder bump, micro bump, and the like. Interconnect structures 308 electrically couple to the GPU chiplets 106. In various embodiments, a layer of dummy silicon 310 (or other gap fill material) occupies areas where the bridge chiplet die and graphics core die(s) are not present (e.g., areas with vertical discrepancy between the GPU chiplets 106 and the active bridge chiplet 118). In this manner, the layer of dummy silicon 310 bring both interconnect bumps associated with communicably and electrically coupling the GPU chiplets 106 and the active bridge chiplet 118 into the same plane and to form a monolithic chip.
As described in more detail below with respect to FIG. 5, the active-bridge- coupled GPU chiplets 300 are constructed using a face-to-back process flow. In particular, the active-bridge-coupled GPU chiplets 200 are oriented such that a face surface F of the active bridge 118 is facing a back surface B of the GPU chiplets 106. Those skilled in the art will recognize that the face surface F (also known as and interchangeably referred to as the “active surface” or the “front surface”) refers to a first surface of a semiconductor die upon which active circuitry 312 and 314 (e.g., functional elements, wirings, and the like) are positioned. Similarly, the back surface B (also known as and interchangeably referred to as the “bottom surface”) refers to a second surface opposite that of the face surface F on the semiconductor dies.
In various embodiments, the components as illustrated in FIG. 3 interface electrically with other electrical structure, such as circuit boards, substrates, or other structures, by way of interconnect structures 306 and 308 (e.g., solder balls and the like). For example, as illustrated in FIG. 3, the active-bridge-coupled GPU chiplets 300 are mounted on another device, such as circuit board 316. However, those skilled in the art will appreciate that in other embodiments various types of interconnect structures such as pins, land grid array structures, other interconnects, and the like are used without departing from the scope of this disclosure.
The active bridge chiplet 118, such as described above with respect to FIGSs. 1-3, thus provides communications between routing fabric of two or more dies and provides coherent L3 memory access with uniform memory access behavior (or mostly uniform memory access behavior). Those skilled in the art will recognize that the performance of a processing system generally scales linearly based on the number of GPU chiplets utilized by nature of physical duplication (e.g., as the number of GPU chiplets increases, so does the number of memory PHYs, workgroup processors (WGPs) 202, and the like).
Referring now to FIGS. 4A-4G, illustrated are block diagrams of a face-to-face process flow for fabricating active-bridge-coupled GPU chiplets (e.g., active-bridge- coupled GPU chiplets 200 of FIG. 2). Those skilled in the art will recognize that in some embodiments the active-bridge-coupled GPU chiplets described herein are fabricated as single units and in other embodiments are fabricated en masse in a wafer-like structure (e.g., a reconstituted wafer) in what amounts to a wafer level process.
At FIG. 4A, illustrated is a sectional view depicting a temporary carrier wafer 402. In various embodiments, the temporary carrier wafer 402 is constructed of glass, silicon, other types of carrier wafer materials, and the like. A plurality of known-good-dies (KGDs) of graphics core dies (GCDs) 404 (e.g., GPU chiplets 106 as previously described with respect to FIGS. 1-3) are bonded to the temporary carrier wafer 402. In various embodiments, each of the graphics core dies 404 (e.g., GPU chiplets 106) is constructed without any through silicon vias (TSVs) and is also interchangeably referred to herein as a “TSV-free GCD” or a “TSV-free GPU chiplet”. The bonding of FIG. 4A includes temporarily bonding a face surface F of the graphics core dies 404 to the temporary carrier wafer 402. In various embodiments, temporarily bonding the graphics core dies 404 includes using an adhesive, such as a light or heat activated adhesive, a two-sided tape, or other type of joining technique than can be subsequently undone. At FIG. 4B, the graphics core dies 404 of FIG. 4A are thinned and one or more gap fill dielectric layers 406 are deposited on top of the temporary carrier wafer 402.
In various embodiments, the gap filling of FIG. 4B includes applying the one or more gap fill dielectric layers 406 using spin coating and baking techniques, other dielectric layer deposition techniques, and the like. Further, in various embodiments, the gap filling of FIG. 4B includes applying one or more gap fill dielectric layers 406 such as to form a planar surface by filling in areas with vertical discrepancy between the temporary carrier wafer 402 and the back surface B of the graphics core dies 404. In some embodiments, when the applying of the one or more gap fill dielectric layers 406 covers up the back surface B of the graphics core dies 404, the surface of the one or more gap fill dielectric layers 406 is subjected to a grinding process to expose the substrate portion of the back surface B of the graphics core dies 404. At this point, the combination of the graphics core dies 404 and the one or more gap fill dielectric layers 406 make up a reconstituted unit (or wafer if performed on a wafer- level basis) that is separate-able from the temporary carrier wafer 402.
At FIG. 4C, carrier flipping is applied and the temporary carrier wafer 402 is separated from the combination of the graphics core dies 404 and the one or more gap fill dielectric layers 406. In various embodiments, the separating of FIG. 4C includes applying a process appropriate for the technique originally used to join the temporary carrier wafer 402 to the graphics core dies 404. For example, in various embodiments, the separating of FIG. 4C includes thermal release tapes or adhesives, light- or heat-activated adhesive releases, and the like. Additionally, the combination of the graphics core dies 404 and the one or more gap fill dielectric layers 406 is bonded to a second carrier wafer 408 with the back surface B of the graphics core dies 404 bonding to the second carrier wafer 408 for mechanical support. In this manner, the front surface F of the graphics core dies 404 are exposed.
At FIG. 4D, a known-good-die of an active bridge die 410 (e.g., active bridge chiplet 118 as previously described with respect to FIGS. 1-3) is bonded to the second carrier wafer 408. The active bridge die 410 includes TSVs 204, as previously described with respect to FIG. 2. In various embodiments, the operations of FIG. 4D includes bonding a face surface F of the active bridge die 410 to a face surface F of the graphics core dies 404 (hence the “face-to-face process” term referred to herein).
At FIG. 4E, the active bridge die 410 is thinned to expose the TSVs 204 and an additional one or more gap fill dielectric layers 406 are deposited over the graphics core dies 404 and existing gap fill dielectric layers 406 of FIGS. 4A-4D. In various embodiments, the gap filling of FIG. 4E includes applying the one or more gap fill dielectric layers 406 using spin coating and baking techniques, other dielectric layer deposition techniques, and the like. Further, in various embodiments, the gap filling of FIG. 4B includes applying one or more gap fill dielectric layers 406 such as to form a planar surface by filling in areas with vertical discrepancy between the graphics core dies 404 and the back surface B of the active bridge die 410. In some embodiments, when the applying of the one or more gap fill dielectric layers 406 covers up the back surface B of the active bridge die 410, the surface of the one or more gap fill dielectric layers 406 is subjected to a grinding process to expose the substrate portion of the back surface B of the active bridge die 410. Further, one or more through dielectric vias 412 (such as the TDVs 206 as previously described with respect to FIG. 2) are etched in the one or more gap fill dielectric layers 406 that tunnel through the one or more gap fill dielectric layers 406 down to the graphics core dies 404. For example, in some embodiments, the one or more gap fill dielectric layers 406 are suitably masked and lithographically patterned, such as by way of photolithography to establish voids / openings to the graphics core dies 404. In various embodiments, the void(s) resulting from etching through the one or more gap fill dielectric layers 406 are filled with Copper or other material with high thermal and/or electrical conductivity to form conductive pillars (e.g., TDVs 412) that communicably couple to the graphics core dies 404.
At FIG. 4F, a redistribution layer (RDL) structure 414 is fabricated at the bottom surface of the active bridge die 410 and the one or more gap fill dielectric layers 406. In various embodiments, the RDL structure 414 includes one or more conductor structures 416 that are coupled to the TDVs 412 through the gap fill dielectric layers 406 and one or more conductor structures 418 coupled to the TSVs 204 of the active bridge die 410. It will be appreciated that, in various embodiments, the RDL structure 414 is fabricated with design rules for small spaces associated with I/O mappings of the graphics core dies 404. In some embodiments, the one or more conductor structures 416 and 418 are bump pads formed by masking and plating processes to establish laterally extending conductors.
At FIG. 4G, the bumping process is completed by applying conductor bumps 420 to the one or more conductor structures 416 and 418 (e.g., bump pads) using known solder plating, pick and place, or printing and reflow techniques. At this stage, the active-bridge-coupled GPU chiplets are complete and ready to be mounted to a substrate, such as the circuit board 216 illustrated in FIG. 2.
Referring now to FIGS. 5A-5D, illustrated are block diagrams of a face-to-back process flow for fabricating active-bridge-coupled GPU chiplets (e.g., active-bridge- coupled GPU chiplets 300 of FIG. 3). Those skilled in the art will recognize that the active-bridge-coupled GPU chiplets described herein in some embodiments are fabricated as single units and in other embodiments are fabricated en masse in a wafer-like structure (e.g., a reconstituted wafer) in what amounts to a wafer level process.
At FIG. 5A, illustrated is a sectional view depicting a temporary carrier wafer 502. In various embodiments, the temporary carrier wafer 502 is constructed of glass, silicon, other types of carrier wafer materials, and the like. A plurality of known-good-dies (KGDs) of graphics core dies (GCDs) 504 (e.g., GPU chiplets 106 as previously described with respect to FIGS. 1-3) are bonded to the temporary carrier wafer 502. In various embodiments, in contrast to the embodiments of FIGS.
2 and 4A-4G, each GPU chiplet 106 includes through silicon vias (TSVs) 506. The bonding of FIG. 5A includes temporarily bonding a face surface F of the graphics core dies 504 to the temporary carrier wafer 502. In various embodiments, temporarily bonding the graphics core dies 504 includes using an adhesive, such as a light or heat activated adhesive, a two-sided tape, or other type of joining technique than can be subsequently undone.
At FIG. 5B, the graphics core dies 504 of FIG. 5A are thinned to expose the TSVs 506 such that the TSVs 506 pass through the GPU chiplets 106 from a face surface F to a back surface B. Further, one or more gap fill dielectric layers 508 are deposited on top of the temporary carrier wafer 502. In various embodiments, the gap filling of FIG. 5B includes applying the one or more gap fill dielectric layers 508 using spin coating and baking techniques, other dielectric layer deposition techniques, and the like. Further, in various embodiments, the gap filling of FIG. 5B includes applying one or more gap fill dielectric layers 508 such as to form a planar surface by filling in areas with vertical discrepancy between the temporary carrier wafer 502 and the back surface B of the graphics core dies 504. In some embodiments, when the applying of the one or more gap fill dielectric layers 508 covers up the back surface B of the graphics core dies 504, the surface of the one or more gap fill dielectric layers 508 is subjected to a grinding process to expose the substrate portion of the back surface B of the graphics core dies 504. At this point, the combination of the graphics core dies 504 and the one or more gap fill dielectric layers 508 make up a reconstituted unit (or wafer if performed on a wafer-level basis) that is separate-able from the temporary carrier wafer 502.
At FIG. 5C, a known-good-die of an active bridge die 510 (e.g., active bridge chiplet 118 as previously described with respect to FIGS. 1-3) is bonded to the graphics core dies 504. In various embodiments, such as previously described with respect to FIG. 3, the active bridge die 510 is constructed without any TSVs and is also interchangeably referred to herein as a “TSV-free active bridge die” or a “TSV- free active bridge chiplet”. In various embodiments, the operations of FIG. 5C includes bonding a face surface F of the active bridge die 510 to a back surface B of the graphics core dies 504 (hence the “face-to-back process” term referred to herein).
Additionally, in some embodiments, the operations of FIG. 5C include depositing one or more layers of dummy silicon 512 on top of the graphics core dies 504 and existing one or more gap fill dielectric layers 508. With reference to FIG. 5D below, the one or more layers of dummy silicon 512 provide structural integrity for the reconstituted unit after binding to the second carrier wafer 514. The one or more layers of dummy silicon 512 improve thermal performance by conducting heat away the graphics core dies 504 during operation. In other embodiments, an additional one or more layers of gap fill dielectric are deposited instead of the one or more layers of dummy silicon 512.
At FIG. 5D, carrier flipping is applied and the temporary carrier wafer 502 is separated from the combination of the graphics core dies 504, the one or more gap fill dielectric layers 508, and the one or more layers of dummy silicon 512. In various embodiments, the separating of FIG. 5D includes applying a process appropriate for the technique originally used to join the temporary carrier wafer 502 to the graphics core dies 504. For example, in various embodiments, the separating of FIG. 5D includes thermal release tapes or adhesives, light- or heat-activated adhesive releases, and the like. Additionally, the combination of the graphics core dies 504, the one or more gap fill dielectric layers 508, and the one or more layers of dummy silicon 512 is bonded to a second carrier wafer 514 with the back surface B of the active bridge die 510 bonding to the second carrier wafer 514 for mechanical support. In this manner, the front surface F of the graphics core dies 504 are exposed. Additionally, the bumping process is completed by applying conductor bumps 516 to the face surface F of the graphics core dies 504 using known solder plating, pick and place, or printing and reflow techniques. At this stage, the active-bridge-coupled GPU chiplets are complete and ready to be mounted to a substrate, such as the circuit board 316 illustrated in FIG. 3.
Those skilled in the art will recognize that although FIGS. 1-5D are described here in the specific context of a rectangular active bridge chiplet die 118 spanning across the middle of two or three GPU chiplets, various other configurations, die shapes, and in other embodiments different geometries are utilized without departing from the scope of this disclosure. For example, in some embodiments, GPU chiplets are fabricated to include active bridge chiplets at one or more corners of a square GPU chiplet such that multiple GPU chiplets are tiled together in a chiplet array. Similarly, in other embodiments, GPU chiplets are fabricated to include active bridge chiplets spanning an entire side of a GPU chiplet such that multiple GPU chiplets are strung together in a long row/column configuration with an intervening active bridge chiplet.
Accordingly, as discussed herein, an active bridge chiplet deploys monolithic GPU functionality using a set of interconnected GPU chiplets in a manner that makes the GPU chiplet implementation appear as a traditional monolithic GPU from a programmer model/developer perspective. The scalable data fabric of one GPU chiplet is able to access the lower level cache(s) on the active bridge chiplet in nearly the same time as to access the lower level cache on its same chiplet, and thus allows the GPU chiplets to maintain cache coherency without requiring additional inter- chiplet coherency protocols. This low-latency, inter-chiplet cache coherency in turn enables the chiplet-based system to operate as a monolithic GPU from the software developer’s perspective, and thus avoids chiplet-specific considerations on the part of a programmer or developer.
As disclosed herein, in some embodiments a method includes: bonding a first GPU chiplet and a second GPU chiplet to a temporary carrier wafer; bonding a face surface of an active bridge chiplet to a face surface of the first and second GPU chiplets, wherein the active bridge chiplet includes a level of cache memory that is shared by the first and second GPU chiplets; and mounting the first and second GPU chiplets to a carrier substrate. In one aspect, bonding the first GPU chiplet and the second GPU chiplet to the temporary carrier wafer includes bonding the face surface of the first and second GPU chiplets to the temporary carrier wafer. In another aspect, the method includes: depositing one or more gap fill dielectric layers on top of the temporary carrier wafer. In another aspect, depositing one or more gap fill dielectric layers on top of the temporary carrier wafer includes forming a planar surface by filling in areas with a vertical discrepancy between a surface of the temporary carrier wafer and a back surface of the first and second GPU chiplets.
In one aspect, the method includes: performing carrier flipping by separating the face surface of the first and second GPU chiplets from the temporary carrier wafer; and bonding a back surface of the first and second GPU chiplets to a second carrier wafer. In another aspect, the method includes: etching one or more voids through the one or more gap fill dielectric layers; and filling the one or more voids with a conductive material to form a set of conductive pillars that communicably couple to at least one of the first and second GPU chiplets. In still another aspect, the set of conductive pillars include through dielectric vias. In yet another aspect, the method includes: fabricating a redistribution layer structure at a back surface of the active bridge chiplet.
In one aspect, fabricating the redistribution layer structure further includes positioning a conductor structure on top of each of the set of conductive pillars. In another aspect, the method includes: thinning, subsequent to bonding the face surface of the active bridge chiplet to the face surface of the first and second GPU chiplets, the active bridge chiplet to expose a set of through silicon vias (TSVs) extending from the face surface of the active bridge chiplet to a back surface opposite the face surface of the active bridge chiplet. In yet another aspect, mounting the first and second GPU chiplets includes mounting active-bridge-coupled GPU chiplets including the first and second GPU chiplets on a circuit board.
In some embodiments, a method of forming an active-bridge-coupled GPU chiplet unit, includes: bonding a first GPU chiplet and a second GPU chiplet to a temporary carrier wafer; bonding a face surface of an active bridge chiplet to a back surface of the first and second GPU chiplets, wherein the active bridge chiplet includes a level of cache memory that is shared by the first and second GPU chiplets; and mounting the active-bridge-coupled GPU chiplet unit including the first and second GPU chiplets to a carrier substrate. In one aspect, the method includes: thinning the first and second GPU chiplets to expose a set of through silicon vias (TSVs) extending from a face surface to the back surface opposite the face surface of the first and second GPU chiplets. In another aspect, the first GPU chiplet and the second GPU chiplet to the temporary carrier wafer includes bonding a face surface of the first and second GPU chiplets to the temporary carrier wafer.
In one aspect, the method includes: depositing one or more gap fill dielectric layers on top of the temporary carrier wafer to form a planar surface by filling in areas with a vertical discrepancy between a surface of the temporary carrier wafer and a back surface of the first and second GPU chiplets. In another aspect, the method includes performing carrier flipping by separating the face surface of the first and second GPU chiplets from the temporary carrier wafer; and bonding a back surface of the active bridge chiplet to a second carrier wafer. In still another aspect, the method includes attaching one or more layers of dummy silicon between the second carrier wafer and a back surface of first and second GPU chiplets. In still another aspect, the method includes: coupling one or more conductor structures to a face surface of the first and second GPU chiplets.
In some embodiments, a processor is formed by a method of forming active- bridge-coupled GPU chiplets, the method including: bonding a first GPU chiplet and a second GPU chiplet to a temporary carrier wafer; bonding a face surface of an active bridge chiplet to a face surface of the first and second GPU chiplets, wherein the active bridge chiplet includes a level of cache memory that is shared by the first and second GPU chiplets; and mounting the first and second GPU chiplets to a carrier substrate. In one aspect, bonding the first GPU chiplet and the second GPU chiplet to the temporary carrier wafer includes bonding the face surface of the first and second GPU chiplets to the temporary carrier wafer.
A computer readable storage medium may include any non-transitory storage medium, or combination of non-transitory storage media, accessible by a computer system during use to provide instructions and/or data to the computer system. Such storage media can include, but is not limited to, optical media (e.g., compact disc (CD), digital versatile disc (DVD), Blu-Ray disc), magnetic media (e.g., floppy disc , magnetic tape, or magnetic hard drive), volatile memory (e.g., random access memory (RAM) or cache), non-volatile memory (e.g., read-only memory (ROM) or Flash memory), or microelectromechanical systems (MEMS)-based storage media. The computer readable storage medium may be embedded in the computing system (e.g., system RAM or ROM), fixedly attached to the computing system (e.g., a magnetic hard drive), removably attached to the computing system (e.g., an optical disc or Universal Serial Bus (USB)-based Flash memory), or coupled to the computer system via a wired or wireless network (e.g., network accessible storage (NAS)).
In some embodiments, certain aspects of the techniques described above are implemented by one or more processors of a processing system executing software. The software includes one or more sets of executable instructions stored or otherwise tangibly embodied on a non-transitory computer readable storage medium. The software can include the instructions and certain data that, when executed by the one or more processors, manipulate the one or more processors to perform one or more aspects of the techniques described above. The non-transitory computer readable storage medium can include, for example, a magnetic or optical disk storage device, solid state storage devices such as Flash memory, a cache, random access memory (RAM) or other non-volatile memory device or devices, and the like. The executable instructions stored on the non-transitory computer readable storage medium are in source code, assembly language code, object code, or other instruction format that is interpreted or otherwise executable by one or more processors. Note that not all of the activities or elements described above in the general description are required, that a portion of a specific activity or device may not be required, and that one or more further activities may be performed, or elements included, in addition to those described. Still further, the order in which activities are listed are not necessarily the order in which they are performed. Also, the concepts have been described with reference to specific embodiments. However, one of ordinary skill in the art appreciates that various modifications and changes can be made without departing from the scope of the present disclosure as set forth in the claims below. Accordingly, the specification and figures are to be regarded in an illustrative rather than a restrictive sense, and all such modifications are intended to be included within the scope of the present disclosure.
Benefits, other advantages, and solutions to problems have been described above with regard to specific embodiments. However, the benefits, advantages, solutions to problems, and any feature(s) that may cause any benefit, advantage, or solution to occur or become more pronounced are not to be construed as a critical, required, or essential feature of any or all the claims. Moreover, the particular embodiments disclosed above are illustrative only, as the disclosed subject matter may be modified and practiced in different but equivalent manners apparent to those skilled in the art having the benefit of the teachings herein. No limitations are intended to the details of construction or design herein shown, other than as described in the claims below. It is therefore evident that the particular embodiments disclosed above may be altered or modified and all such variations are considered within the scope of the disclosed subject matter. Accordingly, the protection sought herein is as set forth in the claims below.

Claims

WHAT IS CLAIMED IS:
1. A method of forming active-bridge-coupled GPU chiplets, comprising: bonding a first GPU chiplet and a second GPU chiplet to a temporary carrier wafer; bonding a face surface of an active bridge chiplet to a face surface of the first and second GPU chiplets, wherein the active bridge chiplet includes a level of cache memory that is shared by the first and second GPU chiplets; and mounting the first and second GPU chiplets to a carrier substrate.
2. The method of claim 1 , wherein bonding the first GPU chiplet and the second GPU chiplet to the temporary carrier wafer comprises bonding the face surface of the first and second GPU chiplets to the temporary carrier wafer.
3. The method of claim 2, further comprising: depositing one or more gap fill dielectric layers on top of the temporary carrier wafer.
4. The method of claim 3, wherein depositing one or more gap fill dielectric layers on top of the temporary carrier wafer includes forming a planar surface by filling in areas with a vertical discrepancy between a surface of the temporary carrier wafer and a back surface of the first and second GPU chiplets.
5. The method of claim 3, further comprising: performing carrier flipping by separating the face surface of the first and second GPU chiplets from the temporary carrier wafer; and bonding a back surface of the first and second GPU chiplets to a second carrier wafer.
6. The method of claim 3, further comprising: etching one or more voids through the one or more gap fill dielectric layers; and filling the one or more voids with a conductive material to form a set of conductive pillars that communicably couple to at least one of the first and second GPU chiplets.
7. The method of claim 6, wherein the set of conductive pillars comprise through dielectric vias.
8. The method of claim 6, further comprising: fabricating a redistribution layer structure at a back surface of the active bridge chiplet.
9. The method of claim 8, wherein fabricating the redistribution layer structure further comprises positioning a conductor structure on top of each of the set of conductive pillars.
10. The method of claim 1 , further comprising: thinning, subsequent to bonding the face surface of the active bridge chiplet to the face surface of the first and second GPU chiplets, the active bridge chiplet to expose a set of through silicon vias (TSVs) extending from the face surface of the active bridge chiplet to a back surface opposite the face surface of the active bridge chiplet.
11. The method of claim 1 , wherein mounting the first and second GPU chiplets includes mounting active-bridge-coupled GPU chiplets including the first and second GPU chiplets on a circuit board.
12. A method of forming an active-bridge-coupled GPU chiplet unit, comprising: bonding a first GPU chiplet and a second GPU chiplet to a temporary carrier wafer; bonding a face surface of an active bridge chiplet to a back surface of the first and second GPU chiplets, wherein the active bridge chiplet includes a level of cache memory that is shared by the first and second GPU chiplets; and mounting the active-bridge-coupled GPU chiplet unit including the first and second GPU chiplets to a carrier substrate.
13. The method of claim 12, further comprising: thinning the first and second GPU chiplets to expose a set of through silicon vias (TSVs) extending from a face surface to the back surface opposite the face surface of the first and second GPU chiplets.
14. The method of claim 12, wherein bonding the first GPU chiplet and the second GPU chiplet to the temporary carrier wafer comprises bonding a face surface of the first and second GPU chiplets to the temporary carrier wafer.
15. The method of claim 14, further comprising: depositing one or more gap fill dielectric layers on top of the temporary carrier wafer to form a planar surface by filling in areas with a vertical discrepancy between a surface of the temporary carrier wafer and a back surface of the first and second GPU chiplets.
16. The method of claim 15, further comprising: performing carrier flipping by separating the face surface of the first and second GPU chiplets from the temporary carrier wafer; and bonding a back surface of the active bridge chiplet to a second carrier wafer.
17. The method of claim 16, further comprising: attaching one or more layers of dummy silicon between the second carrier wafer and a back surface of first and second GPU chiplets.
18. The method of claim 12, further comprising: coupling one or more conductor structures to a face surface of the first and second GPU chiplets.
19. A processor formed by a method of forming active-bridge-coupled GPU chiplets, the method comprising: bonding a first GPU chiplet and a second GPU chiplet to a temporary carrier wafer; bonding a face surface of an active bridge chiplet to a face surface of the first and second GPU chiplets, wherein the active bridge chiplet includes a level of cache memory that is shared by the first and second GPU chiplets; and mounting the first and second GPU chiplets to a carrier substrate.
20. The processor of claim 19, wherein bonding the first GPU chiplet and the second
GPU chiplet to the temporary carrier wafer comprises bonding the face surface of the first and second GPU chiplets to the temporary carrier wafer.
PCT/US2020/052444 2019-09-27 2020-09-24 Fabricating active-bridge-coupled gpu chiplets WO2021061949A1 (en)

Priority Applications (4)

Application Number Priority Date Filing Date Title
EP20867435.8A EP4035019A4 (en) 2019-09-27 2020-09-24 Fabricating active-bridge-coupled gpu chiplets
KR1020227011449A KR20220072838A (en) 2019-09-27 2020-09-24 Building active bridged GPU chiplets
JP2022517359A JP2022549787A (en) 2019-09-27 2020-09-24 Fabrication of Active Bridged GPU Chiplets
CN202080067204.0A CN114467166A (en) 2019-09-27 2020-09-24 Manufacturing active bridge coupled GPU chiplets

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US16/585,480 2019-09-27
US16/585,480 US20210098419A1 (en) 2019-09-27 2019-09-27 Fabricating active-bridge-coupled gpu chiplets

Publications (1)

Publication Number Publication Date
WO2021061949A1 true WO2021061949A1 (en) 2021-04-01

Family

ID=75161698

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2020/052444 WO2021061949A1 (en) 2019-09-27 2020-09-24 Fabricating active-bridge-coupled gpu chiplets

Country Status (6)

Country Link
US (1) US20210098419A1 (en)
EP (1) EP4035019A4 (en)
JP (1) JP2022549787A (en)
KR (1) KR20220072838A (en)
CN (1) CN114467166A (en)
WO (1) WO2021061949A1 (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102020128855A1 (en) * 2020-05-21 2021-11-25 Taiwan Semiconductor Manufacturing Co., Ltd. CHIPLETS 3D SOIC SYSTEM INTEGRATION AND MANUFACTURING PROCESS
US11462495B2 (en) * 2020-05-21 2022-10-04 Taiwan Semiconductor Manufacturing Co., Ltd. Chiplets 3D SoIC system integration and fabrication methods

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150243528A1 (en) * 2014-02-27 2015-08-27 Palo Alto Research Center Incorporated Fabrication Method For Microelectronic Components And Microchip Inks Used In Electrostatic Assembly
US20170365580A1 (en) 2016-04-21 2017-12-21 Micron Technology, Inc. Semiconductor package and fabrication method thereof
US20180102251A1 (en) * 2016-10-07 2018-04-12 Invensas Bonding Technologies, Inc. Direct-bonded native interconnects and active base die
US20180102338A1 (en) * 2016-10-06 2018-04-12 Advanced Micro Devices, Inc. Circuit board with bridge chiplets
US20180233470A1 (en) * 2017-02-15 2018-08-16 Infineon Technologies Ag Handling thin wafer during chip manufacture
US20180366436A1 (en) 2017-06-15 2018-12-20 Invensas Corporation Multi-Chip Modules Formed Using Wafer-Level Processing of a Reconstitute Wafer
WO2019032322A1 (en) * 2017-08-11 2019-02-14 Advanced Micro Devices, Inc. Molded chip combination
WO2019132971A1 (en) 2017-12-29 2019-07-04 Intel Corporation Microelectronic assemblies

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9443783B2 (en) * 2012-06-27 2016-09-13 Taiwan Semiconductor Manufacturing Company, Ltd. 3DIC stacking device and method of manufacture
US8922243B2 (en) * 2012-12-23 2014-12-30 Advanced Micro Devices, Inc. Die-stacked memory device with reconfigurable logic
US9768145B2 (en) * 2015-08-31 2017-09-19 Taiwan Semiconductor Manufacturing Company, Ltd. Methods of forming multi-die package structures including redistribution layers
US11289424B2 (en) * 2018-11-29 2022-03-29 Taiwan Semiconductor Manufacturing Company, Ltd. Package and method of manufacturing the same
US10803548B2 (en) * 2019-03-15 2020-10-13 Intel Corporation Disaggregation of SOC architecture
US11063019B2 (en) * 2019-07-17 2021-07-13 Taiwan Semiconductor Manufacturing Company, Ltd. Package structure, chip structure and method of fabricating the same

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150243528A1 (en) * 2014-02-27 2015-08-27 Palo Alto Research Center Incorporated Fabrication Method For Microelectronic Components And Microchip Inks Used In Electrostatic Assembly
US20170365580A1 (en) 2016-04-21 2017-12-21 Micron Technology, Inc. Semiconductor package and fabrication method thereof
US20180102338A1 (en) * 2016-10-06 2018-04-12 Advanced Micro Devices, Inc. Circuit board with bridge chiplets
US20180102251A1 (en) * 2016-10-07 2018-04-12 Invensas Bonding Technologies, Inc. Direct-bonded native interconnects and active base die
US20180233470A1 (en) * 2017-02-15 2018-08-16 Infineon Technologies Ag Handling thin wafer during chip manufacture
US20180366436A1 (en) 2017-06-15 2018-12-20 Invensas Corporation Multi-Chip Modules Formed Using Wafer-Level Processing of a Reconstitute Wafer
WO2019032322A1 (en) * 2017-08-11 2019-02-14 Advanced Micro Devices, Inc. Molded chip combination
WO2019132971A1 (en) 2017-12-29 2019-07-04 Intel Corporation Microelectronic assemblies

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4035019A4

Also Published As

Publication number Publication date
KR20220072838A (en) 2022-06-02
CN114467166A (en) 2022-05-10
JP2022549787A (en) 2022-11-29
EP4035019A1 (en) 2022-08-03
US20210098419A1 (en) 2021-04-01
EP4035019A4 (en) 2023-11-08

Similar Documents

Publication Publication Date Title
US8736068B2 (en) Hybrid bonding techniques for multi-layer semiconductor stacks
TWI599012B (en) Semiconductor package and fabrication method thereof
US9495498B2 (en) Universal inter-layer interconnect for multi-layer semiconductor stacks
US11507527B2 (en) Active bridge chiplet with integrated cache
US8445918B2 (en) Thermal enhancement for multi-layer semiconductor stacks
US20090196086A1 (en) High bandwidth cache-to-processing unit communication in a multiple processor/cache system
TW202209325A (en) Dram chiplet structure and method for manufacturing the same
US20140264836A1 (en) System-in-package with interposer pitch adapter
TWI591773B (en) Die stacking techniques in bga memory package for small footprint cpu and memory motherboard design
US20240330196A1 (en) Gpu chiplets using high bandwidth crosslinks
US20230352412A1 (en) Multiple die package using an embedded bridge connecting dies
WO2021061949A1 (en) Fabricating active-bridge-coupled gpu chiplets
JP2024511776A (en) Die stacking for modular parallel processors

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20867435

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022517359

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 20227011449

Country of ref document: KR

Kind code of ref document: A

WWE Wipo information: entry into national phase

Ref document number: 2020867435

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 2020867435

Country of ref document: EP

Effective date: 20220428