WO2016000181A1 - 一种3d堆叠器件、芯片及通信方法 - Google Patents

一种3d堆叠器件、芯片及通信方法 Download PDF

Info

Publication number
WO2016000181A1
WO2016000181A1 PCT/CN2014/081286 CN2014081286W WO2016000181A1 WO 2016000181 A1 WO2016000181 A1 WO 2016000181A1 CN 2014081286 W CN2014081286 W CN 2014081286W WO 2016000181 A1 WO2016000181 A1 WO 2016000181A1
Authority
WO
WIPO (PCT)
Prior art keywords
antenna
resource
storage
subunit
unit
Prior art date
Application number
PCT/CN2014/081286
Other languages
English (en)
French (fr)
Inventor
陈少杰
韦竹林
张树杰
赵俊峰
何睿
杨伟
林芃
Original Assignee
华为技术有限公司
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 华为技术有限公司 filed Critical 华为技术有限公司
Priority to PCT/CN2014/081286 priority Critical patent/WO2016000181A1/zh
Priority to CN201480038640.XA priority patent/CN105393353B/zh
Publication of WO2016000181A1 publication Critical patent/WO2016000181A1/zh

Links

Classifications

    • HELECTRICITY
    • H01ELECTRIC ELEMENTS
    • H01LSEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L25/00Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof
    • HELECTRICITY
    • H01ELECTRIC ELEMENTS
    • H01LSEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L2924/00Indexing scheme for arrangements or methods for connecting or disconnecting semiconductor or solid-state bodies as covered by H01L24/00
    • H01L2924/0001Technical content checked by a classifier
    • H01L2924/0002Not covered by any one of groups H01L24/00, H01L24/00 and H01L2224/00

Definitions

  • the present invention relates to the field of electronic technologies, and in particular, to a 3D stacked device, a chip, and a communication method. Background technique
  • the thickness of the existing terminals is thinner and thinner based on convenience of carrying or aesthetics. This requires that the integrated chip function inside the terminal becomes more and more abundant and smaller and smaller. Inside the integrated chip, the computing unit and the memory unit are the most important components, and the interconnection method has a great influence on the volume of the integrated chip.
  • wired interconnect technology includes an on-chip 3D package, a 2.5D package, and an off-chip interconnect structure.
  • the 3D package refers to a packaging technology in which two or more chips are stacked in the same direction in the same package without changing the size of the package, thereby realizing the doubling of the storage capacity;
  • the length of the interconnect is significantly shortened, the signal is transmitted faster and the interference is less; again, multiple different functional chips can be stacked together to enable more functions in a single package; finally, the chip in 3D package is used.
  • each core and each storage subunit must exist between each A fixed input and output path, each core is connected to the storage sub-unit through a through-silicon via (TSV), and is switched by a mechanical switch, and the transmission path is selected by the selector.
  • TSV through-silicon via
  • the rate of the mechanical switch is limited, the bandwidth of the wired transmission is also limited, and because there are many input and output channels, it is necessary to reserve a plurality of 10 (in/out, abbreviated as 10) ports, and there are many wires, resulting in complicated interconnection structure and integration.
  • Embodiments of the present invention provide a 3D stacked device, a chip, and a communication method, which can solve the problem that the internal transmission bandwidth of the chip is limited, and the transmission path cannot be dynamically allocated according to application requirements.
  • a first aspect of the embodiments of the present invention provides a 3D stacked device, which may include:
  • a first substrate configured to carry a computing unit, where the computing unit includes a core
  • a second substrate configured to carry a storage unit
  • the storage unit includes a storage subunit
  • a first antenna array is located on the first substrate, electrically connected to the computing unit, and directed to the second substrate, Transmitting data output by the computing unit and/or receiving data transmitted by the second antenna array;
  • the second antenna array is located on the second substrate, electrically connected to the storage unit, and directed to the first substrate, for receiving data transmitted by the first antenna array and/or receiving the storage The data output by the unit;
  • an adjusting unit configured to adjust data transmission between the transmissions of the antennas in the first antenna array and/or the second antenna array, wherein the transmission parameters of the antennas include phase parameters.
  • the antenna in the first antenna array is a through-silicon via on the first substrate that is directed toward the second substrate and is not in contact with the second substrate.
  • the antenna in the second antenna array is a through hole on the second substrate that is directed to the first substrate and is not in contact with the first substrate.
  • the adjusting unit is further configured to:
  • the adjusting unit includes:
  • a monitoring subunit configured to acquire resource occupation information of the storage subunit in the storage unit, and a recording subunit, configured to generate a resource status table according to the resource occupation information acquired by the monitoring subunit, where the resource status table is At least one of resource usage status information, resource usage information, or temperature information of the storage subunit;
  • a receiving subunit configured to receive a resource request of the computing unit, where the resource request includes at least one of a resource request request, a resource release request, or a resource lock request;
  • Determining a subunit configured to perform a topological calculation on the transmission path according to the resource request and the resource status table, and determine between the kernel in the computing unit and the storage subunit in the storage unit Transmission path and transmission parameters of the antenna;
  • Adjusting a subunit configured to adjust transmission parameters of the antenna according to the transmission path determined by the determining subunit and transmission parameters of the antenna to ensure the kernel in the computing unit and the storage subunit in the storage unit Complete the data transfer.
  • the transmission parameters of the antenna further include:
  • the adjusting unit further includes: the determining subunit being the core and the computing unit The storage subunit in the storage unit allocates a transmission path.
  • the adjusting unit further includes:
  • a data moving subunit configured to move data in the storage unit according to the data moving instruction of the computing unit
  • the first antenna array and the The two antenna arrays are in a medium whose impedance value reaches a preset threshold.
  • the first The antenna array and the second antenna array perform data transmission in a near field communication manner.
  • a second aspect of the embodiments of the present invention provides a chip, which may include:
  • a 3D stacked device according to the first aspect of the present invention or any one of the first aspects.
  • a third aspect of the embodiments of the present invention provides a communication method, which is applicable to the 3D stacking device according to the first aspect of the present invention or the implementation manner of any of the first aspect, which may include:
  • a resource request includes at least one of a resource request request, a resource release request, or a resource lock request;
  • the transmission path performs topology calculation, determines a kernel in the computing unit, and generates the resource status table according to the acquired resource occupation information, where the resource status table includes resource usage status information, resource usage rate information, or the storage sub At least one of the temperature information of the unit.
  • the transmission parameter of the antenna further includes:
  • FIG. 1 is a schematic structural diagram of a 3D stacked device according to an embodiment of the present invention.
  • FIG. 2 is a schematic structural diagram of another 3D stacked device according to an embodiment of the present invention.
  • FIG. 3 is a schematic structural diagram of still another 3D stacked device according to an embodiment of the present invention.
  • FIG. 4 is a schematic structural diagram of still another 3D stacked device according to an embodiment of the present invention.
  • FIG. 5 is a schematic diagram of a configuration of an adjustment unit in a 3D stacked device according to an embodiment of the present invention
  • FIG. 6 is a schematic diagram of a configuration of another adjustment unit in a 3D stacked device according to an embodiment of the present invention
  • the 3D stacked device includes:
  • the first substrate 10 is used to carry the computing unit 20, and the computing unit includes a core 21;
  • the second substrate 30 is configured to carry the storage unit 40.
  • the storage unit includes a storage subunit 41.
  • the first antenna array 51 is located on the first substrate 10, electrically connected to the computing unit 20, and pointed to
  • the second substrate 30 is configured to transmit data output by the computing unit 20 and/or receive data transmitted by the second antenna array 52;
  • the second antenna array 52 is located on the second substrate 30, electrically connected to the storage unit 40, and is directed to the first substrate 10 for receiving data transmitted by the first antenna array 51 and/or Or receiving data output by the storage unit 40;
  • the adjusting unit 60 is configured to adjust transmission parameters of the antennas in the first antenna array 51 and/or the second antenna array 52 to ensure the core 21 and the storage subunit 40 in the computing unit 20
  • the data transmission is completed between the storage sub-units 41, wherein the transmission parameters of the antenna may be packaged.
  • the transmission parameters of the antenna may further include a number parameter of the docked antenna, a transmit power parameter of the docked antenna, and the like.
  • the electrical signal sent by the computing unit 20 may be an instruction or output electrical signal from the core 21, which may be converted by the antenna in the first antenna array 51 into a radio frequency signal and then received by the antenna in the second antenna array 52 and The RF signal is reconverted to an electrical signal output to the storage unit 40 and finally to the storage subunit 41.
  • the storage sub-unit 41 may be a single storage element, or may be a storage cluster composed of a plurality of interconnected storage elements, such as a memory cluster composed of a plurality of memory chips (DIE) or a plurality of storage clusters storing DIEs such as FLASH. .
  • DIE memory chips
  • the adjusting unit 60 may adjust only the transmission parameters of the antennas in the first antenna array 51 or only the second when adjusting the transmission parameters of the antennas in the first antenna array 51 and/or the second antenna array 52. Transmission parameters of the antennas in the antenna array 52. Of course, it is also possible to simultaneously adjust the transmission parameters of the antennas in the two antenna arrays, and only need to ensure the cores 21 in the calculation unit 20 and the storage subunits in the storage unit 40. 41 can complete the data transmission.
  • the antenna in the first antenna array 51 may be a through-silicon via on the first substrate 10 that is directed toward the second substrate 30 and not in contact with the second substrate 30, the second antenna
  • the antenna in the array 52 is a through-silicon via on the second substrate 30 that is directed toward the first substrate 10 and is not in contact with the first substrate 10. In this way, it is no longer necessary to configure additional antennas for the computing unit 20 and the storage unit 40, which saves resources, improves the integration of the antennas, and reduces the size and cost of the 3D stacked devices and chips.
  • the adjusting unit 60 can be divided into two adjusting subunits, which are respectively connected to the first antenna array 51 and the second antenna array 52 to adjust the transmission parameters of the antenna. At this time, the two adjusting subunits are respectively carried on the first substrate 10 And the second substrate 30, of course, the adjusting unit 60 may also be separately located on the first substrate 10 or the second substrate 30, and then the control end of the adjusting unit 60 and the first antenna array 51 and the second antenna array The antennas in 52 can be connected separately.
  • FIG. 2 is a schematic diagram of a composition of a 3D stacked device according to an embodiment of the present invention.
  • the 3D stacked device includes:
  • the first substrate 10 is configured to carry the computing unit 20, and the computing unit 20 includes a first core 21 and a second core 22;
  • the second substrate 30 is configured to carry the storage unit 40, and the storage unit 40 includes a first storage sub-unit 41 and a second storage sub-unit 42;
  • a first antenna array 51 located on the first substrate 10, electrically connected to the computing unit 20, and directed to the second substrate 30 for transmitting data output by the computing unit 20 and/or receiving a second Data transmitted by the antenna array 52;
  • the second antenna array 52 is located on the second substrate 30, electrically connected to the storage unit 40, and is directed to the first substrate 10 for receiving data transmitted by the first antenna array 51 and/or Or receiving data output by the storage unit;
  • the first antenna array 51 is composed of an antenna 511 and an antenna 512
  • the second antenna array 52 is composed of an antenna 521 and an antenna 522.
  • the first antenna array 51 and the second antenna array 52 are used together to obtain an antenna.
  • Array 50 is composed of an antenna 511 and an antenna 512
  • the second antenna array 52 is composed of an antenna 521 and an antenna 522.
  • the adjusting unit 60 is configured to adjust transmission parameters of the antennas in the first antenna array 51 and/or the second antenna array 52 to ensure a core in the computing unit 20 and a storage subunit in the storage unit 40
  • the data transmission is completed, wherein the transmission parameters of the antenna include but are not limited to phase parameters.
  • the transmission parameters of the antenna may further include a numbering parameter of the docked antenna, a transmission power parameter of the docked antenna, and the like.
  • the electrical signals from the computing unit 20 can be converted by the antennas in the first antenna array 51 into RF signals for transmission, then received by the antennas in the second antenna array 52 and reconverted into RF signals for output to the storage unit 40.
  • the storage subunit may be a single storage element, or a storage cluster composed of a plurality of interconnected storage elements, such as a memory cluster composed of a plurality of memory chips (DIEs) or a plurality of storage clusters storing DIEs such as FLASH.
  • DIEs memory chips
  • FLASH a plurality of storage clusters storing DIEs
  • the adjusting unit 60 may adjust only the transmission parameters of the antennas in the first antenna array 51 or only the second when adjusting the transmission parameters of the antennas in the first antenna array 51 and/or the second antenna array 52.
  • Transmission parameters of the antenna in the antenna array 52 of course, it is also possible to adjust two antenna arrays simultaneously In the transmission parameters of the antenna, it is only necessary to ensure that the core in the computing unit 20 and the storage subunit in the storage unit 40 can complete the data transmission.
  • the first core 21 when the first core 21 needs to access data in the second storage subunit 42, if the antennas in the first antenna array 51 and the second antenna array 52 are idle in frequency or time at this time, the first core 21 can actively request data transmission through the antenna 511 and the antenna 522. At this time, only the transmission parameters such as the phase of the antenna 511 can be adjusted, so that the antenna 511 and the antenna 522 are docked to complete the data transmission, and only the transmission parameters of the antenna 522 can be adjusted.
  • the antenna 522 and the antenna 511 are connected to each other to complete data transmission, and the transmission parameters of the antenna 511 and the antenna 522 can be adjusted at the same time so that the antenna 511 and the antenna 522 are docked to complete data transmission, and the phase adjustment of the antenna in the antenna array can also be referred to
  • the phase-weighting method of the antenna array in the prior art realizes the phase adjustment of a specific antenna, that is, the specific phase adjustment of a specific antenna is realized by adjusting the phase of two or more antennas in the array (the cooperation of the main lobe and the side lobe, etc.).
  • the adjusting unit 60 can establish a suitable dynamic link for the core or application by adjusting the phase of the antenna to meet the requirements of the kernel or the application, and the transmission parameters of the adjusted antenna, such as docking.
  • the antenna number parameter and the antenna phase parameter may be preset or manually input, or may be selected by the computing unit 20 according to the resource occupation information of the storage unit 40.
  • the adjustment unit 60 may also be configured to comprehensively consider the requirements of the kernel and the storage subunit.
  • the storage situation is automatically calculated and assigned. For example, when the preset mode is used, the transmission priority of a certain core and different storage subunits and the corresponding docking antenna and phase information can be set first.
  • the configuration adjustment unit 60 comprehensively considers the requirements of the kernel and the storage condition of the storage subunit for automatic calculation and allocation, thereby better performing adaptive matching of resources, so that different kernels and different storage subunits can be optimally adapted.
  • the adjusting unit 60 can be divided into two adjusting subunits, which are respectively connected to the first antenna array 51 and the second antenna array 52 to adjust the transmission parameters of the antenna.
  • the two adjusting subunits are respectively carried on the first substrate 10 And the second substrate 30, of course, the adjusting unit 60 may be separately located on the first substrate 10 or the second substrate 30, and then the control end of the adjusting unit 60 and the first antenna array 51 and the second antenna array The antennas in 52 can be connected separately.
  • the establishment of a wireless dynamic link can be achieved by adjusting the transmission parameters of the antenna, and the transmission rate is high.
  • the antenna can support one-to-many transmission.
  • one core of the computing unit can communicate with multiple storage sub-units in a broadcast manner, and a single storage sub-unit can also simultaneously transmit data to multiple cores for calculation and processing, thereby achieving multiple Enter the effect of multiple outputs.
  • the input and output channels are only established when data transmission is required, and no selector is required for path selection.
  • the interconnection structure is simple, saving 10 ports and chip volume. At the same time, it can also be compatible with Complementary Metal Oxide Semiconductor (CMOS) process.
  • CMOS Complementary Metal Oxide Semiconductor
  • the same set of antennas can be shared at different frequencies or different times to realize frequency division multiplexing and time division multiplexing. For example, if multiple cores want to use the antenna 511 and the antenna 521 to communicate with the first storage subunit 41 at the same time, different communication frequencies can be configured for multiple cores, and then multiple cores can simultaneously use the antenna 511 using different frequencies. And the antenna 521 is implemented to communicate with the first storage subunit 41.
  • the same communication frequency can also be used, but the time when the plurality of cores use the antenna 511 and the antenna 521 to communicate with the storage subunit is shifted, such as configuring the first core 21
  • the signal is transmitted by using a certain frequency through the antenna 511 between 9:00 and 3:00, and then the signal is received by the antenna 521 and converted into an electrical signal and output to the first storage subunit 41, and the second core passes through the antenna at 10:11 to 11:00.
  • 511 uses the frequency to transmit a signal to enable communication with the first memory subunit 41.
  • the core in the computing unit 20 can be used in conjunction with the antenna in the first antenna array 51 by using a wireless transceiver driver 10, and the storage subunit in the storage unit 40 can also drive 10 circuits through wireless transceiver. Used in conjunction with an antenna in the second antenna array 52.
  • the antenna in the first antenna array 51 is a through-silicon via on the first substrate 10 that is directed toward the second substrate 30 and is not in contact with the second substrate 30, and the second antenna array
  • the antenna in 52 is a through-silicon via on the second substrate 30 that is directed toward the first substrate 10 and is not in contact with the first substrate 10.
  • the first substrate 10 and the third substrate 30 can use a high-impedance substrate with an impedance value of 750 Ohm, which is 100 times that of a common P substrate, and the first antenna array and the second antenna.
  • the array is in a medium where the impedance value reaches a preset threshold. For example, if the impedance value reaches a preset threshold such as 750 Ohm. Thereby, the transmission loss between the first antenna array 51 and the second antenna array 52 can be greatly reduced.
  • the first antenna array 51 and the second antenna array 52 can perform data transmission in a near field communication manner. Because it is a near-field communication in the ultra-high frequency band, the on-chip area and power consumption load, such as the antenna radiation power, are very small, so it has the significant advantages of small area, low power consumption, and high bandwidth.
  • the computing unit 20 includes only two cores
  • the The storage unit 40 includes only two storage subunits. Due to the rapid development of the existing electronic chip technology, chips of 4 cores, 8 cores or even more cores have been widely used, and the storage subunits are similar. Moreover, the advantages of the present invention are more obvious when the number of cores is larger. However, in view of the convenience of description, this embodiment is only illustrated by a dual-core scenario, and the principle is similar in a multi-core scenario. In addition, the communication between the single-core and the single-storage sub-units can also be implemented by using the wireless communication method described in this embodiment. For details, refer to the embodiment shown in FIG. 1 and related descriptions, and details are not described herein again.
  • the single-port replaces the multi-port mechanical switching mode, which does not require switching, has a fast transmission rate, and can achieve multi-input and multi-output effects, and supports frequency division multiplexing and time division multiplexing.
  • the 3D stacked device includes:
  • the first substrate 10 is configured to carry the computing unit 20, and the computing unit 20 includes a first core 21 and a second core 22;
  • the second substrate 30 is configured to carry the storage unit 40, and the storage unit 40 includes a first storage sub-unit 41 and a second storage sub-unit 42;
  • a first antenna array 51 located on the first substrate 10, electrically connected to the computing unit 20, and directed to the second substrate 30 for transmitting data output by the computing unit 20 and/or receiving a second Data transmitted by the antenna array 52;
  • the second antenna array 52 is located on the second substrate 30, electrically connected to the storage unit 40, and is directed to the first substrate 10 for receiving data transmitted by the first antenna array 51 and/or Or receiving data output by the storage unit;
  • the first antenna array 51 is composed of an antenna 511 and an antenna 512
  • the second antenna array 52 is composed of an antenna 521 and an antenna 522.
  • the first antenna array 51 and the second antenna array 52 are used together to obtain an antenna.
  • Array 50 is composed of an antenna 511 and an antenna 512
  • the second antenna array 52 is composed of an antenna 521 and an antenna 522.
  • the adjusting unit 60 is configured to adjust transmission parameters of the antennas in the first antenna array 51 and/or the second antenna array 52 to ensure storage in the computing unit 20 and storage in the storage unit 40
  • the subunit completes data transmission, wherein the transmission parameters of the antenna include, but are not limited to, phase parameters.
  • the adjusting unit 60 is further configured to:
  • the kernel and/or the storage subunit may be configured with corresponding priorities, and dynamically adjusted according to the priority. If the event processed by the first kernel 21 is more urgent, the configuration may be configured. The high priority is 9, and the second core 22 has a lower priority of 8.
  • the storage subunit is also configurable. For example, the first storage subunit 41 is configured with a priority of 7, and the second storage subunit is configured with a priority.
  • the first core 21 and the second core 22 simultaneously request to transmit data to the first storage subunit 41 using the antenna 511 and the antenna 521, the first core 21 is preferentially satisfied, and the first core 21 is configured for the first core 21.
  • the second core 22 can be configured with the second core 22 - day Line 512 - antenna 522 - the transmission path of the second storage sub-unit 42, or may wait for the first core 21 to complete the data transfer or after the first core 21 issues a resource release request, the second core 22 reuses the second core 22
  • the second core 22 issues a resource request request, and the application is at 10 o'clock -
  • the antenna 511, the antenna 521, and the first storage subunit 41 are used for communication at 11 o'clock, the antenna 511, the antenna 521, and the first storage subunit 41 will not be allocated to the second core 22, and of course, if the second core 22 is prioritized. If the level is higher, the requirement of the second core 22 can be preferentially satisfied.
  • the resource occupation information of the storage subunit such as the resource usage status, the resource usage rate, or the temperature information of the storage subunit, may be prioritized during dynamic adjustment. For example, when the first core 21 requests to use the antenna 511 and the antenna 521 to transmit data to the first storage subunit 41, if the resource usage state of the first storage subunit 41 is currently in use, or the resources of the first storage subunit 41 If the usage rate has reached 95%, or the first storage subunit 42 has reached a temperature of 60 degrees Celsius due to simultaneous communication with multiple cores or other reasons, the adjustment unit 60 may allocate the first core 21 to the first core 21.
  • the resource request and the resource occupation information can be considered in a balanced manner.
  • the priorities of the cores and the storage subunits are the same, and only the kernel or the storage subunit should be allocated according to the timing. The best transmission path. If the first core 21 requests to communicate with the first storage subunit 41 through the antennas 511 and 521, if the antenna 511 is already occupied, it is preferred to allocate the currently idle antenna. If the first storage subunit 41 has no free storage space, then The first core 21 is allocated a storage sub-unit having a free storage space. If the resource usage state or the resource usage rate or the temperature of the first storage sub-unit 41 reaches a warning threshold, the first kernel 21 may be allocated another storage.
  • the unit is the second storage subunit; if the first core 21 and the second core 22 simultaneously request to communicate with the first storage subunit 41 through the antennas 511 and 521, the first core 21 can be assigned a relatively close antenna 511 and an antenna. 521. Communicate with the first storage subunit 41, and allocate the antennas 512 and 522 with the closer distance to the second core 22 to implement communication with the second storage subunit 42.
  • the adjusting unit 60 may include but is not limited to:
  • the monitoring sub-unit 61 is configured to acquire the resource occupation information of the storage sub-unit in the storage unit, and the recording sub-unit 62 is configured to generate a resource status table according to the resource occupation information acquired by the monitoring sub-unit 61, where the resource status is
  • the table includes at least one of resource usage status information, resource usage information, or temperature information of the storage subunit;
  • the resource usage status may include, but is not limited to, used and unused, or may be a predetermined usage state (used by a certain kernel for a certain period of time), etc.; the resource usage rate may be the usage rate of the current storage subunit, such as 1G.
  • the temperature information of the storage subunit may indicate the temperature of the current storage subunit. If a certain temperature threshold is exceeded, other storage subunits may be selected to establish a link.
  • the receiving subunit 63 is configured to receive a resource request of the computing unit 20, where the resource request includes at least one of a resource request request, a resource release request, or a resource lock request;
  • the resource request request may be used to apply for the use of resources, for example, the first kernel 21 may request to use the antenna 511 and the antenna 521 and the storage subunit 41 to communicate at a certain time or frequency through a resource request request; the resource release request may be used to apply for release of the current Occupied resources, such as the first kernel 21 began to apply at 9 o'clock -10 points use antenna 511 and antenna 521 and storage sub-unit 41 for communication, but at 9:30, the data to be transmitted has been transmitted, then the first core 21 can issue a resource release request to release the occupied resources.
  • the resource lock request can be used to apply for prohibiting the release of certain resources for a certain period of time, which is equivalent to scheduling or locking the resource usage time, such as the first core 21 9:10 am in idle state or using the transmission path of antenna 512, antenna 522 and second storage subunit 42 to transmit data, if the first core needs to use antenna 511 and antenna at the next 10:11 to 11 o'clock 521 and the storage sub-unit 41 communicate, the first core 21 can request to use the antenna 511 and the antenna 521 and the storage sub-unit 41 to communicate at 10:11 to 11 o'clock through the resource lock request, and between 10:11 and 11 o'clock. The other cores cannot communicate using the antenna 511 and the antenna 521 and the storage subunit 41.
  • a determining sub-unit 64 configured to perform topology calculation on the transmission path according to the resource request and the resource status table, and determine between a kernel in the computing unit 20 and a storage sub-unit in the storage unit 40 Transmission path and transmission parameters of the antenna;
  • the topology calculation needs to comprehensively consider the resource request of the kernel and the resource status table of the storage subunit, and the ultimate purpose is to allocate an optimal transmission path for the kernel.
  • all possible transmission paths can be calculated first, such as the first core 21 - the antenna 511 - the antenna 521 - the first storage subunit 41; the first core 21 - the antenna 511 - the antenna 521 - the second storage The sub-unit 42; the first core 21 - the antenna 511 - the antenna 521 - the first storage sub-unit 41, etc., then allocates to the kernel the transmission path of the currently idle transmission path with the shortest transmission distance, the lowest transmission power consumption, and the best signal strength.
  • the adjusting subunit 65 is configured to adjust transmission parameters of the antenna according to the transmission path determined by the determining subunit 64 and the transmission parameters of the antenna to ensure the kernel in the computing unit 20 and the storage subunit in the storage unit 40 Complete the data transfer.
  • the transmission parameter of the antenna may include, but is not limited to: a number parameter of the antenna that needs to be docked in the transmission path determined by the determining subunit, a phase parameter of the antenna that needs to be docked, and a transmit power parameter of the antenna that needs to be docked .
  • the adjusting unit 70 can also receive the resource request sent by the multiple cores in the computing unit 20, perform dynamic link establishment according to the resource request and the resource occupation information to achieve reasonable allocation of resources, and improve transmission performance and efficiency. It should be noted that the adjustment unit 60 can be independently set and carried.
  • the first substrate 10 can also be carried on the second substrate 30, and connected to the computing unit 20 and the storage unit 40 by corresponding input and output pins.
  • the sub-units that are divided into functions can be respectively carried on the first substrate 10 and the second substrate 30.
  • the adjustment unit 60 can be divided into the receiving sub-unit 63 for receiving the resource request of the kernel in the computing unit, and
  • the receiving subunit is carried on the first substrate 10, and then the monitoring subunit 61 and the recording subunit 62 are configured to acquire resource occupation information of the storage subunit in the storage unit 40, and generate a resource status table, the monitoring subunit 61 and the recording subunit.
  • 62 can be carried on the second substrate 30, and then the determination sub-unit 64 calculates the transmission parameters of the transmission path and the antenna, and outputs the transmission parameters of the antenna to the adjustment sub-unit 65.
  • the determination sub-unit 64 can be disposed on the first substrate 10.
  • the adjustment subunit 65 can be disposed on the same substrate as the first antenna array 51 or the second antenna array 52, and different subunits can be connected through corresponding input and output pins.
  • the adjustment unit 60 can be independently arranged, disposed off-chip, or integrated in the chip, thereby reducing the volume of the motherboard.
  • the 4 is a schematic diagram of a composition of a 3D stacked device according to an embodiment of the present invention.
  • the core of the computing unit 10, CORE issues a memory access instruction to the memory DIE through the memory controller, and the memory control is performed.
  • the device may be integrated in the computing unit 10 or may be independently configured.
  • the storage unit 40 includes a Read-Only Memory (ROM) and a RAM, and specifically may store DIEs such as FLASH and Memory DIE.
  • ROM Read-Only Memory
  • RAM random access memory
  • DIEs such as FLASH and Memory DIE.
  • the 3D stacked device specifically includes:
  • a first substrate 10 a computing unit 20 composed of four COREs and a corresponding number of memory controllers, a second substrate 30, a memory unit 40 composed of 8 memory DIEs and 8 memory DIEs, and an antenna composed of 8 antennas
  • the TSV formation between them can be referred to as a TSV antenna, and the antenna array is in a high-impedance dielectric layer to reduce transmission power consumption.
  • memory DIE and the storage DIE may be carried on the same substrate or on different substrates, and the memory DIE and the storage DIE 10 ports may be connected to the antenna to achieve CORE data transmission.
  • Each TSV is used as a single stage antenna.
  • the antenna is integrated on-chip for on-chip communication.
  • the height of the TSV is optimized to be configurable at 140 GHz with a high radiation effect.
  • both sides of the first substrate and the second substrate are required to have TSV for connecting 10 circuits in 3D synthesis.
  • the multi-point to multi-point wireless interconnect solution applies to the interconnection between many cores and memory clusters with low loss and high configurability.
  • the first substrate 10 and the second substrate 30 are both high-impedance silicon substrates. It has an impedance of 750 Ohm, which is more than 100 times that of a normal P substrate. Therefore, the transmission loss is greatly reduced.
  • memory DIE and DIE 3D stacking can facilitate the virtualization of micro resources and achieve fine-grained hard partitioning.
  • the input of the dynamic interconnect layer between the first substrate 10 and the second substrate 30 is connected to the physical pins of each layer of the DIE, and the output of the memory controller connected to the CORE is output through the 3D TSV mode.
  • the adjustment unit 60 controls the dynamic interconnect layer, adjusts the adaptive switching of the hardware path in real time, and implements a combination of resource allocation specific CORE, specific memory DIE, and specific storage DIE.
  • the interface of the adjustment unit 60 and the upper layer CORE may include an input interface for receiving a resource request of the CORE; the interface with the dynamic interconnect layer may include a control bus, a built-in self-test (Built-in Self Test) test control line Data shift control lines, etc., can control the adjustment of the dynamic interconnect layer path.
  • the interface with the dynamic interconnect layer may include a control bus, a built-in self-test (Built-in Self Test) test control line Data shift control lines, etc., can control the adjustment of the dynamic interconnect layer path.
  • the adjustment unit 60 may establish a resource status table to include resource usage status information, resource usage information, and temperature information of the storage subunit, etc., which may serve as a basis for resource management.
  • the adjustment unit 60 allocates transmission parameters of the transmission path and the antenna according to the resource request of the CORE and the resource occupation information of the memory DIE, and adjusts the to-be-connected on the first substrate 10 and the second substrate 30 according to the transmission parameters of the antenna.
  • Antenna phase and transmit power, CORE can generate access commands according to application requirements, and can be transmitted to the wireless transceiver 10 port through the memory controller.
  • the wireless transceiver 10 driver can drive the wireless transceiver 10 port to pre-allocate the access command via the first substrate 10.
  • the TSV antenna is transmitted, and then received by the pre-assigned TSV antenna on the second substrate 30 and transmitted to the memory DIE via the wireless transceiver 10 port, and the memory DIE returns the data to the CORE through the above transmission path. data transmission.
  • the way CORE accesses the stored DIE is similar, and will not be described here.
  • the adjustment unit includes:
  • the monitoring sub-unit 61 is configured to acquire the resource occupation information of the storage sub-unit in the storage unit, and the recording sub-unit 62 is configured to generate a resource status table according to the resource occupation information acquired by the monitoring sub-unit 61, where the resource status is
  • the table includes at least one of resource usage status information, resource usage information, or temperature information of the storage subunit;
  • the receiving sub-unit 63 is configured to receive a resource request of the computing unit, where the resource request includes at least one of a resource request request, a resource release request, or a resource lock request;
  • a determining subunit 64 configured to perform a topological calculation on the transmission path according to the resource request and the resource status table, and determine a transmission path between a kernel in the computing unit and a storage subunit in the storage unit And transmission parameters of the antenna;
  • the adjusting subunit 65 is configured to adjust transmission parameters of the antenna according to the transmission path determined by the determining subunit and transmission parameters of the antenna to ensure that the kernel in the computing unit and the storage subunit in the storage unit complete data transmission .
  • the transmission parameter of the antenna may include, but is not limited to: a number parameter of the antenna that needs to be docked in the transmission path determined by the determining subunit, a phase parameter of the antenna that needs to be docked, and the antenna that needs to be docked Transmit power parameters.
  • the determining subunit 64 is specifically configured to send the transmission parameter to the adjusting subunit 65, so that the adjusting subunit 65 adjusts a phase, a transmitting power, and the like of the antenna corresponding to the antenna number according to the transmission parameter.
  • the antenna is composed of TSV, the angle is generally fixed. Therefore, when adjusting the phase of the antenna, the phase of the docking antenna can be adjusted according to the cooperation of the existing antenna array.
  • the adjustment unit includes:
  • the monitoring sub-unit 61 is configured to acquire the resource occupation information of the storage sub-unit in the storage unit, and the recording sub-unit 62 is configured to generate a resource status table according to the resource occupation information acquired by the monitoring sub-unit 61, where the resource status is The table includes at least one of resource usage status information, resource usage information, or temperature information of the storage subunit;
  • the receiving sub-unit 63 is configured to receive a resource request of the computing unit, where the resource request includes at least one of a resource request request, a resource release request, or a resource lock request;
  • a determining subunit 64 configured to perform a topological calculation on the transmission path according to the resource request and the resource status table, and determine a transmission path between a kernel in the computing unit and a storage subunit in the storage unit And transmission parameters of the antenna;
  • the adjusting subunit 65 is configured to adjust transmission parameters of the antenna according to the transmission path determined by the determining subunit and transmission parameters of the antenna to ensure that the kernel in the computing unit and the storage subunit in the storage unit complete data transmission .
  • the adjustment unit further includes:
  • a built-in self-test sub-unit 66 configured to test the transmission status of the transmission path and report to the determining sub-unit 64 so that the determining sub-unit 65 is a core in the computing unit and a storage sub-unit in the storage unit Allocating transmission paths;
  • a data moving subunit 67 configured to: move data in the storage unit according to a data moving instruction of the computing unit; cache data in data, or if the storage unit includes at least two storage subunits,
  • the embodiment of the present invention further includes a chip including the 3D stacked device according to any of the above embodiments of the present invention.
  • FIG. 7 is a schematic flowchart of a communication method applied to a 3D stacked device according to an embodiment of the present invention.
  • the method includes:
  • S702 Receive a resource request of the computing unit, where the resource request includes at least one of a resource request request, a resource release request, or a resource lock request.
  • S703 Perform topology calculation on the transmission path according to the resource request and the resource occupation information, and transmit parameters of the antenna, where the transmission parameter of the antenna includes a phase parameter.
  • the resource status table may be generated according to the acquired resource occupation information, where the resource status table includes resource usage status information, resource usage rate information, or temperature information of the storage subunit. At least one. After the resource list is generated, it is more convenient to call the unit for resource management, and can also be displayed to the user by configuring the display.
  • the transmission parameters of the antenna may also include but are not limited to:
  • the present invention has the following advantages:
  • connection dynamic resource allocation according to application requirements, single port replacement multi-port mechanical switching mode, no switch switching, fast transmission rate, multi-input and multi-output effects, frequency division multiplexing and time division multiplexing.
  • the storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

Landscapes

  • Engineering & Computer Science (AREA)
  • Microelectronics & Electronic Packaging (AREA)
  • Physics & Mathematics (AREA)
  • Condensed Matter Physics & Semiconductors (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Hardware Design (AREA)
  • Power Engineering (AREA)
  • Variable-Direction Aerials And Aerial Arrays (AREA)

Abstract

本发明实施例提供了一种3D堆叠器件,包括:第一基板,用于承载计算单元;第二基板,用于承载存储单元;第一天线阵列,位于第一基板上,与计算单元电连接,且指向第二基板,用于发射计算单元输出的数据和/或接收第二天线阵列发射的数据;第二天线阵列,位于第二基板上,与存储单元电连接,且指向第一基板,用于接收第一天线阵列发射的数据和/或接收存储单元输出的数据;调节单元,用于调节第一天线阵列和/或第二天线阵列中的天线的传输参数,以确保计算单元中的内核与存储单元中的存储子单元之间完成数据传输,本发明实施例还提供了一种芯片及通信方法,采用本发明,可提升芯片内部的传输带宽,且传输通路可根据应用需求进行动态分配。

Description

一种 3D堆叠器件、 芯片及通信方法
技术领域
本发明涉及电子技术领域, 尤其涉及一种 3D堆叠器件、 芯片及通信方法。 背景技术
随着用户对终端轻便化的要求越来越高, 如手机、 平板电脑、 笔记本或电 视等, 基于方便携带或美观的考虑, 现有终端的机身厚度越来越薄。 这就要求 终端内部的集成芯片功能越来越丰富, 且体积越来越小。 而在集成芯片内部, 计算单元和存储单元作为最重要的组成部分, 其互连方式对集成芯片的体积影 响很大。
在现有技术中, 有线互连技术包含片内的 3D封装、 2.5D封装以及片外互 连结构。 其中, 3D封装, 是指在不改变封装体尺寸的前提下, 在同一个封装体 内于垂直方向叠放两个以上芯片的封装技术, 从而可实现存储容量的倍增; 由 于将芯片直接互连, 互连线长度显著缩短, 信号传输得更快且所受干扰更小; 再次, 可将多个不同功能芯片堆叠在一起, 使单个封装体实现更多的功能; 最 后, 釆用 3D封装的芯片还有功耗低、 速度快等优点, 可使得电子信息产品的尺 寸和重量减小数十倍。 因此, 3D封装技术正被越来越广泛的应用。 但是, 对于 具备至少两个内核 (Core ) 的计算单元与具备至少两个存储子单元的存储单元 而言, 现有的 3D封装技术中, 每个内核与每个存储子单元之间都必须存在固定 的输入输出通路, 每个内核通过硅通孔(Through Si Via, 简称 TSV ) 与存储子 单元连接, 通过机械开关切换, 由选择器选择传输通路。 但是机械开关的速率 有限, 有线传输的带宽也有限, 且由于输入输出通路较多, 需要预留多个 10 ( in/out, 简称 10 )端口, 走线较多, 导致互联结构复杂, 集成度有限, 芯片体 积较大, 计算和存储融合的数据存储系统架构不灵活, 无法根据应用需求进行 动态资源分配, 另开关切换时仅单通道存在数据传输, 不能充分发挥芯片的性 能。 发明内容 本发明实施例提供了一种 3D堆叠器件、 芯片及通信方法, 可解决芯片内部 传输带宽有限, 传输通路无法根据应用需求进行动态分配的问题。
本发明实施例第一方面提供了一种 3D堆叠器件, 可包括:
第一基板, 用于承载计算单元, 所述计算单元包括一个内核;
第二基板, 用于承载存储单元, 所述存储单元包括一个存储子单元; 第一天线阵列, 位于所述第一基板上, 与所述计算单元电连接, 且指向所 述第二基板, 用于发射所述计算单元输出的数据和 /或接收第二天线阵列发射的 数据;
所述第二天线阵列, 位于所述第二基板上, 与所述存储单元电连接, 且指 向所述第一基板, 用于接收所述第一天线阵列发射的数据和 /或接收所述存储单 元输出的数据;
调节单元, 用于调节所述第一天线阵列和 /或第二天线阵列中的天线的传输 之间完成数据传输, 其中, 所述天线的传输参数包括相位参数。
在第一方面的第一种可能的实现方式中, 所述第一天线阵列中的天线为所 述第一基板上指向所述第二基板且不与所述第二基板接触的硅通孔, 所述第二 天线阵列中的天线为所述第二基板上指向所述第一基板且不与所述第一基板接 触的石圭通孔。
结合第一方面或结合第一方面的第一种可能的实现方式, 在第二种可能的 实现方式中, 所述调节单元还用于:
获取所述存储单元中的所述存储子单元的资源占用信息, 接收所述计算单 元的资源请求, 根据所述资源请求和所述资源占用信息调整所述计算单元中的 结合第一方面的第二种可能的实现方式, 在第三种可能的实现方式中, 所 述调节单元包括:
监控子单元, 用于获取所述存储单元中所述存储子单元的资源占用信息; 记录子单元, 用于根据所述监控子单元获取的资源占用信息, 生成资源状 态表, 所述资源状态表包括资源使用状态信息、 资源使用率信息或所述存储子 单元的温度信息中的至少一种; 接收子单元, 用于接收所述计算单元的资源请求, 所述资源请求包括资源 申请请求、 资源释放请求或资源锁定请求中的至少一种;
确定子单元, 用于根据所述资源请求和所述资源状态表, 对传输通路进行 拓朴计算, 确定所述计算单元中的所述内核和所述存储单元中的所述存储子单 元之间的传输通路及天线的传输参数;
调整子单元, 用于根据所述确定子单元确定的传输通路和天线的传输参数, 调整天线的传输参数以确保所述计算单元中的所述内核与所述存储单元中的所 述存储子单元完成数据传输。
结合第一方面的第三种可能的实现方式, 在第四种可能的实现方式中, 所 述天线的传输参数还包括:
所述确定子单元确定的传输通路中需要对接的天线的编号参数和所述需要 对接的天线的发射功率参数。
结合第一方面的第三或第四种可能的实现方式, 在第五种可能的实现方式 中, 所述调节单元还包括: 以便所述确定子单元为所述计算单元中的所述内核和所述存储单元中的所述存 储子单元分配传输通路。
结合第一方面的第五种可能的实现方式, 在第六种可能的实现方式中, 所 述调节单元还包括:
数据搬移子单元, 用于根据所述计算单元的数据搬移指令, 搬移所述存储 单元中的数据;
緩冲子单元, 用于在所述数据搬移子单元搬移所述存储单元中的数据时緩 存数据, 或者若所述存储单元包括至少两个存储子单元, 则在所述计算单元调 结合第一方面或结合第一方面的第一或第二或第三或第四或第五或第六种 可能的实现方式, 在第七种可能的实现方式中, 所述第一天线阵列和所述第二 天线阵列处于阻抗值达到预设阔值的介质中。
结合第一方面或结合第一方面的第一或第二或第三或第四或第五或第六或 第七种可能的实现方式, 在第八种可能的实现方式中, 所述第一天线阵列和所 述第二天线阵列以近场通信的方式进行数据传输。 本发明实施例第二方面提供了一种芯片, 可包括:
如本发明实施例第一方面或第一方面任一实现方式所述的 3D堆叠器件。 本发明实施例第三方面提供了一种通信方法, 应用于如本发明实施例第一 方面或第一方面任一实现方式所述的 3D堆叠器件上, 可包括:
获取所述存储单元中存储子单元的资源占用信息;
接收所述计算单元的资源请求, 所述资源请求包括资源申请请求、 资源释 放请求或资源锁定请求中的至少一种;
根据所述资源请求和所述资源占用信息, 对传输通路进行拓朴计算, 确定 的传输参数, 所述天线的传输参数包括相位参数;
根据确定的传输通路和天线的传输参数, 调整天线的传输参数以确保所述 在第三方面的第一种可能的实现方式中, 在所述根据所述资源请求和所述 资源占用信息, 对传输通路进行拓朴计算, 确定所述计算单元中的内核和所述 根据获取的资源占用信息, 生成资源状态表, 所述资源状态表包括资源使 用状态信息、 资源使用率信息或所述存储子单元的温度信息中的至少一种。
结合第三方面或结合第三方面的第一种可能的实现方式, 在第二种可能的 实现方式中, 所述天线的传输参数还包括:
编号参数和发射功率参数。
实施本发明实施例, 具有如下有益效果:
通过在计算单元和存储单元之间配置可实现无线通信的第一天线阵列和第
资源分配, 传输速率快。 附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案, 下面将对实施 例或现有技术描述中所需要使用的附图作简单地介绍, 显而易见地, 下面描述 中的附图仅仅是本发明的一些实施例, 对于本领域普通技术人员来讲, 在不付 出创造性劳动的前提下, 还可以根据这些附图获得其他的附图。
图 1是本发明实施例提供的一种 3D堆叠器件的组成示意图;
图 2是本发明实施例提供的另一种 3D堆叠器件的组成示意图;
图 3是本发明实施例提供的又一种 3D堆叠器件的组成示意图;
图 4是本发明实施例提供的又一种 3D堆叠器件的组成示意图;
图 5是本发明实施例提供的 3D堆叠器件中的一种调节单元的组成示意图; 图 6是本发明实施例提供的 3D堆叠器件中的另一种调节单元的组成示意图; 图 7是本发明实施例提供的应用于 3D堆叠器件的通信方法的流程示意图。 具体实施方式
下面将结合本发明实施例中的附图, 对本发明实施例中的技术方案进行清 楚、 完整地描述, 显然, 所描述的实施例仅仅是本发明一部分实施例, 而不是 全部的实施例。 基于本发明中的实施例, 本领域普通技术人员在没有作出创造 性劳动的前提下所获得的所有其他实施例, 都属于本发明保护的范围。
请参照图 1, 为本发明实施例提供的一种 3D堆叠器件的组成示意图, 在本 实施例中, 所述 3D堆叠器件包括:
第一基板 10, 用于承载计算单元 20, 所述计算单元包括一个内核 21 ;
第二基板 30,用于承载存储单元 40,所述存储单元包括一个存储子单元 41 ; 第一天线阵列 51, 位于所述第一基板 10上, 与所述计算单元 20电连接, 且指向所述第二基板 30, 用于发射所述计算单元 20输出的数据和 /或接收第二 天线阵列 52发射的数据;
所述第二天线阵列 52, 位于所述第二基板 30上, 与所述存储单元 40电连 接, 且指向所述第一基板 10, 用于接收所述第一天线阵列 51发射的数据和 /或 接收所述存储单元 40输出的数据;
调节单元 60, 用于调节所述第一天线阵列 51和 /或第二天线阵列 52中的天 线的传输参数, 以确保所述计算单元 20中的所述内核 21与所述存储子单元 40 中的所述存储子单元 41之间完成数据传输, 其中, 所述天线的传输参数可以包 括但不限于相位参数。 例如, 所述天线的传输参数还可以包括对接的天线的编 号参数、 对接的天线的发射功率参数等。
计算单元 20发出的电信号可以是内核 21发出的指令或输出的电信号, 可 以由第一天线阵列 51中的天线转换为射频信号发射出去, 然后由第二天线阵列 52中的天线接收并将射频信号重新转换为电信号输出至存储单元 40, 最后传递 至存储子单元 41。
存储子单元 41可以是单一的存储元件, 也可以由多个互连的存储元件组成 的存储簇, 如多个内存棵片 ( DIE )组成的内存簇或多个存储 DIE如 FLASH组 成的存储簇。
所述调节单元 60在调节所述第一天线阵列 51和 /或第二天线阵列 52中的天 线的传输参数时, 可以仅调节第一天线阵列 51中的天线的传输参数, 或者仅调 节第二天线阵列 52中的天线的传输参数, 当然, 也可以同时调节两个天线阵列 中的天线的传输参数, 只需要确保计算单元 20中的所述内核 21与存储单元 40 中的所述存储子单元 41可完成数据传输即可。
可选地, 所述第一天线阵列 51 中的天线可以为所述第一基板 10上指向所 述第二基板 30且不与所述第二基板 30接触的硅通孔, 所述第二天线阵列 52中 的天线为所述第二基板 30上指向所述第一基板 10且不与所述第一基板 10接触 的硅通孔。 这样, 无需再为计算单元 20和存储单元 40配置额外的天线, 节省 了资源, 还可以提升天线的集成度, 减少 3D堆叠器件和芯片的体积和成本。
所述调节单元 60可分为两个调节子单元, 分别与第一天线阵列 51和第二 天线阵列 52连接以调节天线的传输参数, 此时, 两个调节子单元分别承载于第 一基板 10和第二基板 30上, 当然, 所述调节单元 60也可以单独位于第一基板 10或第二基板 30上,然后将所述调节单元 60的控制端与第一天线阵列 51和第 二天线阵列 52中的天线分别连接即可。
通过在计算单元和存储单元之间配置可实现无线通信的第一天线阵列和第
资源分配, 传输速率快。 请参照图 2, 为本发明实施例提供的一种 3D堆叠器件的组成示意图, 在本 实施例中, 所述 3D堆叠器件包括:
第一基板 10, 用于承载计算单元 20, 所述计算单元 20包括第一内核 21和 第二内核 22;
第二基板 30, 用于承载存储单元 40, 所述存储单元 40包括第一个存储子 单元 41和第二存储子单元 42;
第一天线阵列 51, 位于所述第一基板 10上, 与所述计算单元 20电连接, 且指向所述第二基板 30, 用于发射所述计算单元 20输出的数据和 /或接收第二 天线阵列 52发射的数据;
所述第二天线阵列 52, 位于所述第二基板 30上, 与所述存储单元 40电连 接, 且指向所述第一基板 10, 用于接收所述第一天线阵列 51发射的数据和 /或 接收所述存储单元输出的数据;
其中, 所述第一天线阵列 51由天线 511和天线 512组成, 所述第二天线阵 列 52由天线 521和天线 522组成, 所述第一天线阵列 51和第二天线阵列 52配 合使用便得到天线阵列 50。
调节单元 60, 用于调节所述第一天线阵列 51和 /或第二天线阵列 52中的天 线的传输参数, 以确保所述计算单元 20中的内核与所述存储单元 40中的存储 子单元完成数据传输, 其中, 所述天线的传输参数包括但不限于相位参数。 例 如, 所述天线的传输参数还可以包括对接的天线的编号参数、 对接的天线的发 射功率参数等。
计算单元 20发出的电信号可以由第一天线阵列 51 中的天线转换为射频信 号发射出去, 然后由第二天线阵列 52中的天线接收并将射频信号重新转换为电 信号输出至存储单元 40。
存储子单元可以是单一的存储元件, 也可以由多个互连的存储元件组成的 存储簇, 如多个内存棵片 ( DIE )组成的内存簇或多个存储 DIE如 FLASH组成 的存储簇。
所述调节单元 60在调节所述第一天线阵列 51和 /或第二天线阵列 52中的天 线的传输参数时, 可以仅调节第一天线阵列 51中的天线的传输参数, 或者仅调 节第二天线阵列 52中的天线的传输参数, 当然, 也可以同时调节两个天线阵列 中的天线的传输参数, 只需要确保计算单元 20中的内核与存储单元 40中的存 储子单元可完成数据传输即可。
例如, 当第一内核 21需要访问第二存储子单元 42中的数据时, 若此时第 一天线阵列 51和第二天线阵列 52中的天线在频率或时间上均存在空闲, 则第 一内核 21可以主动请求通过天线 511和天线 522完成数据传输, 此时, 可以仅 调节天线 511的传输参数如相位等使得天线 511和天线 522对接, 完成数据传 输, 也可以仅调节天线 522的传输参数使得天线 522和天线 511对接, 完成数 据传输, 还可以同时调节天线 511和天线 522的传输参数使得天线 511和天线 522对接, 完成数据传输, 且对于天线阵列中的天线相位的调节, 还可以参照现 有技术中天线阵列的相位加权方式实现特定天线的相位调节, 即通过对阵列中 两个或以上的天线相位的调节 (主瓣和副瓣的配合等) 来实现特定天线的特定 相位调节。
当某内核或某应用的需求发生变化时, 所述调节单元 60可通过调节天线的 相位为该内核或应用建立合适的动态链路以满足内核或应用的需求, 调节的天 线的传输参数如对接的天线编号参数和天线相位参数可预先设定或人工输入, 或者还可以由计算单元 20根据存储单元 40的资源占用信息进行选择, 当然还 可以配置调节单元 60综合考虑内核的需求以及存储子单元的存储情况进行自动 计算和分配。 例如釆用预先设定的方式时, 可首先设定某内核与不同存储子单 元的传输优先级及相应的对接天线和相位信息, 当该内核需要进行数据传输时, 便可以根据预先设定的传输优先级、 对接天线和相位信息完成数据调用, 若传 输优先级最高的存储子单元无法完成数据传输时, 便可以尝试与优先级为次高 级的存储子单元完成数据传输。 而配置调节单元 60综合考虑内核的需求以及存 储子单元的存储情况进行自动计算和分配则可以更好完成资源的自适应匹配, 使得不同内核和不同存储子单元实现最佳的适配。 所述调节单元 60可分为两个 调节子单元, 分别与第一天线阵列 51和第二天线阵列 52连接以调节天线的传 输参数, 此时, 两个调节子单元分别承载于第一基板 10和第二基板 30上, 当 然, 所述调节单元 60也可以单独位于第一基板 10或第二基板 30上, 然后将所 述调节单元 60的控制端与第一天线阵列 51和第二天线阵列 52中的天线分别连 接即可。
通过天线的传输参数调节便可以实现无线动态链路的建立, 传输速率高。 且天线可支持一对多传输, 如计算单元的一个内核可以以广播的方式与多个存 储子单元通信, 单一的存储子单元也可以将数据同时发送给多个内核进行计算 处理, 从而实现多输入多输出的效果。 而输入输出通路只有在需要进行数据传 输时才建立, 无需选择器进行通路选择, 互连结构简单, 节省了 10端口和芯片 体积。 同时还可以兼容互补金属氧化物半导体 ( Complementary Metal Oxide Semiconductor, 简称 CMOS ) 工艺, 而内核较多时, 可共用同一组天线在不同 频率或不同时间进行通信, 从而实现频分复用和时分复用。 例如, 多个内核想 同时使用天线 511与天线 521与第一存储子单元 41进行通信, 则可以为多个内 核配置不同的通信频率, 则此时多个内核可使用不同的频率同时使用天线 511 和天线 521实现与第一存储子单元 41通信,当然,也可以使用相同的通信频率, 但错开多个内核使用天线 511与天线 521与该存储子单元通信的时间, 如配置 第一内核 21在 9点 -10点之间通过天线 511使用某频率发射信号, 然后由天线 521接收该信号并转换为电信号输出至第一存储子单元 41, 而第二内核在 10点 -11点再通过天线 511使用该频率发射信号,实现与第一存储子单元 41的通信。
可选地, 所述计算单元 20中的内核可通过无线收发驱动 10电路与第一天 线阵列 51 中的天线进行配合使用, 所述存储单元 40中的存储子单元同样可通 过无线收发驱动 10电路与第二天线阵列 52中的天线进行配合使用。
可选地, 所述第一天线阵列 51 中的天线为所述第一基板 10上指向所述第 二基板 30且不与所述第二基板 30接触的硅通孔, 所述第二天线阵列 52中的天 线为所述第二基板 30上指向所述第一基板 10且不与所述第一基板 10接触的硅 通孔。 这样, 无需再为计算单元 20和存储单元 40配置额外的天线, 节省了资 源, 还可以提升天线的集成度, 减少 3D堆叠器件和芯片的体积和成本。
可选地, 所述第一基板 10和所述第三基板 30可使用高阻抗基板, 其阻抗 值可达到 750Ohm,是普通 P基板的 100倍, 所述第一天线阵列和所述第二天线 阵列处于阻抗值达到预设阔值的介质中。 如阻抗值达到预设阔值如 750Ohm的 介质中。从而可大大减小第一天线阵列 51与第二天线阵列 52之间的传输损耗。
可选地, 所述第一天线阵列 51和所述第二天线阵列 52可以近场通信的方 式进行数据传输。 因为是超高频段的近场通信, 片上面积和功耗负荷如天线辐 射功率非常小, 因此具备面积小, 功耗低, 带宽高的显著优点。
需要说明的是, 在本实施例中, 所述计算单元 20仅包括两个内核, 所述存 储单元 40仅包括两个存储子单元, 由于现有电子芯片技术的飞速发展, 4核、 8 核甚至更多内核的芯片都已被广泛应用, 存储子单元的情况类似。 且本发明在 内核越多时优势越明显, 但鉴于描述的便利性, 本实施例仅以双内核的场景进 行举例说明, 多内核的场景下原理类似。 此外, 单内核和单存储子单元的通信 同样可以釆用本实施例中所述的无线通信方式来实现, 具体可参照图 1 所示的 实施例及其相关描述, 此处不再赘述。
在本实施例中, 通过在计算单元和存储单元之间配置可实现无线通信的第 参数以确保计算单元中的内核与存储单元中的存储子单元完成数据传输, 从而 用需求进行动态资源分配,单端口替换多端口的机械切换方式,无需开关切换, 传输速率快, 且可以实现多输入多输出的效果, 支持频分复用和时分复用。
请参照图 3, 为本发明实施例提供的另一种 3D堆叠器件的组成示意图, 在 本实施例中, 所述 3D堆叠器件包括:
第一基板 10, 用于承载计算单元 20, 所述计算单元 20包括第一内核 21和 第二内核 22;
第二基板 30, 用于承载存储单元 40, 所述存储单元 40包括第一个存储子 单元 41和第二存储子单元 42;
第一天线阵列 51, 位于所述第一基板 10上, 与所述计算单元 20电连接, 且指向所述第二基板 30, 用于发射所述计算单元 20输出的数据和 /或接收第二 天线阵列 52发射的数据;
所述第二天线阵列 52, 位于所述第二基板 30上, 与所述存储单元 40电连 接, 且指向所述第一基板 10, 用于接收所述第一天线阵列 51发射的数据和 /或 接收所述存储单元输出的数据;
其中, 所述第一天线阵列 51由天线 511和天线 512组成, 所述第二天线阵 列 52由天线 521和天线 522组成, 所述第一天线阵列 51和第二天线阵列 52配 合使用便得到天线阵列 50。
调节单元 60, 用于调节所述第一天线阵列 51和 /或第二天线阵列 52中的天 线的传输参数, 以确保所述计算单元 20中的内核与所述存储单元 40中的存储 子单元完成数据传输, 其中, 所述天线的传输参数包括但不限于相位参数。 所述调节单元 60还用于:
获取所述存储单元 40中存储子单元的资源占用信息,接收所述计算单元 20 的资源请求, 根据所述资源请求和所述资源占用信息调整所述计算单元 20中的 其中, 调节单元 60在综合考虑资源请求和资源占用信息时, 可以为内核和 /或存储子单元配置相应的优先级, 根据优先级来进行动态调整, 如第一内核 21 处理的事件较紧急, 则可以为其配置较高的优先级如 9, 而第二内核 22优先级 较低为 8, 存储子单元同样可配置优先级, 如将第一存储子单元 41配置优先级 为 7, 第二存储子单元配置优先级为 6, 则第一内核 21和第二内核 22同时请求 使用天线 511和天线 521向第一存储子单元 41传输数据时, 则优先满足第一内 核 21, 为第一内核 21配置第一内核 21—天线 511—天线 521—第一存储子单元 41 的传输通路, 此时, 可为第二核内核 22配置第二内核 22—天线 512—天线 522—第二存储子单元 42的传输通路, 或者也可以等待第一内核 21数据传输完 成后或者在第一内核 21发出资源释放请求后, 第二内核 22再使用第二内核 22 一天线 511—天线 521—第一存储子单元 41的传输通路。 若第一内核 21发出资 源锁定请求, 请求锁定天线 511、 天线 521和第一存储子单元 41在 10点 -11点 不被占用时,则第二内核 22发出资源申请请求,申请在 10点 -11点使用天线 511、 天线 521和第一存储子单元 41通信时, 则天线 511、 天线 521和第一存储子单 元 41将不会被分配给第二内核 22, 当然若第二内核 22的优先级较高, 则可以 优先满足第二内核 22的需求。
而若存储子单元的优先级较高时, 则在动态调整时可以优先考虑存储子单 元的资源占用信息如资源使用状态、 资源使用率或存储子单元的温度信息等。 例如, 第一内核 21请求使用天线 511和天线 521向第一存储子单元 41传输数 据时, 若此时第一存储子单元 41的资源使用状态为正在使用, 或者第一存储子 单元 41的资源使用率如容量已使用达到 95%, 或者第一存储子单元 42由于同 时与多个内核通信或其他原因导致温度达到 60摄氏度时, 则调节单元 60可以 将为第一内核 21分配第一内核 21—天线 511—天线 521—第二存储子单元 42的 传输通路, 或者等待直至第一存储子单元 41的资源占用信息到达警戒阔值以下 时再为第一内核 21分配第一内核 21—天线 511—天线 521—第一存储子单元 41 的传输通路。 需要说明的是, 此处涉及的各项警戒阔值如 95%和 60摄氏度仅用 于举例说明, 不对本发明实施例的具体应用和配置进行任何限定。
当然,除了配置优先级之外,还可以均衡的考虑资源请求和资源占用信息, 此时可理解为各个内核和存储子单元的优先级均相同, 只需要根据时序尽量为 内核或存储子单元分配最佳的传输通路。如第一内核 21请求通过天线 511和 521 与第一存储子单元 41通信, 若天线 511已被占用, 则优先考虑分配当前空闲的 天线, 若第一存储子单元 41 已无空闲存储空间, 则为第一内核 21分配具备空 闲存储空间的存储子单元, 若第一存储子单元 41的资源使用状态或资源使用率 或温度任意项达到警戒阔值, 则可以为第一内核 21分配其他存储子单元如第二 存储子单元; 若第一内核 21与第二内核 22同时请求通过天线 511和 521与第 一存储子单元 41通信, 则可以为第一内核 21分配距离较近的天线 511和天线 521,实现与第一存储子单元 41通信,而为第二内核 22分配距离较近的天线 512 和 522, 实现与第二存储子单元 42通信。
可选地, 所述调节单元 60可以包括但不限于:
监控子单元 61, 用于获取所述存储单元中存储子单元的资源占用信息; 记录子单元 62, 用于根据所述监控子单元 61获取的资源占用信息, 生成资 源状态表, 所述资源状态表包括资源使用状态信息、 资源使用率信息或所述存 储子单元的温度信息中的至少一种;
资源使用状态可以包括但不限于已使用和未使用, 或者还可以是被预定使 用状态 (被某内核预定在某段时间使用)等; 资源使用率可以是当前存储子单 元的使用率,如 1G容量的随机存取存储器( Random Access Memory,简称 RAM ), 已使用 512M, 则使用率为 50%, 可继续建立链路使用, 若使用率超过 90%则可 以考虑其他存储子单元建立链路; 存储子单元的温度信息可以表明当前存储子 单元的温度, 若超过一定温度阔值则可以选择其他存储子单元建立链路。
接收子单元 63, 用于接收所述计算单元 20的资源请求, 所述资源请求包括 资源申请请求、 资源释放请求或资源锁定请求中的至少一种;
资源申请请求可用于申请使用资源, 如第一内核 21可通过资源申请请求申 请在某段时间或某个频率使用天线 511和天线 521以及存储子单元 41进行通信; 资源释放请求可用于申请释放当前占用的资源, 如第一内核 21开始申请在 9点 -10点使用天线 511和天线 521以及存储子单元 41进行通信,但在 9点 30分时, 需要传输的数据已经传输完成, 则此时第一内核 21可发出资源释放请求以释放 占用的资源, 以便其他内核使用天线 511和天线 521以及存储子单元 41进行通 信; 资源锁定请求可用于申请在某段时间内禁止释放某些资源, 相当于预定或 锁定资源使用时间, 如第一内核 21在 9点 -10点处于空闲态或正使用天线 512、 天线 522以及第二存储子单元 42的传输通路传输数据, 若第一内核在接下来的 10点 -11点的时间需要使用天线 511和天线 521以及存储子单元 41进行通信, 则第一内核 21可通过资源锁定请求,请求在 10点 -11点使用天线 511和天线 521 以及存储子单元 41进行通信,则在 10点 -11点之间,其他内核无法使用天线 511 和天线 521以及存储子单元 41进行通信。
确定子单元 64, 用于根据所述资源请求和所述资源状态表, 对传输通路进 行拓朴计算, 确定所述计算单元 20中的内核和所述存储单元 40中的存储子单 元之间的传输通路及天线的传输参数;
可选地, 拓朴计算需综合考虑内核的资源请求以及存储子单元的资源状态 表, 其最终目的是为内核分配最佳的传输通路。 在拓朴计算时, 可先将所有可 能的传输通路计算出来,如第一内核 21—天线 511—天线 521—第一存储子单元 41 ; 第一内核 21—天线 511—天线 521—第二存储子单元 42; 第一内核 21—天 线 511—天线 521—第一存储子单元 41等, 然后为内核分配当前空闲的传输通 路中传输距离最短、 传输功耗最低且信号强度最佳的传输通路。
调整子单元 65,用于根据所述确定子单元 64确定的传输通路和天线的传输 参数,调整天线的传输参数以确保所述计算单元 20中的内核与所述存储单元 40 中的存储子单元完成数据传输。
所述天线的传输参数可以包括但不限于: 所述确定子单元确定的传输通路 中需要对接的天线的编号参数、 所述需要对接的天线的相位参数和所述需要对 接的天线的发射功率参数。
且调节单元 70还可以接收计算单元 20中多个内核发出的资源请求, 根据 资源请求和上述的资源占用信息来进行动态链路的建立以实现资源的合理分配, 提升传输的性能和效率。 需要说明的是, 所述调节单元 60可以独立设置, 承载 于所述第一基板 10上, 也可以承载于所述第二基板 30上, 釆用相应的输入输 出引脚与计算单元 20和存储单元 40分别连接。 还可以按功能划分为独立的子 单元分别承载于第一基板 10和第二基板 30上, 例如可将所述调节单元 60划分 为接收子单元 63用于接收计算单元中内核的资源请求, 且接收子单元承载于第 一基板 10上, 然后划分出监控子单元 61和记录子单元 62获取存储单元 40中 存储子单元的资源占用信息, 并生成资源状态表, 监控子单元 61和记录子单元 62可承载于第二基板 30上, 然后划分出确定子单元 64计算传输通路和天线的 传输参数, 并输出天线的传输参数至调整子单元 65, 确定子单元 64可配置在第 一基板 10上、 第二基板 30或其他基板上, 调整子单元 65可与所述第一天线阵 列 51或第二天线阵列 52配置在同一基板上, 不同子单元之间通过相应的输入 输出引脚连接即可。 当然, 所述调节单元 60可以独立设置, 配置在片外, 也可 以集成在片内, 从而减少主板体积。
请参照图 4, 为本发明实施例提供的又一种 3D堆叠器件的组成示意图, 在 本实施例中,计算单元 10中的内核即 CORE通过内存控制器向内存 DIE发出内 存访问指令, 内存控制器可集成在计算单元 10中, 也可以独立设置, 存储单元 40包括只读存储记忆体( Read-Only Memory, 简称 ROM )和 RAM, 具体可为 存储 DIE如 FLASH和内存 DIE。 由此以第一基板 10和第二基板 30为界, 由上 至下依次可形成计算层、 动态互连层和存储层。
所述 3D堆叠器件具体包括:
第一基板 10、由 4个 CORE以及数量对应的内存控制器组成的计算单元 20、 第二基板 30、 由 8个内存 DIE和 8个存储 DIE组成的存储单元 40、 由 8个天线 组成的天线阵列 50、 调节单元 60以及用于驱动 CORE和内存 DIE无线收发 10 端口的无线收发 10驱动 70, 其中, 两个或以上的内存 DIE可构成内存簇, 天 线由第一基板 10和第二基板 30之间的 TSV形成, 可称为 TSV天线, 且天线阵 列处于高阻抗介质层以降低传输功耗。
需要说明的是, 内存 DIE和存储 DIE可以承载于同一基板上, 也可以承载 于不同的基板上, 内存 DIE和存储 DIE的 10端口均可以与天线连接以实现和 CORE的数据传输。
每个 TSV被用作单级天线。 天线集成在片内用于片内通信。 TSV的高度经 过优化可配置在 140GHz具有较高的辐射效应。且第一基板和第二基板的两侧都 需要具有 TSV, 用于在 3D合成时连接 10电路。
多点到多点的无线互连方案应用在众核与内存簇之间的互连, 具有低损耗 和高可配置性。
任何一个内核可以自由访问任何一个内存簇, 可实现 M-to-N的网络要求。 第一基板 10和第二基板 30均为高阻抗硅基板。 具有 750Ohm的阻抗, 是 普通 P基板的 100倍以上。 因此传输损耗大幅降低。
而不同 CORE、 内存 DIE和存储 DIE的 3D堆叠, 可便于微观资源的虚拟 化, 实现细粒度的硬分区。
第一基板 10和第二基板 30之间的动态互连层的输入连接各层 DIE的物理 管脚, 通过 3D TSV的方式, 输出连接 CORE的内存控制器的存储 10上。
调节单元 60可控制动态互连层, 实时调整硬件通路的自适应切换, 实现资 源分配特定 CORE、 特定内存 DIE和特定存储 DIE的组合。
调节单元 60与上层 CORE的接口可包括输入接口,用于接收 CORE的资源 请求; 与动态互连层的接口可包括控制总线、 内建自测(Built-in Self Test, 检测 BIST )测试控制线、 数据搬移控制线等, 可控制动态互连层路径的调整。
调节单元 60可建立资源状态表以包括资源使用状态信息、 资源使用率信息 和存储子单元的温度信息等, 这些信息可作为资源管理的判断依据。
在数据传输之前,由调节单元 60根据 CORE的资源请求以及内存 DIE的资 源占用信息分配传输通路和天线的传输参数, 并根据天线的传输参数调节第一 基板 10和第二基板 30上的待对接天线的相位和发射功率等, CORE可根据应用 需求产生访问指令, 经过内存控制器传输至无线收发 10端口, 无线收发 10驱 动可驱动无线收发 10端口将访问指令经第一基板 10上预先分配好的 TSV天线 发射出去, 然后由第二基板 30上预先分配好的 TSV天线接收并经无线收发 10 端口传输至内存 DIE, 内存 DIE再将数据经上述传输通路返回给 CORE, 完成 数据传输。 CORE访问存储 DIE的方式类似, 此处不再赘述。
请参照图 5, 为本发明实施例提供的 3D堆叠器件中的一种调节单元的组成 示意图, 在本实施例中, 所述调节单元包括:
监控子单元 61, 用于获取所述存储单元中存储子单元的资源占用信息; 记录子单元 62, 用于根据所述监控子单元 61获取的资源占用信息, 生成资 源状态表, 所述资源状态表包括资源使用状态信息、 资源使用率信息或所述存 储子单元的温度信息中的至少一种;
接收子单元 63, 用于接收所述计算单元的资源请求, 所述资源请求包括资 源申请请求、 资源释放请求或资源锁定请求中的至少一种;
确定子单元 64, 用于根据所述资源请求和所述资源状态表, 对传输通路进 行拓朴计算, 确定所述计算单元中的内核和所述存储单元中的存储子单元之间 的传输通路及天线的传输参数;
调整子单元 65, 用于根据所述确定子单元确定的传输通路和天线的传输参 数, 调整天线的传输参数以确保所述计算单元中的内核与所述存储单元中的存 储子单元完成数据传输。
可选地, 所述天线的传输参数可以包括但不限于: 所述确定子单元确定的 传输通路中需要对接的天线的编号参数、 所述需要对接的天线的相位参数和所 述需要对接的天线的发射功率参数。
所述确定子单元 64具体可用于将所述传输参数发送给所述调整子单元 65, 以便所述调整子单元 65根据所述传输参数调节所述天线编号对应的天线的相位 和发射功率等。
由于天线由 TSV组成,其角度一般固定不变,因此具体在调节天线相位时, 可根据现有的天线阵列的配合来调整对接天线的相位。
请参照图 6, 为本发明实施例提供的 3D堆叠器件中的另一种调节单元的组 成示意图, 在本实施例中, 所述调节单元包括:
监控子单元 61, 用于获取所述存储单元中存储子单元的资源占用信息; 记录子单元 62, 用于根据所述监控子单元 61获取的资源占用信息, 生成资 源状态表, 所述资源状态表包括资源使用状态信息、 资源使用率信息或所述存 储子单元的温度信息中的至少一种; 接收子单元 63, 用于接收所述计算单元的资源请求, 所述资源请求包括资 源申请请求、 资源释放请求或资源锁定请求中的至少一种;
确定子单元 64, 用于根据所述资源请求和所述资源状态表, 对传输通路进 行拓朴计算, 确定所述计算单元中的内核和所述存储单元中的存储子单元之间 的传输通路及天线的传输参数;
调整子单元 65, 用于根据所述确定子单元确定的传输通路和天线的传输参 数, 调整天线的传输参数以确保所述计算单元中的内核与所述存储单元中的存 储子单元完成数据传输。
所述调节单元还包括:
内建自测子单元 66, 用于测试传输通路的传输状况并上报至所述确定子单 元 64以便所述确定子单元 65为所述计算单元中的内核和所述存储单元中的存 储子单元分配传输通路;
数据搬移子单元 67, 用于根据所述计算单元的数据搬移指令, 搬移所述存 储单元中的数据; 数据时緩存数据, 或者若所述存储单元包括至少两个存储子单元, 则在所述计 本发明实施例还包括一种芯片, 所述芯片包括如本发明上述任意实施例所 述的 3D堆叠器件。
请参照图 7, 为本发明实施例提供的应用于 3D堆叠器件的通信方法的流程 示意图, 在本实施例中, 所述方法包括:
5701 , 获取所述存储单元中存储子单元的资源占用信息。
5702, 接收所述计算单元的资源请求, 所述资源请求包括资源申请请求、 资源释放请求或资源锁定请求中的至少一种。
5703 ,根据所述资源请求和所述资源占用信息,对传输通路进行拓朴计算, 天线的传输参数, 所述天线的传输参数包括相位参数。
5704, 根据确定的传输通路和天线的传输参数, 调整天线的传输参数以确 可选地, 在执行步骤 S703时, 还可以根据获取的资源占用信息, 生成资源 状态表, 所述资源状态表包括资源使用状态信息、 资源使用率信息或所述存储 子单元的温度信息中的至少一种。 生成资源列表后更加便于调用单元进行资源 管理, 同时还可以通过配置显示器显示给用户查看。
所述天线的传输参数还可以包括但不限于:
编号参数和发射功率参数。
需要说明的是, 本说明书中的各个实施例均釆用递进的方式描述, 每个实 施例重点说明的都是与其它实施例的不同之处, 各个实施例之间相同相似的部 分互相参见即可。 对于装置实施例而言, 由于其与方法实施例基本相似, 所以 描述的比较简单, 相关之处参见方法实施例的部分说明即可。
通过上述实施例的描述, 本发明具有以下优点:
通过在计算单元和存储单元之间配置天线阵列, 并通过调节单元调节天线
接, 可根据应用需求进行动态资源分配, 单端口替换多端口的机械切换方式, 无需开关切换, 传输速率快, 且可以实现多输入多输出的效果, 支持频分复用 和时分复用。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程, 是可以通过计算机程序来指令相关的硬件来完成, 所述的程序可存储于一计算 机可读取存储介质中,该程序在执行时,可包括如上述各方法的实施例的流程。 其中, 所述的存储介质可为磁碟、 光盘、 只读存储记忆体(Read-Only Memory, 简称 ROM )或随机存取存储器( Random Access Memory, 简称 RAM )等。
以上所揭露的仅为本发明较佳实施例而已, 当然不能以此来限定本发明之 权利范围, 因此依本发明权利要求所作的等同变化,仍属本发明所涵盖的范围。

Claims

权 利 要 求
1、 一种 3D堆叠器件, 其特征在于, 包括:
第一基板, 用于承载计算单元, 所述计算单元包括一个内核;
第二基板, 用于承载存储单元, 所述存储单元包括一个存储子单元; 第一天线阵列, 位于所述第一基板上, 与所述计算单元电连接, 且指向所 述第二基板, 用于发射所述计算单元输出的数据和 /或接收第二天线阵列发射 的数据;
所述第二天线阵列, 位于所述第二基板上, 与所述存储单元电连接, 且指 向所述第一基板, 用于接收所述第一天线阵列发射的数据和 /或接收所述存储 单元输出的数据;
调节单元, 用于调节所述第一天线阵列和 /或第二天线阵列中的天线的传 元之间完成数据传输, 其中, 所述天线的传输参数包括相位参数。
2、 如权利要求 1所述的 3D堆叠器件, 其特征在于, 所述第一天线阵列 中的天线为所述第一基板上指向所述第二基板且不与所述第二基板接触的硅 通孔,所述第二天线阵列中的天线为所述第二基板上指向所述第一基板且不与 所述第一基板接触的硅通孔。
3、 如权利要求 1或 2所述的 3D堆叠器件, 其特征在于, 所述调节单元 还用于:
获取所述存储单元中的所述存储子单元的资源占用信息,接收所述计算单 元的资源请求,根据所述资源请求和所述资源占用信息调整所述计算单元中的 所述内核和所 i# A储 元 Φ ό^ι所 i# 储子 元之间
4、如权利要求 3所述的 3D堆叠器件,其特征在于, 所述调节单元包括: 监控子单元, 用于获取所述存储单元中所述存储子单元的资源占用信息; 记录子单元, 用于根据所述监控子单元获取的资源占用信息, 生成资源状 态表, 所述资源状态表包括资源使用状态信息、 资源使用率信息或所述存储子 单元的温度信息中的至少一种;
接收子单元, 用于接收所述计算单元的资源请求, 所述资源请求包括资源 申请请求、 资源释放请求或资源锁定请求中的至少一种;
确定子单元, 用于根据所述资源请求和所述资源状态表,对传输通路进行 拓朴计算,确定所述计算单元中的所述内核和所述存储单元中的所述存储子单 元之间的传输通路及天线的传输参数;
调整子单元,用于根据所述确定子单元确定的传输通路和天线的传输参数, 调整天线的传输参数以确保所述计算单元中的所述内核与所述存储单元中的 所述存储子单元完成数据传输。
5、 如权利要求 4所述的 3D堆叠器件, 其特征在于, 所述天线的传输参 数还包括:
所述确定子单元确定的传输通路中需要对接的天线的编号参数和所述需 要对接的天线的发射功率参数。
6、 如权利要求 4或 5所述的 3D堆叠器件, 其特征在于, 所述调节单元 还包括:
内建自测子单元,用于测试传输通路的传输状况并上报至所述确定子单元 以便所述确定子单元为所述计算单元中的所述内核和所述存储单元中的所述 存储子单元分配传输通路。
7、 如权利要求 6所述的 3D堆叠器件, 其特征在于, 所述调节单元还包 括:
数据搬移子单元, 用于根据所述计算单元的数据搬移指令,搬移所述存储 单元中的数据;
緩冲子单元,用于在所述数据搬移子单元搬移所述存储单元中的数据时緩 存数据, 或者若所述存储单元包括至少两个存储子单元, 则在所述计算单元调
8、 如权利要求 1-7任一项所述的 3D堆叠器件, 其特征在于, 所述第一天 线阵列和所述第二天线阵列处于阻抗值达到预设阔值的介质中。
9、 如权利要求 1-8任一项所述的 3D堆叠器件, 其特征在于, 所述第一天 线阵列和所述第二天线阵列以近场通信的方式进行数据传输。
10、 一种芯片, 其特征在于, 包括:
如权利要求 1-9任一项所述的 3D堆叠器件。
11、一种通信方法,应用于如权利要求 1-9任一项所述的 3D堆叠器件上, 其特征在于, 包括:
获取所述存储单元中存储子单元的资源占用信息;
接收所述计算单元的资源请求, 所述资源请求包括资源申请请求、 资源释 放请求或资源锁定请求中的至少一种;
根据所述资源请求和所述资源占用信息,对传输通路进行拓朴计算,确定 线的传输参数, 所述天线的传输参数包括相位参数;
根据确定的传输通路和天线的传输参数,调整天线的传输参数以确保所述
12、 如权利要求 11所述的通信方法, 其特征在于, 在所述根据所述资源 请求和所述资源占用信息,对传输通路进行拓朴计算,确定所述计算单元中的 包括:
根据获取的资源占用信息, 生成资源状态表, 所述资源状态表包括资源使 用状态信息、 资源使用率信息或所述存储子单元的温度信息中的至少一种。
13、 如权利要求 11或 12所述的通信方法, 其特征在于, 所述天线的传输 参数还包括:
编号参数和发射功率参数。
PCT/CN2014/081286 2014-06-30 2014-06-30 一种3d堆叠器件、芯片及通信方法 WO2016000181A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2014/081286 WO2016000181A1 (zh) 2014-06-30 2014-06-30 一种3d堆叠器件、芯片及通信方法
CN201480038640.XA CN105393353B (zh) 2014-06-30 2014-06-30 一种3d堆叠器件、芯片及通信方法

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2014/081286 WO2016000181A1 (zh) 2014-06-30 2014-06-30 一种3d堆叠器件、芯片及通信方法

Publications (1)

Publication Number Publication Date
WO2016000181A1 true WO2016000181A1 (zh) 2016-01-07

Family

ID=55018266

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2014/081286 WO2016000181A1 (zh) 2014-06-30 2014-06-30 一种3d堆叠器件、芯片及通信方法

Country Status (2)

Country Link
CN (1) CN105393353B (zh)
WO (1) WO2016000181A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106303657A (zh) * 2016-08-18 2017-01-04 北京奇虎科技有限公司 一种连麦直播的方法及主播端设备
CN112994768A (zh) * 2021-02-05 2021-06-18 北京航空航天大学 一种基于矩阵求逆的短距离并行无线传输系统及方法

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109558370A (zh) * 2017-09-23 2019-04-02 成都海存艾匹科技有限公司 三维计算封装

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1839474A (zh) * 2004-01-28 2006-09-27 松下电器产业株式会社 模块及使用它的安装构造体
WO2009114965A1 (en) * 2008-03-19 2009-09-24 Acm Research (Shanghai) Inc. Electrochemical deposition system
CN103107166A (zh) * 2013-01-23 2013-05-15 华中科技大学 一种三维堆叠封装芯片中的电感及无线耦合通信系统

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1839474A (zh) * 2004-01-28 2006-09-27 松下电器产业株式会社 模块及使用它的安装构造体
WO2009114965A1 (en) * 2008-03-19 2009-09-24 Acm Research (Shanghai) Inc. Electrochemical deposition system
CN103107166A (zh) * 2013-01-23 2013-05-15 华中科技大学 一种三维堆叠封装芯片中的电感及无线耦合通信系统

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106303657A (zh) * 2016-08-18 2017-01-04 北京奇虎科技有限公司 一种连麦直播的方法及主播端设备
CN112994768A (zh) * 2021-02-05 2021-06-18 北京航空航天大学 一种基于矩阵求逆的短距离并行无线传输系统及方法
CN112994768B (zh) * 2021-02-05 2022-08-30 北京航空航天大学 一种基于矩阵求逆的短距离并行无线传输系统及方法

Also Published As

Publication number Publication date
CN105393353A (zh) 2016-03-09
CN105393353B (zh) 2018-06-15

Similar Documents

Publication Publication Date Title
US10110499B2 (en) QoS in a system with end-to-end flow control and QoS aware buffer allocation
US9886409B2 (en) System and method for configuring a channel
CN110691419B (zh) 平衡用于双连接的上行链路传输
US9606916B2 (en) Semiconductor devices including application processor connected to high-bandwidth memory and low-bandwidth memory, and channel interleaving method thereof
CN109643704A (zh) 用于管理多芯片封装上的专用功率门控的方法和设备
KR101736593B1 (ko) 온-다이 상호접속부의 아키텍처
CN107111573B (zh) 三维3d集成电路及其中使用的方法
JP2013524519A (ja) インターフェイスダイと複数のダイスタックとの間の同時通信を可能にする装置、スタック型デバイス内のインターリーブされた導電経路、ならびにその形成方法および動作方法
TW201333710A (zh) 利用封裝體上的輸入/輸出介面之於封裝體中的封裝晶片至晶粒的互連技術
TW201515176A (zh) 具有一控制器及一記憶體堆疊之彈性記憶體系統
US11295828B2 (en) Multi-chip programming for phased array
CN103988442B (zh) 用于互连封装内芯片的设备及系统
US8713248B2 (en) Memory device and method for dynamic random access memory having serial interface and integral instruction buffer
WO2016000181A1 (zh) 一种3d堆叠器件、芯片及通信方法
US20170141025A1 (en) Semiconductor device and semiconductor package
US10762012B2 (en) Memory system for sharing a plurality of memories through a shared channel
US20230403608A1 (en) Data router connectivity to wireless communication slices
KR102036693B1 (ko) 반도체 메모리 시스템 및 그의 동작 방법
CN112805820A (zh) 芯片制造方法及芯片结构
KR20230015068A (ko) 무선 통신 시스템에서 측정 보고(measurement report, MR)를 수행하기 위한 장치 및 방법
CN115985357A (zh) 三维堆叠式射频接口存储器和射频数据处理装置
KR20130072853A (ko) 패키지 및 메모리 패키지

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 201480038640.X

Country of ref document: CN

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

Ref document number: 14896404

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 14896404

Country of ref document: EP

Kind code of ref document: A1