WO2005064796A1 - Load-aware circuit arrangement - Google Patents
Load-aware circuit arrangement Download PDFInfo
- Publication number
- WO2005064796A1 WO2005064796A1 PCT/IB2004/052710 IB2004052710W WO2005064796A1 WO 2005064796 A1 WO2005064796 A1 WO 2005064796A1 IB 2004052710 W IB2004052710 W IB 2004052710W WO 2005064796 A1 WO2005064796 A1 WO 2005064796A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- circuit arrangement
- load
- circuit
- buffer
- arrangement according
- Prior art date
Links
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03K—PULSE TECHNIQUE
- H03K19/00—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits
- H03K19/0175—Coupling arrangements; Interface arrangements
- H03K19/0185—Coupling arrangements; Interface arrangements using field effect transistors only
- H03K19/018585—Coupling arrangements; Interface arrangements using field effect transistors only programmable
Definitions
- the present invention relates a circuit arrangement comprising at least one circuit component at which a load is applied that can vary during operation of the circuit arrangement. Furthermore, the present invention relates to a method of controlling power consumption of such a circuit arrangement, such as for example a field programmable gate array (FPGA).
- FPGA field programmable gate array
- ASICs application specific integrated circuits
- FPGAs can perform different functions depending on a configuration bit stream which is loaded.
- the circuit components inside the FPGA like buffers, logic gates, connection boxes, switch boxes etc., have different input load (fan-in) and output load (fan-out) depending on the configuration which is determined by the configuration bit stream loaded into the FPGA.
- Conventional methods in FPGA circuit design have always designed the components for the worst-case load.
- this method still suffers from the overhead of large capacitances associated with over- designed components designed to drive the worst-case load. It is therefore an object of the present invention to provide a circuit arrangement and method of controlling power consumption by means of which over-design of components can be at least reduced. This object is achieved by a circuit arrangement as claimed in claim 1 and by a method as claimed in claim 1 1. Accordingly, the problem of over-design is solved by tailoring the components to have just sufficient drive capacity depending on the potential load, which is determined by examining the actual load applied at the at least one circuit component. Thereby, component design can be adapted for lowest power-delay-product in different load situations ranging from very low to worst-case loading.
- the determination means may be configured to determine the load based on a configuration information loaded to the circuit arrangement.
- this configuration information may be stored in a configuration memory.
- the configuration information may comprise a configuration bit stream defining at least one of an input load and an output load of the at least one component.
- a configuration information as used for example in FPGAs or other configurable circuit arrangements can be used to adjust the drive capacity of the individual components to thereby optimize the power consumption by tailoring the components so as to provide sufficient drive capacity for the selected configuration.
- the adjusting means may be configured to vary a buffer size or a buffer number of the at least one component.
- control signal may be generated by the adjusting means for switching on or off the buffers or buffer sections.
- the control signal may be derived from a most significant bit signal of a selection signal derived from the determination means.
- selection signals supplied from the configuration memory e.g. of an FPGA can be directly used to switch track buffers into stand-by mode. This leads to a considerable reduction in the active energy consumption. This reduction is obtained at a small area overhead for the buffer.
- the adjusting means may be configured to vary a threshold voltage of circuit elements of the circuit arrangement. This may be achieved by changing at least one bias voltage responsive to the determination means.
- buffers By applying the bias voltage, buffers can be kept smaller in size and can thus have lower power-delay-product and faster speed. Hence, based on the actual configuration, buffers can be optimized for lowest power-delay-product at the same or higher speed. Further advantageous developments are defined in the dependent claims.
- FIG. 2B shows a buffer driving fan-out path as used in FPGAs
- Fig. 3 shows a configuration aware connection box according to a first preferred embodiment
- Fig. 4 shows a configuration aware buffer circuit according to a second preferred embodiment
- Fig. 5 shows a more detailed view of a programmable buffer section as used in the second preferred embodiment
- Figs. 6 and 7 show diagrams of delay vs capacitive load for a conventional and a programmable buffer according to the second preferred embodiment for different load ranges
- Figs. 8 and 9 show diagrams of power-delay-product vs capacitive load for a conventional and a programmable buffer according to the second preferred embodiment for different load ranges
- FIG. 10 shows a buffer circuit with varying threshold voltage according to a third preferred embodiment
- Figs. 11 and 12 show diagrams of normalized delay for different bias voltages at different capacitive loads
- Figs. 13 and 14 show diagrams of normalized power-delay-product for different bias voltages at different capacitive loads.
- the FPGA circuit arrangement comprises logic blocks 20, input/output blocks (not shown) and programmable routing.
- a so-called island-style FPGA is shown, where the logic blocks 20 are surrounded by pre-fabricated wiring segments 10 on all four sides.
- Input or output terminals of the logic blocks 20 can be connected to wiring segments 10 comprising a plurality of routing wires in the channel adjacent to the logic blocks 20 via a connection block of programmable switches.
- a switch box 30 is provided at every intersection of a horizontal and a vertical channel.
- the FPGA interconnect can be configured by programming the switch boxes 30 to achieve a predetermined circuit configuration.
- FIG. 2A shows a connection box used to connect the logic block 20 to the wiring segments 10 of Fig. 1.
- routing wires 301 of a wiring segment 10 are connected via track buffers 304 and a multiplexing circuit 60 controlled by selection signals SO, SI. which are derived from a configuration information loaded to the FPGA and which may be stored in respective memory cells, e.g. Static Random Access Memory (SRAM) cells 302, to an input port of the logical block 20.
- SRAM Static Random Access Memory
- Fig. 2B shows a schematic diagram of an internal portion of one of the switch boxes 30 of Fig.
- a buffer 304 is used to drive programmable switches SI to S4 which are controlled by respective selection signals CM1 to CM4 which are derived from the configuration information loaded to the FPGA.
- Such buffers 304 of connection boxes as shown in Fig. 1 and fan-out paths and/or switch boxes 30 as shown in Fig. 2 are provided on FPGAs in large numbers. It is therefore desirable to reduce the amount of energy consumed in these components to achieve a reduction in the overall energy consumed by the FPGA. Reducing the amount of energy is especially critical in FPGAs, since a three order of magnitude difference exists between the energy consumption of FPGAs and ASICs.
- tailoring for sufficient drive capacity can be achieved by varying the size and/or number of the buffers 304.
- the drive capacity or drive strength is varied based on the potential load which is applied to a component or which a component has to drive.
- Fig. 3 shows a proposed modification of the connection box 30 of Fig. 2A according to the first preferred embodiment, wherein the selection signals SO and SI which are supplied from a configuration memory are directly used for controlling the track buffers 304, e.g. for setting them into a stand-by mode. This can be achieved by providing controllable switching elements, e.g.
- a reduction in the active energy consumption can be achieved.
- using only the MSB selection signal SI to put track buffers into the stand-by state provides the advantage of less energy consumption at absolutely no area overhead.
- not all non-used track buffers are turned off, but only half of the total number of buffers.
- a dedicated decoding circuit can be provided for decoding the selection signals SO and SI to provide control signals for the switching elements 305 in a manner that only the used track buffer, i.e. the track buffer of the signal line which is switched through the multiplexer, is kept in an active state.
- Fig. 4 shows a programmable structure of the buffers 304 according to the second preferred embodiment.
- the programmable buffer 304 consists of two small inverters 3040 which are always in an active state.
- the other buffer stages or buffer sections 3041 to 3046 are programmable or controllable to be switched on or off.
- the programmable buffer 304 is configured in such a way, that its delay corresponds to the conventional buffers when all its buffer stages 3041 to 3046 are turned on. This configuration is used for worst-case loading.
- the capacitor CL in Fig. 4 represents the capacitive load to be driven by the programmable buffer 304.
- Fig. 5 shows a more detailed view of the buffer stages 3041 to 3046 of Fig.
- a control signal CMN which is used to turn on or off the programmable buffer stages 3041 to 3046 is generated at a decoding or control circuit 50 based on a configuration information supplied from the configuration memory 40 of the FPGA.
- CMOS buffer circuit comprising a series connection of two p-channel transistors MP1 and MP2 and two n-channel transistors MN1 and MN2, wherein the control signal CMN is supplied to one of the transistors and an inverted version of the control signal CMN is supplied to another one of the transistors of opposite channel polarity.
- these two controlled transistors can be switched on or off by the selection signal CMN to respectively activate or deactivate the buffer stage.
- simulations may be performed. Possible results of such simulations are shown in the following Figs. 6 to 9.
- FIGS. 8 and 9 show plots of power-delay-product (which is indicative of energy consumption) vs capacitive loads for the different buffer configurations in a 0.13um CMOS technology.
- the capacitive load CL at the output of the programmable buffer 304 of Fig. 4 has been swept from lOfF to 2pF to mimic the variation of the load from the lowest load to the worst-case load. From Figs. 6 to 9, it can be gathered that the configuration "PRGl 10000" leads to the lowest energy consumption at an acceptable delay for loads in the range of 10 to 40 fF.
- the programmable buffer can be tuned for having an acceptable delay and the least energy consumption.
- Fig. 10 shows a schematic circuit diagram of a multi-stage buffer circuit, wherein n-well and p-well bias voltages VNW and VPW can be controlled to change the threshold voltage of individual transistor elements or other semiconductor elements.
- the control circuit 50 is used in this third embodiment to generate or supply the bias voltages VNW and VPW based on the configuration information supplied from the configuration memory 40.
- Fig. 1 1 to 14 show diagrams indicating delay and PDP, respectively, of the bias- voltage-controlled buffer circuit of Fig.
- the different areas in Fig. 11 indicate averages of normalized delays ranging from 0.7 to 0.8 in the left upper area, from 0.8 to 0.9 in the dark left area, and from 0.9 to 1 in the middle grey area.
- the average of the normalized delay ranges from 0.9 to 0.95 in the small dark area in the upper left portion, from 0.95 to 1 in the small white area in the upper left portion, and from 0.85 to 0.9 in the remaining area.
- Fig. 11 indicate averages of normalized delays ranging from 0.7 to 0.8 in the left upper area, from 0.8 to 0.9 in the dark left area, and from 0.9 to 1 in the middle grey area.
- the average of the normalized delay ranges from 0.9 to 0.95 in the small dark area in the upper left portion, from 0.95 to 1 in the small white area in the upper left portion, and from 0.85 to 0.9 in the remaining area.
- the average of the normalized PDP ranges from 0.94 to 0.98 in the small white areas at the upper left corner and the upper and lower right corners, from 0.9 to 0.94 in the remaining white areas, from 0.86 to 0.9 in the grey area, and from 0.82 to 0.86 in the middle dark area.
- the average normalized PDP ranges from 0.8 to 0.99 in the grey area in the upper left portion, from 1.56 to 1.75 in the dark area, from 1.18 to 1.37 in the white area in the middle portion and from 1.37 to 1.56 in the white area in the lower right corner of the diagram.
- the proposed buffer can be faster than the conventional buffer and can have a smaller PDP.
- the proposed buffer is faster and has a lower power-delay- product (PDP).
- the bias voltages can be generated on-chip by using the threshold drops of the
- the bias voltage not necessarily has to be generated by a reference voltage generator, but could as well be generated by a logic circuit which may be provided for example in the control circuit 50 of Fig. 10. Then, the logic circuit responds to a changing load of the buffer, which can be determined by observing the configuration memory 40 of the FPGA which controls the switches that the buffer drives, by changing the bias voltages VNW applied to the n-well and VPW applied to the p-well of the buffer circuit of Fig. 10.
- the proposed tailoring of the circuit components for sufficient drive can be achieved either by varying the size of the buffers as proposed in the first and second embodiments or by adjusting the threshold voltage as proposed in the third embodiment or even by doing both in combination.
- energy efficiency can be achieved by varying the drive strength based on the potential load that a component has to drive or which is supplied to a component.
- the proposed scheme not only reduces the energy consumption of FPGAs but also reduces off-state leakage and noise generation due to the lower time derivative (dl/dt) of the current. This lower time derivative means that the buffer can drain less current from the power supply per unit of time which results in a lower supply bounds and electromagnetic interference (EMI).
- EMI electromagnetic interference
- the present invention is not restricted to the above embodiments but can applied for design of any circuit component where potential load at run-time can be determined.
- the proposed scheme can be applied in eFPGA circuits which are part of ASICs.
- the NMOS and PMOS transistors not necessarily need to be placed between another transistor and ground and another transistor and power supply, but can also be placed between the output node of a buffer or buffer stage and the bottom transistor, or between the output node and the top transistor.
- the proposed scheme can be applied to the design of any load-sensitive bit configuration aware components for low energy circuit arrangements.
- circuit components such as buffers, logic gates, connection boxes, switch boxes etc., which have different fan-in and fan-out load depending on the configuration, can be controlled by determining the expected load of the component and/or by dynamically sizing the drive power of the component that is sufficient to handle the load with acceptable delay.
- the embodiments may thus vary within the scope of the attached claims.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Logic Circuits (AREA)
- Semiconductor Integrated Circuits (AREA)
- Design And Manufacture Of Integrated Circuits (AREA)
Abstract
Description
Claims
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2006546423A JP2007517450A (en) | 2003-12-23 | 2004-12-08 | Load recognition circuit device |
EP04801500A EP1700377A1 (en) | 2003-12-23 | 2004-12-08 | Load-aware circuit arrangement |
US10/583,808 US7741866B2 (en) | 2003-12-23 | 2004-12-08 | Load-aware circuit arrangement |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP03104934 | 2003-12-23 | ||
EP03104934.9 | 2003-12-23 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2005064796A1 true WO2005064796A1 (en) | 2005-07-14 |
Family
ID=34717244
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/IB2004/052710 WO2005064796A1 (en) | 2003-12-23 | 2004-12-08 | Load-aware circuit arrangement |
Country Status (5)
Country | Link |
---|---|
US (1) | US7741866B2 (en) |
EP (1) | EP1700377A1 (en) |
JP (1) | JP2007517450A (en) |
CN (1) | CN1898869A (en) |
WO (1) | WO2005064796A1 (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8030968B1 (en) * | 2010-04-07 | 2011-10-04 | Intel Corporation | Staged predriver for high speed differential transmitter |
US8643418B2 (en) * | 2011-06-02 | 2014-02-04 | Micron Technology, Inc. | Apparatus and methods for altering the timing of a clock signal |
JP6056632B2 (en) * | 2013-04-22 | 2017-01-11 | 富士通株式会社 | Data holding circuit and semiconductor integrated circuit device |
US9490805B2 (en) * | 2014-09-02 | 2016-11-08 | Integrated Device Technology, Inc. | Low power driver with programmable output impedance |
US9419588B1 (en) | 2015-02-21 | 2016-08-16 | Integrated Device Technology, Inc. | Output driver having output impedance adaptable to supply voltage and method of use |
US9407268B1 (en) | 2015-04-29 | 2016-08-02 | Integrated Device Technology, Inc. | Low voltage differential signaling (LVDS) driver with differential output signal amplitude regulation |
JP6924621B2 (en) * | 2017-06-12 | 2021-08-25 | 日立Astemo株式会社 | Electronic controls, in-vehicle systems, and power supplies |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5666078A (en) * | 1996-02-07 | 1997-09-09 | International Business Machines Corporation | Programmable impedance output driver |
US5994922A (en) * | 1996-02-08 | 1999-11-30 | Kabushiki Kaisha Toshiba | Output buffer, semiconductor integrated circuit having output buffer and driving ability adjusting method for output buffer |
US6075379A (en) * | 1998-01-22 | 2000-06-13 | Intel Corporation | Slew rate control circuit |
US6087847A (en) * | 1997-07-29 | 2000-07-11 | Intel Corporation | Impedance control circuit |
US6331785B1 (en) * | 2000-01-26 | 2001-12-18 | Cirrus Logic, Inc. | Polling to determine optimal impedance |
Family Cites Families (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5220516A (en) * | 1989-02-21 | 1993-06-15 | International Business Machines Corp. | Asynchronous staging of objects between computer systems in cooperative processing systems |
US5134311A (en) * | 1990-06-07 | 1992-07-28 | International Business Machines Corporation | Self-adjusting impedance matching driver |
US5220216A (en) * | 1992-01-02 | 1993-06-15 | Woo Ann K | Programmable driving power of a CMOS gate |
US5801548A (en) * | 1996-04-11 | 1998-09-01 | Xilinx Inc | Configurable performance-optimized programmable logic device |
KR100356576B1 (en) * | 2000-09-15 | 2002-10-18 | 삼성전자 주식회사 | programmable data output circuit with programmable on chip termination operation and method therefore |
JP3670563B2 (en) * | 2000-09-18 | 2005-07-13 | 株式会社東芝 | Semiconductor device |
US6445245B1 (en) * | 2000-10-06 | 2002-09-03 | Xilinx, Inc. | Digitally controlled impedance for I/O of an integrated circuit device |
DE60239447D1 (en) * | 2001-01-09 | 2011-04-28 | Broadcom Corp | SUBMICRON INPUT / OUTPUT CIRCUIT WITH HIGH INPUT VOLTAGE COMPATIBILITY |
US6545522B2 (en) * | 2001-05-17 | 2003-04-08 | Intel Corporation | Apparatus and method to provide a single reference component for multiple circuit compensation using digital impedance code shifting |
DE10139126A1 (en) * | 2001-08-09 | 2003-02-20 | Ciba Sc Pfersee Gmbh | A four-step method for preparation of compositions containing polysiloxanes and fluoropolymers useful for treatment of fiber materials, e.g. flat textile articles with superior in oil repelling action |
KR100495660B1 (en) * | 2002-07-05 | 2005-06-16 | 삼성전자주식회사 | Semiconductor integrated circuit having on-die termination circuit |
KR100505645B1 (en) * | 2002-10-17 | 2005-08-03 | 삼성전자주식회사 | Output driver capable of controlling slew rate of output signal according to operating frequency information or CAS latency information |
-
2004
- 2004-12-08 US US10/583,808 patent/US7741866B2/en active Active
- 2004-12-08 JP JP2006546423A patent/JP2007517450A/en not_active Withdrawn
- 2004-12-08 EP EP04801500A patent/EP1700377A1/en not_active Withdrawn
- 2004-12-08 WO PCT/IB2004/052710 patent/WO2005064796A1/en not_active Application Discontinuation
- 2004-12-08 CN CNA2004800384951A patent/CN1898869A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5666078A (en) * | 1996-02-07 | 1997-09-09 | International Business Machines Corporation | Programmable impedance output driver |
US5994922A (en) * | 1996-02-08 | 1999-11-30 | Kabushiki Kaisha Toshiba | Output buffer, semiconductor integrated circuit having output buffer and driving ability adjusting method for output buffer |
US6087847A (en) * | 1997-07-29 | 2000-07-11 | Intel Corporation | Impedance control circuit |
US6075379A (en) * | 1998-01-22 | 2000-06-13 | Intel Corporation | Slew rate control circuit |
US6331785B1 (en) * | 2000-01-26 | 2001-12-18 | Cirrus Logic, Inc. | Polling to determine optimal impedance |
Also Published As
Publication number | Publication date |
---|---|
EP1700377A1 (en) | 2006-09-13 |
US7741866B2 (en) | 2010-06-22 |
US20070115026A1 (en) | 2007-05-24 |
CN1898869A (en) | 2007-01-17 |
JP2007517450A (en) | 2007-06-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
KR100888806B1 (en) | Semiconductor integrated circuit device | |
KR100625153B1 (en) | Semiconductor integrated circuit | |
KR100562636B1 (en) | Power up circuit in semiconductor device | |
KR0153603B1 (en) | Power-up reset signal generating circuit of semiconductor apparatus | |
US7262642B2 (en) | Semiconductor integrated circuit comprising first and second transmission systems | |
US5497117A (en) | Input sense circuit having selectable thresholds | |
US7131074B2 (en) | Nested voltage island architecture | |
US7741866B2 (en) | Load-aware circuit arrangement | |
US7656210B2 (en) | Semiconductor integrated circuit | |
JP2006135560A (en) | Level shift circuit and semiconductor integrated circuit device including it | |
US7629812B2 (en) | Switching circuits and methods for programmable logic devices | |
KR100699832B1 (en) | Multi-Threshold complementary metal-oxide semiconductor control circuit | |
KR100226486B1 (en) | Semiconductor circuit for high power voltage | |
US20180109262A1 (en) | Techniques For Handling High Voltage Circuitry In An Integrated Circuit | |
US7589562B2 (en) | I/O cell capable of finely controlling drive strength | |
KR100587087B1 (en) | An internal voltage generator for a semiconductor device | |
US5933050A (en) | Semiconductor circuit | |
JP3055223B2 (en) | Buffer circuit | |
US6433594B1 (en) | Semiconductor integrated circuit and semiconductor integrated circuit system | |
KR100232523B1 (en) | Low power consumption type circuit | |
KR100313497B1 (en) | Power transfer switch and control circuit for internal generation power supply | |
KR100444316B1 (en) | Input buffer with delay reduction part of semiconductor memory device to reduce delay of each inverter node | |
KR100608347B1 (en) | Data output buffer control circuit | |
JPH1141090A (en) | Semiconductor device with signal level converting function | |
KR100282228B1 (en) | Data output buffer |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 200480038495.1 Country of ref document: CN |
|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NA NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): BW GH GM KE LS MW MZ NA SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LT LU MC NL PL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2004801500 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2007115026 Country of ref document: US Ref document number: 2006546423 Country of ref document: JP Ref document number: 10583808 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWW | Wipo information: withdrawn in national office |
Ref document number: DE |
|
WWP | Wipo information: published in national office |
Ref document number: 2004801500 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 10583808 Country of ref document: US |