Jump to content

Project Denver

From Wikipedia, the free encyclopedia
Nvidia Denver 1/2
General information
Launched2014 (Denver)
2016 (Denver 2)
Designed byNvidia
Cache
L1cache192KiBper core
(128 KiB I-cache with parity, 64 KiB D-cache with ECC)
L2 cache2MiB@ 2 cores
Architecture and classification
Technology node28 nm (Denver 1) to 16 nm (Denver 2)
Instruction setARMv8-A
Physical specifications
Cores
  • 2
Nvidia Carmel
General information
Launched2018
Designed byNvidia
Max.CPUclock rateto 2.3 GHz
Cache
L1cache192KiBper core
(128 KiB I-cache with parity, 64 KiB D-cache with ECC)
L2 cache2MiB@ 2 cores
L3 cache(4MiB@ 8 cores, T194[1])
Architecture and classification
Technology node12 nm
Instruction setARMv8.2-A
Physical specifications
Cores
  • 2

Project Denveris the codename of acentral processing unitdesigned byNvidiathat implements theARMv8-A64/32-bitinstruction setsusing a combination of simple hardware decoder and software-basedbinary translation(dynamic recompilation) where "Denver's binary translation layer runs in software, at a lower level than the operating system, and stores commonly accessed, already optimized code sequences in a 128 MB cache stored in main memory".[2]Denver is a very wide in-ordersuperscalarpipeline. Its design makes it suitable for integration with otherSIPscores (e.g.GPU,display controller,DSP,image processor,etc.) into onedieconstituting asystem on a chip(SoC).

Project Denver is targeted at mobile computers,personal computers,servers,as well assupercomputers.[3]Respective cores have found integration in theTegraSoC series from Nvidia. Initially Denver cores was designed for the28 nmprocess node (Tegra model T132 aka "Tegra K1").Denver 2was an improved design that built for the smaller, more efficient16 nmnode. (Tegra model T186 aka "Tegra X2").

In 2018, Nvidia released an improved design (codename: "Carmel",based on ARMv8 (64-bit; variant: ARM-v8.2[4]with 10-way superscalar, functional safety, dual execution, parity & ECC) got integrated into the TegraXavierSoC offering a total of 8 cores (or 4 dual-core pairs).[5][failed verification]The Carmel CPU core supports full Advanced SIMD (ARM NEON), VFP (Vector Floating Point), and ARMv8.2-FP16.[6]First published testings of Carmel cores integrated in the Jetson AGX development kit by third party experts took place in September 2018 and indicated a noticeably increased performance as should expected for this real world physical manifestation compared to predecessors systems, despite all doubts the used quickness of such a test setup in general an in particular implies.[7]The Carmel design can be found in the Tegra model T194 ( "Tegra Xavier") that is designed with a 12 nm structure size.

Overview[edit]

  • Pipelined processor with 7-waysuperscalarexecution pipeline
  • 128 KiB instruction + 64 KiB data L1 cache per core (both 4-way), 2MiBL2 cache (16-way shared)[8]
  • Denver also sets aside 128 MiB of main memory as an interpretation cache, which is inaccessible to the main operating system.
  • Running at up to 2.5 GHz[9]
  • ARM code is translated either by a hardware translator or through software emulation to an instruction set that is internal to Project Denver. ARM instructions can be reordered, removed if they do not contribute to the end result, or otherwise optimized if software emulation is used.[2]

Chips[edit]

Adual-coreDenver CPU was paired with aKepler-based GPU solution to form theTegra K1;the dual-core 2.3 GHz Denver-based K1 was first used in the HTCNexus 9tablet, released November 3, 2014.[10][11]Note, however, that the quad-core Tegra K1, while using the same name, isn't based on Denver.

TheNvidia Tegra X2has two Denver2 (ARMv8 64bit) cores inside and another fourA57(ARMv8 64bit) cores using a coherent HMP (Heterogeneous Multi-Processor Architecture) approach.[12]This pairs the units with a Parker-GPU.

TheTegra Xavieris pairing an Nvidia Volta-GPU and several special purpose accelerators with 8 CPU cores with the Carmel design. In this design 4 CarmelASICmacro blocks (with each having 2 cores) are matched to each other with one more crossbar and 4 MiB of L3 memory.

History[edit]

The existence of Project Denver was revealed at the 2011Consumer Electronics Show.[13]In a March 4, 2011 Q&A article CEOJen-Hsun Huangrevealed that Project Denver is a five-year64-bitARMv8-A architectureCPUdevelopment on which hundreds of engineers had already worked for three and half years and which also has32-bitARM instruction set (ARMv7)backward compatibility.[14]Project Denver was started in Stexar Company (Colorado) as an x86-compatible processor using binary translation, similar to projects byTransmeta.Stexar was acquired by Nvidia in 2006.[15][16][17]

According to Tom's Hardware, there are engineers fromIntel,AMD,HP,SunandTransmetaon the Denver team, and they have extensive experience designingsuperscalarCPUs without-of-order execution,very long instruction words(VLIW) andsimultaneous multithreading(SMT).[18]

According to Charlie Demer gian, the Project Denver CPU may internally translate the ARM instructions to an internal instruction set, using firmware in the CPU.[19]Also according to Demer gian, Project Denver was originally intended to support bothARMandx86code usingcode morphingtechnology from Transmeta, but was changed to the ARMv8-A 64-bit instruction set because Nvidia could not obtain a license to Intel'spatents.[19]

The first consumer device shipping with Denver CPU cores, Google'sNexus 9,was announced on October 15, 2014. The tablet was manufactured by HTC and features the dual-core Tegra K1 SoC. The Nexus 9 was the first 64-bit Android device available to consumers.[20]

See also[edit]

References[edit]

  1. ^NVIDIA Jetson AGX Xavier Delivers 32 TeraOps for New Era of AI in Roboticsby Dustin Franklin (Nvidia development team for Jetson), December 12, 2018
  2. ^abWasson, Scott (August 11, 2014)."Nvidia claims Haswell-class performance for Denver CPU core".The Tech Report.RetrievedAugust 14,2014.
  3. ^Dally, Bill (January 5, 2011).""PROJECT DENVER" PROCESSOR TO USHER IN NEW ERA OF COMPUTING ".Official Nvidia blog.
  4. ^NVIDIA Jetson AGX Xavier Delivers 32 TeraOps for New Era of AI in Roboticsby Dustin Franklin (Nvidia development team for Jetson), December 12, 2018
  5. ^NVIDIA Drive Xavier SOC Detailedby Hassan Mujtaba on Jan 8, 2018 via WccfTech
  6. ^NVIDIA Jetson AGX Xavier Delivers 32 TeraOps for New Era of AI in Roboticsby Dustin Franklin (Nvidia development team for Jetson), December 12, 2018
  7. ^"A Quick Test of NVIDIA's" Carmel "CPU Performance".
  8. ^Hachman, Mark (August 11, 2014)."Nvidia reveals PC-like performance for 'Denver' Tegra K1".PC World.RetrievedSeptember 19,2014.
  9. ^Anthony, Sebastian (January 6, 2014)."Tegra K1 64-bit Denver core analysis: Are Nvidia's x86 efforts hidden within?".ExtremeTech.RetrievedJanuary 7,2014.
  10. ^"Nexus 9 storms through Geekbench, Tegra K1 outperforms Apple iPhone 6's A8".16 October 2014.
  11. ^Shimpi, Anand (January 5, 2014)."NVIDIA Announces Tegra K1 SoC with Optional Denver CPU Cores".Anandtech.RetrievedJanuary 6,2014.
  12. ^NVIDIA Unveils Tegra Parker SOC at Hot Chips – Built on 16nm TSMC Process, Features Pascal and Denver 2 Duo Architecture,August 22, 2016
  13. ^http:// nvidia /object/ces2011.htmlNvidia's press conference webcast
  14. ^Takahashi, Dean (March 4, 2011)."Q&A: Nvidia chief explains his strategy for winning in mobile computing".
  15. ^Valich, Theo (December 12, 2011)."NVIDIA Project Denver" Lost in Rockies ", to Debut in 2014-15".
  16. ^Miller, Paul (October 19, 2006)."NVIDIA has x86 CPU in the works?".Engadget.RetrievedOctober 19,2013.
  17. ^Valich, Theo (March 20, 2013)."New Tegra Roadmap Reveals Logan, Parker and Kayla CUDA Strategy".
  18. ^Parrish, Kevin (October 14, 2013)."64-bit Nvidia Tegra 6" Parker "Chip May Arrive in 2014. Devices with a 64-bit Tegra 6 could launch before the end of 2014".Tom's Hardware & ExtremeTech.RetrievedOctober 19,2013.
  19. ^abDemer gian, Charlie (August 5, 2011)."What is Project Denver based on?".Semiaccurate.
  20. ^Amadeo, Ron (October 15, 2014)."Google announces Nexus 6, Nexus 9, Nexus Player, and Android 5.0 Lollipop".

External links[edit]