Difference between revisions of "GPU"

From Chessprogramming wiki
Jump to: navigation, search
(Hardware Model)
(Programming Model)
Line 79: Line 79:
=Programming Model=
=Programming Model=
A [https://en.wikipedia.org/wiki/Parallel_programming_model parallel programming model] for GPGPU can be [https://en.wikipedia.org/wiki/Task_parallelism task-parallel], [https://en.wikipedia.org/wiki/Data_parallelism data-parallel], a mixture of both, or with libraries and offload-directives also [https://en.wikipedia.org/wiki/Implicit_parallelism implicitly-parallel].
A [https://en.wikipedia.org/wiki/Parallel_programming_model parallel programming model] for GPGPU can be [https://en.wikipedia.org/wiki/Data_parallelism data-parallel], [https://en.wikipedia.org/wiki/Task_parallelism task-parallel], a mixture of both, or with libraries and offload-directives also [https://en.wikipedia.org/wiki/Implicit_parallelism implicitly-parallel]. Single GPU threads (resp. work-items in OpenCL) are coupled to a block (resp. work-group in OpenCL), these can be usually synchronized and have access to the same scratch-pad memory, with an architecture limit of how many threads a block can hold.
=Memory Model=
=Memory Model=

Revision as of 09:17, 23 October 2021

Home * Hardware * GPU

GPU (Graphics Processing Unit),
a specialized processor primarily intended to fast image processing. GPUs may have more raw computing power than general purpose CPUs but need a specialized and parallelized way of programming. Leela Chess Zero has proven that a Best-first Monte-Carlo Tree Search (MCTS) with deep learning methodology will work with GPU architectures.


In the 1970s and 1980s RAM was expensive and Home Computers used custom graphics chips to operate directly on registers/memory without a dedicated frame buffer, like TIAin the Atari VCS gaming system, GTIA+ANTIC in the Atari 400/800 series, or Denise+Agnus in the Commodore Amiga series. The 1990s would make 3D graphics and 3D modeling more popular, especially for video games. Cards specifically designed to accelerate 3D math, such as the 3dfx Voodoo2, were used by the video game community to play 3D graphics. Some game engines could use instead the SIMD-capabilities of CPUs such as the Intel MMX instruction set or AMD's 3DNow! for real-time rendering. Sony's 3D capable chip used in the PlayStation (1994) and Nvidia's 2D/3D combi chips like NV1 (1995) coined the term GPU for 3D graphics hardware acceleration. With the advent of the unified shader architecture, like in Nvidia Tesla (2006), ATI/AMD TeraScale (2007) or Intel GMA X3000 (2006), GPGPU frameworks like CUDA and OpenCL emerged and gained in popularity.

GPU in Computer Chess

There are in main three approaches how to use a GPU for Chess:

  • As an accelerator in Lc0: run a neural network for position evaluation on GPU.
  • Offload the search in Zeta: run a parallel game tree search with move generation and position evaluation on GPU.
  • As an hybrid in perft_gpu: expand the game tree to a certain degree on CPU and offload to GPU to compute the sub-tree.

GPU Chess Engines


Early efforts to leverage a GPU for general-purpose computing required reformulating computational problems in terms of graphics primitives via graphics APIs like OpenGL or DirextX, followed by first GPGPU frameworks such as Sh/RapidMind or Brook and finally CUDA and OpenCL.

Khronos OpenCL

OpenCL specified by the Khronos Group is widely adopted across all kind of hardware accelerators from different vendors.


AMD supports language frontends like OpenCL, HIP, C++ AMP and with OpenMP offload directives. It offers with ROCm its own parallel compute platform.


Since macOS 10.14 Mojave a transition from OpenCL to Metal is recommended by Apple.


CUDA is the parallel computing platform by Nvidia. It supports language frontends like C, C++, Fortran, OpenCL and offload directives via OpenACC and OpenMP.


Hardware Model

A common scheme on GPUs is to run multiple threads in SIMT fashion and a multitude of SIMT waves on the same SIMD unit to hide memory latencies. Multiple processing elements (GPU cores) are members of a SIMD unit, multiple SIMD units are coupled to a compute unit, with up to hundreds of compute units present on a discrete GPU. The actual SIMD units may have architecture dependent different numbers of cores (SIMD8, SIMD16, SIMD32), and different computation abilities, floating-point and/or integer with specific bit-width of the FPU/ALU. Scalar units present in the compute unit perform special functions the SIMD units are not capable of and MMACUs (matrix-multiply-accumulate-units) are used to speed up neural networks further.

Programming Model

A parallel programming model for GPGPU can be data-parallel, task-parallel, a mixture of both, or with libraries and offload-directives also implicitly-parallel. Single GPU threads (resp. work-items in OpenCL) are coupled to a block (resp. work-group in OpenCL), these can be usually synchronized and have access to the same scratch-pad memory, with an architecture limit of how many threads a block can hold.

Memory Model

OpenCL offers the following memory model for the programmer:

  • __private - usually registers, accessable only by a single work-item resp. thread.
  • __local - scratch-pad memory shared across work-items of a work-group resp. threads of block.
  • __constant - read-only variable.
  • __global - usually VRAM, accessable by all work-items resp. threads.

Here the data for the Nvidia GeForce GTX 580 (Fermi) as an example: [2]

  • 128 KiB private memory per compute unit
  • 48 KiB (16 KiB) local memory per compute unit (configurable)
  • 64 KiB constant memory
  • 8 KiB constant cache per compute unit
  • 16 KiB (48 KiB) L1 cache per compute unit (configurable)
  • 768 KiB L2 cache
  • 1.5 GiB to 3 GiB global memory

Here the data for the AMD Radeon HD 7970 (GCN) as an example: [3]

  • 256 KiB private memory per compute unit
  • 64 KiB local memory per compute unit
  • 64 KiB constant memory
  • 16 KiB constant cache per four compute units
  • 16 KiB L1 cache per compute unit
  • 768 KiB L2 cache
  • 3 GiB to 6 GiB global memory

Instruction Throughput

GPUs are used in HPC environments because of their good FLOP/Watt ratio. The instruction throughput in general depends on the architecture (like Nvidia's Tesla, Fermi, Kepler, Maxwell or AMD's TeraScale, GCN, RDNA), the brand (like Nvidia GeForce, Quadro, Tesla or AMD Radeon, Radeon Pro, Radeon Instinct) and the specific model.

Integer Instruction Throughput

  • INT32
The 32 bit integer performance can be architecture and operation depended less than 32 bit FLOP or 24 bit integer performance.
  • INT64
In general GPU registers and Vector-ALUs are 32 bit wide and have to emulate 64 bit integer operations.
  • INT8
Some architectures offer higher throughput with lower precision. They quadruple the INT8 or octuple the INT4 throughput.

Floating Point Instruction Throughput

  • FP32
Consumer GPU performance is measured usually in single-precision (32 bit) floating point FMA, fused-multiply-add, throughput.
  • FP64
Consumer GPUs have in general a lower ratio (FP32:FP64) for double-precision (64 bit) floating point operations throughput than server brand GPUs.
  • FP16
Some GPGPU architectures offer half-precision (16 bit) floating point operation throughput with an FP32:FP16 ratio of 1:2.


Nvidia TensorCores

With Nvidia Volta series TensorCores were introduced. They offer FP16xFP16+FP32, matrix-multiplication-accumulate-units, used to accelerate neural networks.[4] Turing's 2nd gen TensorCores add FP16, INT8, INT4 optimized computation.[5] Amperes's 3rd gen adds support for BF16, TF32, FP64 and sparsity acceleration.[6]

AMD Matrix Cores

AMD released 2020 its server-class CDNA architecture with Matrix Cores which support MFMA, matrix-fused-multiply-add, operations on various data types like INT8, FP16, BF16, FP32.

Intel XMX Cores

Intel plans XMX, Xe Matrix eXtensions, for its upcoming Xe discrete GPU series.

Throughput Examples

Nvidia GeForce GTX 580 (Fermi, CC 2.0) - 32 bit integer operations/clock cycle per compute unit [7]

   MAD 16
   MUL 16
   ADD 32
   Bit-shift 16
   Bitwise XOR 32

Max theoretic ADD operation throughput: 32 Ops * 16 CUs * 1544 MHz = 790.528 GigaOps/sec

AMD Radeon HD 7970 (GCN 1.0) - 32 bit integer operations/clock cycle per processing element [8]

   MAD 1/4
   MUL 1/4
   ADD 1
   Bit-shift 1
   Bitwise XOR 1

Max theoretic ADD operation throughput: 1 Op * 2048 PEs * 925 MHz = 1894.4 GigaOps/sec

Host-Device Latencies

One reason GPUs are not used as accelerators for chess engines is the host-device latency, aka. kernel-launch-overhead. Nvidia and AMD have not published official numbers, but in practice there is an measurable latency for null-kernels of 5 microseconds [9] up to 100s of microseconds [10]. One solution to overcome this limitation is to couple tasks to batches to be executed in one run [11].

Deep Learning

GPUs are much more suited than CPUs to implement and train Convolutional Neural Networks (CNN), and were therefore also responsible for the deep learning boom, also affecting game playing programs combining CNN with MCTS, as pioneered by Google DeepMind's AlphaGo and AlphaZero entities in Go, Shogi and Chess using TPUs, and the open source projects Leela Zero headed by Gian-Carlo Pascutto for Go and its Leela Chess Zero adaption.


The market is split into two categories, integrated and discrete GPUs. The first being the most important by quantity, the second by performance. Discrete GPUs are divided as consumer brands for playing 3D games, professional brands for CAD/CGI programs and server brands for big-data and number-crunching workloads. Each brand offering different feature sets in driver, VRAM, or computation abilities.


AMD line of discrete GPUs is branded as Radeon for consumer, Radeon Pro for professional and Radeon Instinct for server.


CDNA architecture in MI100 HPC-GPU with Matrix Cores was unveiled in November, 2020.

Navi 2X RDNA 2.0

RDNA 2.0 cards were unveiled on October 28, 2020.

Navi RDNA 1.0

RDNA 1.0 cards were unveiled on July 7, 2019.

Vega GCN 5th gen

Vega cards were unveiled on August 14, 2017.

Polaris GCN 4th gen

Polaris cards were first released in 2016.



Apple released its M1 SoC (system on a chip) with integrated GPU for desktops and notebooks in 2020.

ARM Mali

The Mali GPU variants can be found on various systems on chips (SoCs) from different vendors. Since Midgard (2012) with unified-shader-model OpenCL support is offered.

Valhall (2019)

Bifrost (2016)

Midgard (2012)


Intel Xe 'Gen12'

Intel Xe line of GPUs (released since 2020) is divided as Xe-LP (low-power), Xe-HPG (high-performance-gaming), Xe-HP (high-performace) and Xe-HPC (high-performance-computing).


Nvidia line of discrete GPUs is branded as GeForce for consumer, Quadro for professional and Tesla for server.

Ampere Architecture

The Ampere microarchitecture was announced on May 14, 2020 [12]. The Nvidia A100 GPU based on the Ampere architecture delivers a generational leap in accelerated computing in conjunction with CUDA 11 [13].

Turing Architecture

Turing cards were first released in 2018. They are the first consumer cores to launch with RTX, for raytracing, features. These are also the first consumer cards to launch with TensorCores used for matrix multiplications to accelerate convolutional neural networks. The Turing GTX line of chips do not offer RTX or TensorCores.

Architectural Whitepaper

Volta Architecture

Volta cards were released in 2017. They were the first cards to launch with TensorCores, supporting matrix multiplications to accelerate convolutional neural networks.

Architecture Whitepaper

Pascal Architecture

Pascal cards were first released in 2016.

Architecture Whitepaper

Maxwell Architecture

Maxwell cards were first released in 2014.

Architecture Whitepaper on archiv.org

PowerVR - Imagination Technologies

Imagination Technologies licenses PowerVR IP to third parties (most notable Apple) used for system on a chip (SoC) designs. Since Series5 SGX OpenCL support via licensees is available.

PowerVR Graphics

See also




2008 ...






2015 ...


Chapter 8 in Ross C. Walker, Andreas W. Götz (2016). Electronic Structure Calculations on Graphics Processing Units: From Quantum Chemistry to Condensed Matter Physics. John Wiley & Sons



Forum Posts

2005 ...

2010 ...


Re: Possible Board Presentation and Move Generation for GPUs by Steffan Westcott, CCC, March 20, 2011



2015 ...



Re: How good is the RTX 2080 Ti for Leela? by Ankan Banerjee, CCC, September 16, 2018


2020 ...

External Links




Deep Learning

Game Programming

GitHub - gcp/leela-zero: Go engine with no human-provided knowledge, modeled after the AlphaGo Zero paper

Chess Programming


  1. Graphics processing unit - Wikimedia Commons
  2. CUDA C Programming Guide v7.0, Appendix G.COMPUTE CAPABILITIES
  3. AMD Accelerated Parallel Processing OpenCL Programming Guide rev2.7, Appendix D Device Parameters, Table D.1 Parameters for 7xxx Devices
  5. AnandTech - Nvidia Turing Deep Dive page 6
  6. Wikipedia - Ampere microarchitecture
  7. CUDA C Programming Guide v7.0, Chapter 5.4.1. Arithmetic Instructions
  8. AMD_OpenCL_Programming_Optimization_Guide.pdf 3.0beta, Chapter 2.7.1 Instruction Bandwidths
  9. host-device latencies? by Srdja Matovic, Nvidia CUDA ZONE, Feb 28, 2019
  10. host-device latencies? by Srdja Matovic AMD Developer Community, Feb 28, 2019
  11. Re: GPU ANN, how to deal with host-device latencies? by Milos Stanisavljevic, CCC, May 06, 2018
  12. NVIDIA Ampere Architecture In-Depth | NVIDIA Developer Blog by Ronny Krashinsky, Olivier Giroux, Stephen Jones, Nick Stam and Sridhar Ramaswamy, May 14, 2020
  13. CUDA 11 Features Revealed | NVIDIA Developer Blog by Pramod Ramarao, May 14, 2020
  14. Photon mapping from Wikipedia
  15. Cell (microprocessor) from Wikipedia
  16. Jetson TK1 Embedded Development Kit | NVIDIA
  17. Jetson GPU architecture by Dann Corbit, CCC, October 18, 2016
  18. PowerVR from Wikipedia
  19. Density functional theory from Wikipedia
  20. Yaron Shoham, Sivan Toledo (2002). Parallel Randomized Best-First Minimax Search. Artificial Intelligence, Vol. 137, Nos. 1-2
  21. Alberto Maria Segre, Sean Forman, Giovanni Resta, Andrew Wildenberg (2002). Nagging: A Scalable Fault-Tolerant Paradigm for Distributed Search. Artificial Intelligence, Vol. 140, Nos. 1-2
  22. Tesla K20 GPU Compute Processor Specifications Released | techPowerUp
  23. Parallel Thread Execution from Wikipedia
  24. NVIDIA Compute PTX: Parallel Thread Execution, ISA Version 1.4, March 31, 2009, pdf
  25. ankan-ban/perft_gpu · GitHub
  26. Tensor processing unit from Wikipedia
  27. GeForce 20 series from Wikipedia
  28. Phoronix Test Suite from Wikipedia
  29. kernel launch latency - CUDA / CUDA Programming and Performance - NVIDIA Developer Forums by LukeCuda, June 18, 2018
  30. Re: Generate EGTB with graphics cards? by Graham Jones, CCC, January 01, 2019
  31. Fast perft on GPU (upto 20 Billion nps w/o hashing) by Ankan Banerjee, CCC, June 22, 2013

Up one Level