GPTshop.ai NVIDIA GH200 Linux Benchmarks
Benchmarks by Michael Larabel for a future article.
GPTshop.ai GH200
Processor: ARMv8 Neoverse-V2 @ 3.39GHz (72 Cores), Motherboard: Quanta Cloud QuantaGrid S74G-2U 1S7GZ9Z0000 S7G MB (CG1) (3A06 BIOS), Memory: 1 x 480GB DRAM-6400MT/s, Disk: 960GB SAMSUNG MZ1L2960HCJR-00A07 + 1920GB SAMSUNG MZTL21T9, Graphics: ASPEED, Network: 2 x Mellanox MT2910 + 2 x QLogic FastLinQ QL41000 10/25/40/50GbE
OS: Ubuntu 23.10, Kernel: 6.5.0-15-generic (aarch64), Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1200
Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-fix-cortex-a53-843419 --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-nls --enable-objc-gc=auto --enable-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --program-prefix=aarch64-linux-gnu- --target=aarch64-linux-gnu --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-target-system-zlib=auto -v
Processor Notes: Scaling Governor: cppc_cpufreq performance (Boost: Disabled)
Python Notes: Python 3.11.6
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected
HP Z6 G5 A - Threadripper PRO 7995WX
Processor: AMD Ryzen Threadripper PRO 7995WX 96-Cores @ 6.44GHz (96 Cores / 192 Threads), Motherboard: HP Z6 G5 A Workstation 8B24 (U65 Ver. 01.01.04 BIOS), Chipset: AMD Device 14a4, Memory: 8 x 16GB DRAM-5200MT/s Hynix HMCG78AGBRA190N, Disk: 2 x 1024GB SAMSUNG MZVL21T0HCLR-00BH1, Graphics: NVIDIA RTX A4000 16GB, Audio: NVIDIA GA104 HD Audio, Monitor: ASUS VP28U, Network: Realtek RTL8111/8168/8411
OS: Ubuntu 23.10, Kernel: 6.5.0-17-generic (x86_64), Desktop: GNOME Shell 45.2, Display Server: X Server 1.21.1.7, Display Driver: NVIDIA 535.154.05, OpenGL: 4.6.0, OpenCL: OpenCL 3.0 CUDA 12.2.148, Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 3840x2160
Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v
Processor Notes: Scaling Governor: amd-pstate-epp performance (EPP: performance) - CPU Microcode: 0xa108105
OpenCL Notes: GPU Compute Cores: 6144
Python Notes: Python 3.11.6
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Mitigation of safe RET + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced / Automatic IBRS IBPB: conditional STIBP: always-on RSB filling PBRSB-eIBRS: Not affected + srbds: Not affected + tsx_async_abort: Not affected
7-Zip Compression
This is a test of 7-Zip compression/decompression with its integrated benchmark feature. Learn more via the OpenBenchmarking.org test page.
ACES DGEMM
This is a multi-threaded DGEMM benchmark. Learn more via the OpenBenchmarking.org test page.
Algebraic Multi-Grid Benchmark
AMG is a parallel algebraic multigrid solver for linear systems arising from problems on unstructured grids. The driver provided with AMG builds linear systems for various 3-dimensional problems. Learn more via the OpenBenchmarking.org test page.
ASKAP
ASKAP is a set of benchmarks from the Australian SKA Pathfinder. The principal ASKAP benchmarks are the Hogbom Clean Benchmark (tHogbomClean) and Convolutional Resamping Benchmark (tConvolve) as well as some previous ASKAP benchmarks being included as well for OpenCL and CUDA execution of tConvolve. Learn more via the OpenBenchmarking.org test page.
asmFish
This is a test of asmFish, an advanced chess benchmark written in Assembly. Learn more via the OpenBenchmarking.org test page.
ASTC Encoder
ASTC Encoder (astcenc) is for the Adaptive Scalable Texture Compression (ASTC) format commonly used with OpenGL, OpenGL ES, and Vulkan graphics APIs. This test profile does a coding test of both compression/decompression. Learn more via the OpenBenchmarking.org test page.
Preset: Medium
GPTshop.ai GH200: The test quit with a non-zero exit status. E: ./astcenc: 2: ./astc-encoder-4.0.0/build/Source/astcenc-neon: not found
Preset: Thorough
GPTshop.ai GH200: The test quit with a non-zero exit status. E: ./astcenc: 2: ./astc-encoder-4.0.0/build/Source/astcenc-neon: not found
Preset: Exhaustive
GPTshop.ai GH200: The test quit with a non-zero exit status. E: ./astcenc: 2: ./astc-encoder-4.0.0/build/Source/astcenc-neon: not found
Coremark
This is a test of EEMBC CoreMark processor benchmark. Learn more via the OpenBenchmarking.org test page.
Cpuminer-Opt
Cpuminer-Opt is a fork of cpuminer-multi that carries a wide range of CPU performance optimizations for measuring the potential cryptocurrency mining performance of the CPU/processor with a wide variety of cryptocurrencies. The benchmark reports the hash speed for the CPU mining performance for the selected cryptocurrency. Learn more via the OpenBenchmarking.org test page.
Algorithm: Deepcoin
GPTshop.ai GH200: The test quit with a non-zero exit status. E: ./cpuminer-opt: 3: ./cpuminer: not found
Algorithm: Blake-2 S
GPTshop.ai GH200: The test quit with a non-zero exit status. E: ./cpuminer-opt: 3: ./cpuminer: not found
Algorithm: Myriad-Groestl
GPTshop.ai GH200: The test quit with a non-zero exit status. E: ./cpuminer-opt: 3: ./cpuminer: not found
Algorithm: Triple SHA-256, Onecoin
GPTshop.ai GH200: The test quit with a non-zero exit status. E: ./cpuminer-opt: 3: ./cpuminer: not found
DuckDB
DuckDB is an in-progress SQL OLAP database management system optimized for analytics and features a vectorized and parallel engine. Learn more via the OpenBenchmarking.org test page.
Graph500
This is a benchmark of the reference implementation of Graph500, an HPC benchmark focused on data intensive loads and commonly tested on supercomputers for complex data problems. Graph500 primarily stresses the communication subsystem of the hardware under test. Learn more via the OpenBenchmarking.org test page.
GraphicsMagick
This is a test of GraphicsMagick with its OpenMP implementation that performs various imaging tests on a sample 6000x4000 pixel JPEG image. Learn more via the OpenBenchmarking.org test page.
GROMACS
The GROMACS (GROningen MAchine for Chemical Simulations) molecular dynamics package testing with the water_GMX50 data. This test profile allows selecting between CPU and GPU-based GROMACS builds. Learn more via the OpenBenchmarking.org test page.
Helsing
Helsing is an open-source POSIX vampire number generator. This test profile measures the time it takes to generate vampire numbers between varying numbers of digits. Learn more via the OpenBenchmarking.org test page.
High Performance Conjugate Gradient
HPCG is the High Performance Conjugate Gradient and is a new scientific benchmark from Sandia National Lans focused for super-computer testing with modern real-world workloads compared to HPCC. Learn more via the OpenBenchmarking.org test page.
X Y Z: 144 144 144 - RT: 60
HP Z6 G5 A - Threadripper PRO 7995WX: The test quit with a non-zero exit status. E: cat: 'HPCG-Benchmark*.txt': No such file or directory
John The Ripper
This is a benchmark of John The Ripper, which is a password cracker. Learn more via the OpenBenchmarking.org test page.
libxsmm
Libxsmm is an open-source library for specialized dense and sparse matrix operations and deep learning primitives. Libxsmm supports making use of Intel AMX, AVX-512, and other modern CPU instruction set capabilities. Learn more via the OpenBenchmarking.org test page.
M N K: 128
GPTshop.ai GH200: The test quit with a non-zero exit status. E: Error: no specialized routine found!
M N K: 256
GPTshop.ai GH200: The test quit with a non-zero exit status. E: Error: no specialized routine found!
Liquid-DSP
LiquidSDR's Liquid-DSP is a software-defined radio (SDR) digital signal processing library. This test profile runs a multi-threaded benchmark of this SDR/DSP library focused on embedded platform usage. Learn more via the OpenBenchmarking.org test page.
LULESH
LULESH is the Livermore Unstructured Lagrangian Explicit Shock Hydrodynamics. Learn more via the OpenBenchmarking.org test page.
miniBUDE
MiniBUDE is a mini application for the the core computation of the Bristol University Docking Engine (BUDE). This test profile currently makes use of the OpenMP implementation of miniBUDE for CPU benchmarking. Learn more via the OpenBenchmarking.org test page.
NAS Parallel Benchmarks
NPB, NAS Parallel Benchmarks, is a benchmark developed by NASA for high-end computer systems. This test profile currently uses the MPI version of NPB. This test profile offers selecting the different NPB tests/problems and varying problem sizes. Learn more via the OpenBenchmarking.org test page.
NWChem
NWChem is an open-source high performance computational chemistry package. Per NWChem's documentation, "NWChem aims to provide its users with computational chemistry tools that are scalable both in their ability to treat large scientific computational chemistry problems efficiently, and in their use of available parallel computing resources from high-performance parallel supercomputers to conventional workstation clusters." Learn more via the OpenBenchmarking.org test page.
oneDNN
This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of the Intel oneAPI toolkit. Learn more via the OpenBenchmarking.org test page.
Harness: Convolution Batch Shapes Auto - Data Type: bf16bf16bf16 - Engine: CPU
GPTshop.ai GH200: The test run did not produce a result.
Harness: Deconvolution Batch shapes_1d - Data Type: bf16bf16bf16 - Engine: CPU
GPTshop.ai GH200: The test run did not produce a result.
OpenVINO
This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.
PostgreSQL
This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.
Primesieve
Primesieve generates prime numbers using a highly optimized sieve of Eratosthenes implementation. Primesieve primarily benchmarks the CPU's L1/L2 cache performance. Learn more via the OpenBenchmarking.org test page.
RawTherapee
RawTherapee is a cross-platform, open-source multi-threaded RAW image processing program. Learn more via the OpenBenchmarking.org test page.
rays1bench
This is a test of rays1bench, a simple path-tracer / ray-tracing that supports SSE and AVX instructions, multi-threading, and other features. This test profile is measuring the performance of the "large scene" in rays1bench. Learn more via the OpenBenchmarking.org test page.
Large Scene
GPTshop.ai GH200: The test quit with a non-zero exit status. E: FileNotFoundError: [Errno 2] No such file or directory: './a.out'
Rodinia
Rodinia is a suite focused upon accelerating compute-intensive applications with accelerators. CUDA, OpenMP, and OpenCL parallel models are supported by the included applications. This profile utilizes select OpenCL, NVIDIA CUDA and OpenMP test binaries at the moment. Learn more via the OpenBenchmarking.org test page.
Stockfish
This is a test of Stockfish, an advanced open-source C++11 chess benchmark that can scale up to 512 CPU threads. Learn more via the OpenBenchmarking.org test page.
Stress-NG
Stress-NG is a Linux stress tool developed by Colin Ian King. Learn more via the OpenBenchmarking.org test page.
SVT-AV1
Tachyon
This is a test of the threaded Tachyon, a parallel ray-tracing system, measuring the time to ray-trace a sample scene. The sample scene used is the Teapot scene ray-traced to 8K x 8K with 32 samples. Learn more via the OpenBenchmarking.org test page.
Total Time
GPTshop.ai GH200: The test run did not produce a result. E: ./tachyon-benchmark: 3: ./tachyon: not found
Timed Gem5 Compilation
This test times how long it takes to compile Gem5. Gem5 is a simulator for computer system architecture research. Gem5 is widely used for computer architecture research within the industry, academia, and more. Learn more via the OpenBenchmarking.org test page.
Timed Godot Game Engine Compilation
This test times how long it takes to compile the Godot Game Engine. Godot is a popular, open-source, cross-platform 2D/3D game engine and is built using the SCons build system and targeting the X11 platform. Learn more via the OpenBenchmarking.org test page.
Timed Linux Kernel Compilation
This test times how long it takes to build the Linux kernel in a default configuration (defconfig) for the architecture being tested or alternatively an allmodconfig for building all possible kernel modules for the build. Learn more via the OpenBenchmarking.org test page.
Timed LLVM Compilation
This test times how long it takes to compile/build the LLVM compiler stack. Learn more via the OpenBenchmarking.org test page.
Timed Node.js Compilation
This test profile times how long it takes to build/compile Node.js itself from source. Node.js is a JavaScript run-time built from the Chrome V8 JavaScript engine while itself is written in C/C++. Learn more via the OpenBenchmarking.org test page.
Xcompact3d Incompact3d
Xcompact3d Incompact3d is a Fortran-MPI based, finite difference high-performance code for solving the incompressible Navier-Stokes equation and as many as you need scalar transport equations. Learn more via the OpenBenchmarking.org test page.
Xmrig
Xmrig is an open-source cross-platform CPU/GPU miner for RandomX, KawPow, CryptoNight and AstroBWT. This test profile is setup to measure the Xmlrig CPU mining performance. Learn more via the OpenBenchmarking.org test page.
GPTshop.ai GH200
Processor: ARMv8 Neoverse-V2 @ 3.39GHz (72 Cores), Motherboard: Quanta Cloud QuantaGrid S74G-2U 1S7GZ9Z0000 S7G MB (CG1) (3A06 BIOS), Memory: 1 x 480GB DRAM-6400MT/s, Disk: 960GB SAMSUNG MZ1L2960HCJR-00A07 + 1920GB SAMSUNG MZTL21T9, Graphics: ASPEED, Network: 2 x Mellanox MT2910 + 2 x QLogic FastLinQ QL41000 10/25/40/50GbE
OS: Ubuntu 23.10, Kernel: 6.5.0-15-generic (aarch64), Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1200
Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-fix-cortex-a53-843419 --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-nls --enable-objc-gc=auto --enable-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --program-prefix=aarch64-linux-gnu- --target=aarch64-linux-gnu --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-target-system-zlib=auto -v
Processor Notes: Scaling Governor: cppc_cpufreq performance (Boost: Disabled)
Python Notes: Python 3.11.6
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected
Testing initiated at 5 February 2024 20:48 by user x.
HP Z6 G5 A - Threadripper PRO 7995WX
Processor: AMD Ryzen Threadripper PRO 7995WX 96-Cores @ 6.44GHz (96 Cores / 192 Threads), Motherboard: HP Z6 G5 A Workstation 8B24 (U65 Ver. 01.01.04 BIOS), Chipset: AMD Device 14a4, Memory: 8 x 16GB DRAM-5200MT/s Hynix HMCG78AGBRA190N, Disk: 2 x 1024GB SAMSUNG MZVL21T0HCLR-00BH1, Graphics: NVIDIA RTX A4000 16GB, Audio: NVIDIA GA104 HD Audio, Monitor: ASUS VP28U, Network: Realtek RTL8111/8168/8411
OS: Ubuntu 23.10, Kernel: 6.5.0-17-generic (x86_64), Desktop: GNOME Shell 45.2, Display Server: X Server 1.21.1.7, Display Driver: NVIDIA 535.154.05, OpenGL: 4.6.0, OpenCL: OpenCL 3.0 CUDA 12.2.148, Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 3840x2160
Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v
Processor Notes: Scaling Governor: amd-pstate-epp performance (EPP: performance) - CPU Microcode: 0xa108105
OpenCL Notes: GPU Compute Cores: 6144
Python Notes: Python 3.11.6
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Mitigation of safe RET + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced / Automatic IBRS IBPB: conditional STIBP: always-on RSB filling PBRSB-eIBRS: Not affected + srbds: Not affected + tsx_async_abort: Not affected
Testing initiated at 17 February 2024 17:07 by user phoronix.