clpeak-1 AMD EPYC 7F52 16-Core testing with a ASRockRack ROMED8-2T v1.03 (P3.80 BIOS) and llvmpipe on Debian 12 via the Phoronix Test Suite. gpu-1: Processor: AMD EPYC 7F52 16-Core @ 3.50GHz (16 Cores / 32 Threads), Motherboard: ASRockRack ROMED8-2T v1.03 (P3.80 BIOS), Chipset: AMD Starship/Matisse, Memory: 8 x 32GB DDR4-3200MT/s Samsung M393A4K40EB3-CWE, Disk: 1000GB Samsung SSD 980 PRO 1TB, Graphics: llvmpipe, Audio: NVIDIA GA102 HD Audio, Monitor: SDM-X93, Network: 2 x Intel X550 OS: Debian 12, Kernel: 6.1.0-30-amd64 (x86_64), Desktop: Cinnamon 5.6.8, Display Server: X Server 1.21.1.7, Display Driver: NVIDIA 565.57.01, OpenGL: 4.5 Mesa 22.3.6 (LLVM 15.0.6 256 bits), OpenCL: OpenCL 3.0 CUDA 12.7.33, Compiler: GCC 12.2.0 + CUDA 12.6, File-System: ext4, Screen Resolution: 1280x1024 clpeak 1.1.2 OpenCL Test: Kernel Latency us < Lower Is Better gpu-1 . 6.47 |================================================================= clpeak 1.1.2 OpenCL Test: Integer Compute GIOPS > Higher Is Better gpu-1 . 17811.62 |============================================================= clpeak 1.1.2 OpenCL Test: Integer 24-bit Compute GIOPS > Higher Is Better gpu-1 . 17673.17 |============================================================= clpeak 1.1.2 OpenCL Test: Global Memory Bandwidth GBPS > Higher Is Better gpu-1 . 838.73 |=============================================================== clpeak 1.1.2 OpenCL Test: Double-Precision Compute GFLOPS > Higher Is Better gpu-1 . 641.34 |=============================================================== clpeak 1.1.2 OpenCL Test: Single-Precision Compute GFLOPS > Higher Is Better gpu-1 . 34568.78 |============================================================= clpeak 1.1.2 OpenCL Test: Transfer Bandwidth enqueueReadBuffer GBPS > Higher Is Better gpu-1 . 12.24 |================================================================ clpeak 1.1.2 OpenCL Test: Transfer Bandwidth enqueueWriteBuffer GBPS > Higher Is Better gpu-1 . 15.96 |================================================================