2.1.44558.3 MB
Apache-2.0
classic
core20
A synthetic micro-benchmark that measures peak compute, bandwidth, and matrix throughput of GPUs and CPUs
clpeak — "Compute Latency PEAK". A synthetic micro-benchmark that measures the peak achievable performance of GPU compute devices. It exercises tight vector / MAD / MMA loops and vendor-SDK GEMM libraries (cuBLASLt on NVIDIA, MPS on Apple) to expose what the hardware is capable of — from raw ALU peaks to near-vendor-advertised matrix throughput.
clpeak began as an OpenCL-only tool and is now a multi-backend benchmark — OpenCL, Vulkan, CUDA, ROCm/HIP, Metal, oneAPI/SYCL, plus a native CPU backend — run back-to-back on the same hardware, so cross-stack differences (driver lowering, instruction scheduling, extension exposure) surface alongside the raw peak numbers.
clpeak began as an OpenCL-only tool and is now a multi-backend benchmark — OpenCL, Vulkan, CUDA, ROCm/HIP, Metal, oneAPI/SYCL, plus a native CPU backend — run back-to-back on the same hardware, so cross-stack differences (driver lowering, instruction scheduling, extension exposure) surface alongside the raw peak numbers.
Update History
2.1.3 (445) → 2.1.4 (455)28 Aug 2026, 01:30 UTC
2.1.0 (428) → 2.1.3 (445)24 Aug 2026, 10:15 UTC
2.0.15 (396) → 2.1.0 (428)18 Aug 2026, 11:00 UTC
2.0.14 (389) → 2.0.15 (396)25 Jun 2026, 08:15 UTC
2.0.12 (379) → 2.0.14 (389)23 Jun 2026, 14:15 UTC
1.1.2 (256) → 2.0.12 (379)14 Jun 2026, 08:15 UTC
1.1.2 (256)1 Apr 2026, 21:28 UTC
10 Jul 2019, 13:56 UTC
27 Aug 2026, 20:06 UTC
1 Apr 2026, 21:28 UTC