Files
amd-strix-halo-toolboxes/benchmark/results/gpt-oss-20b-mxfp4__rocm7_rc__hblt0__fa1.log
T
Donato Capitella 765cc5c733 updated benchs
2025-10-12 07:39:42 +01:00

11 lines
756 B
Plaintext

ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 ROCm devices:
Device 0: AMD Radeon Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | fa | mmap | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -: | ---: | --------------: | -------------------: |
| gpt-oss 20B MXFP4 MoE | 11.27 GiB | 20.91 B | ROCm | 99 | 1 | 0 | pp512 | 1566.75 ± 13.55 |
| gpt-oss 20B MXFP4 MoE | 11.27 GiB | 20.91 B | ROCm | 99 | 1 | 0 | tg128 | 67.99 ± 0.01 |
build: a3cb0474 (6735)