Files
amd-strix-halo-toolboxes/benchmark/results/Qwen3-Coder-30B-A3B-Instruct-Q4_K_M__rocm7-nightlies__fa1__longctx32768.log
T
Donato Capitella 2d09b9e6db updated benchmarks
2026-02-05 19:03:13 +00:00

9 lines
689 B
Plaintext

ggml_cuda_init: found 1 ROCm devices:
Device 0: Radeon 8060S Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32
| model | size | params | backend | ngl | n_ubatch | fa | test | t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -------: | -: | --------------: | -------------------: |
| qwen3moe 30B.A3B Q4_K - Medium | 17.35 GiB | 30.53 B | ROCm | 99 | 2048 | 1 | pp2048 @ d32768 | 176.98 ± 4.28 |
| qwen3moe 30B.A3B Q4_K - Medium | 17.35 GiB | 30.53 B | ROCm | 99 | 2048 | 1 | tg32 @ d32768 | 35.43 ± 0.01 |
build: e0c93af2a (7938)