[gfx1201] add tuned ck_gemm_a8w8_blockscale configs for various qwen3 models and default case

[big-yellow-duck]

Motivation

Add tuned configs for gfx1201 ck_gemm_a8w8_blockscale kernel, targeting various Qwen3 model variants. gfx1201 supports the FP8 dtype so these tuned configs speed up the gemm_a8w8_blockscale for inference in vLLM.

Technical Details

Tuning Process

The tuning was performed using the CK GEMM tuner:

cd csrc/ck_gemm_a8w8_blockscale
python gemm_a8w8_blockscale_tune.py \
    -i /app/aiter/aiter/configs/a8w8_blockscale_untuned_gemm.csv \
    -o <output.csv>

The tuned configurations are added to the existing GEMM configuration files and are automatically selected based on the input tensor dimensions and the target architecture (gfx1201).

Test Plan

The tuned kernels were validated using the GEMM test suite:

python op_tests/test_gemm_a8w8_blockscale.py --ck_preshuffle False

Tests cover various matrix dimensions (M: 1-10240, N: 24576, K: 1536) that are representative of Qwen3 inference workloads.

Test Result

dtype	m	n	k	ck_preshuffle	ck us	ck TFLOPS	ck TB/s	ck err
torch.bfloat16	1	24576	1536	False	62.2419	1.21297	0.606509	0
torch.bfloat16	2	24576	1536	False	63.0116	2.39631	0.599125	0
torch.bfloat16	4	24576	1536	False	63.6247	4.74642	0.593399	0
torch.bfloat16	8	24576	1536	False	64.1253	9.41875	0.588863	0
torch.bfloat16	16	24576	1536	False	65.0905	18.5582	0.58032	0
torch.bfloat16	32	24576	1536	False	68.2576	35.3942	0.553754	0
torch.bfloat16	64	24576	1536	False	80.5625	59.9763	0.469785	0
torch.bfloat16	96	24576	1536	False	122.323	59.251	0.309805	0
torch.bfloat16	128	24576	1536	False	149.707	64.5505	0.253464	0
torch.bfloat16	160	24576	1536	False	191.106	63.209	0.198814	0
torch.bfloat16	192	24576	1536	False	225.946	64.1546	0.168375	0
torch.bfloat16	224	24576	1536	False	255.417	66.211	0.14914	0
torch.bfloat16	256	24576	1536	False	282.715	68.3633	0.134913	0
torch.bfloat16	288	24576	1536	False	320.135	67.9191	0.119297	0
torch.bfloat16	320	24576	1536	False	354.852	68.0824	0.107764	0
torch.bfloat16	352	24576	1536	False	386.663	68.7293	0.0990251	0
torch.bfloat16	384	24576	1536	False	415.049	69.8496	0.0923711	0
torch.bfloat16	416	24576	1536	False	451.077	69.6265	0.0851023	0
torch.bfloat16	448	24576	1536	False	484.864	69.7575	0.0792736	0
torch.bfloat16	480	24576	1536	False	516.881	70.1105	0.0744581	0
torch.bfloat16	512	24576	1536	False	544.023	71.0534	0.0708337	0
torch.bfloat16	1024	24576	1536	False	1038.07	74.4739	0.0378794	0
torch.bfloat16	2048	24576	1536	False	2032.53	76.072	0.02012	0
torch.bfloat16	4096	24576	1536	False	3959.11	78.1079	0.0111238	0
torch.bfloat16	6144	24576	1536	False	5945.33	78.0203	0.00793664	0
torch.bfloat16	8192	24576	1536	False	7990.64	77.3999	0.00629882	0
torch.bfloat16	10240	24576	1536	False	10011.5	77.2203	0.00534157	0

All tests pass with zero error, confirming the correctness of the tuned configurations.

Submission Checklist

• Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

[github-actions]

🏷️ CI Guide

Runs automatically on every PR:

• ✅ Pre-checks (submodule verification, code formatting)
• ✅ Aiter op tests (gfx942 + gfx950)
• ✅ Triton tests (only when aiter/ops/triton/** or related paths are changed)

Extended tests (opt-in via labels):

Label	Tests
ci:sglang	SGLang integration tests
ci:atom	ATOM benchmark (DeepSeek-R1 + GPT-OSS)
ci:vllm	vLLM benchmark
ci:all	All of the above

Add labels via the sidebar or gh pr edit 1 --add-label <label>