[config] Add tuned configs for GLM-5.2-FP8 block-scale fp8 GEMM and MoE

[gbyu-amd]

Adds tuned (and source untuned) configs for GLM-5.2-FP8 running TP4 on gfx950:

• a8w8_blockscale_bpreshuffle GEMM: 4 shapes derived from GLM-5.2 linear layers (N6144/K3072, N6144/K4096, N6144/K6144, N7168/K512), swept over M in {1..32768}.
• fused MoE (a8w8_blockscale, per_1x128): hidden_dim=6144, inter_dim=512, expert=257, topk=9, swept over tokens {1..32768}.

Configs are auto-discovered by the model_configs glob merge; no code change required. Verified end-to-end: server startup + gsm8k 3-shot show zero fp8 GEMM / MoE default-config fallbacks, accuracy 0.9356 strict / 0.9348 flexible exact_match.

Motivation

Technical Details

Test Plan

Test Result

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 on MI35X (only when aiter/ops/triton/** or related paths are changed)

Extended tests (opt-in via labels):

Label	Tests
ci:triton-300x	Run an additional Triton test job on MI300X in PRs; main branch always runs both MI35X and MI300X
ci:sglang	SGLang integration tests: DeepSeek-R1-MXFP4 accuracy, Qwen 3.5 accuracy
ci:atom	ATOM benchmark: DeepSeek-R1-0528, GPT-OSS-120B
ci:atom_full	ATOM accuracy suite for PR and main models from ATOM models_accuracy.json
ci:vllm	vLLM benchmark: GPT-OSS-120B, DeepSeek-R1-0528, Kimi-K2.5
ci:all	All standard extended tests (excludes ci:atom_full)

Only add ci:atom_full for FlyDSL or Triton upgrades.
Add labels via the sidebar or gh pr edit 4038 --add-label <label>

[zejunchen-zejun]

#4074
Hi @gbyu-amd
We use online quant to make GLM5.2 FP8 MOE 1x32+E8M0 scale, so I tunned this moe kernel, could you help review

[yifehuan]

LGTM, one thing is the gfx column is not needed in fmoe tuned csv (not found in other csv), but it will auto-derive, it is ok

[zejunchen-zejun]

LGTM, one thing is the gfx column is not needed in fmoe tuned csv (not found in other csv), but it will auto-derive, it is ok

Hi @yifehuan
Thank you for review. This PR will be closed as the new PR is here: #4095. It adds all tuned config. Could you help review this new PR?
Thank you