[AMD] Nightly Test Coverage - Minimax-M3-MXFP8 Accuracy Test (#30613)

This commit is contained in:
YC Yen-Ching Tseng
2026-07-25 23:52:22 -07:00
committed by GitHub
parent 2cbddb842d
commit 1d0cd2e473
3 changed files with 351 additions and 0 deletions
@@ -87,6 +87,8 @@ on:
- nightly-8-gpu-mi35x-glm5-mxfp4-rocm720
# 4-GPU MiniMax-M2.5 (MI35x)
- nightly-4-gpu-mi35x-minimax-m25-rocm720
# 4-GPU MiniMax-M3 MXFP8 (MI35x)
- nightly-4-gpu-mi35x-minimax-m3-rocm720
# 8-GPU MiniMax-M2.7 (MI30x only)
- nightly-8-gpu-minimax-m27-rocm720
# Diffusion (MI30x)
@@ -1890,6 +1892,44 @@ jobs:
echo "$(<github_summary.md )" >> $GITHUB_STEP_SUMMARY || true
exit ${TEST_EXIT_CODE:-0}
# ==============================================================================
# 4-GPU MiniMax-M3 MXFP8 (MI35x)
# ==============================================================================
nightly-4-gpu-mi35x-minimax-m3-rocm720:
if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-4-gpu-mi35x-minimax-m3-rocm720,'))
runs-on: linux-mi35x-gpu-8
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ inputs.ref || github.sha }}
- name: Ensure VRAM is clear
run: bash scripts/ci/amd/ensure_vram_clear.sh rocm
- name: Setup docker (ROCm 7.2)
run: |
touch github_summary.md
bash scripts/ci/amd/amd_ci_start_container.sh --rocm-version rocm720
env:
GITHUB_WORKSPACE: ${{ github.workspace }}
ENABLE_CACHE_HOST: "1"
- name: Install dependencies
run: bash scripts/ci/amd/amd_ci_install_dependency.sh --skip-test-time-deps
- name: Accuracy Test MI35x ROCm 7.2 (4-GPU MiniMax-M3 MXFP8)
timeout-minutes: 120
run: |
> github_summary.md
bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \
-e SGLANG_USE_AITER=1 \
-e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \
python3 run_suite.py --hw amd --suite nightly-amd-4-gpu-mi35x-minimax-m3-tp4 --nightly --timeout-per-file 5400 ${{ (github.event_name == 'schedule' || inputs.continue_on_error) && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$?
echo "$(<github_summary.md )" >> $GITHUB_STEP_SUMMARY || true
exit ${TEST_EXIT_CODE:-0}
# ==============================================================================
# 8-GPU MiniMax-M2.7 (MI30x only)
# ==============================================================================
@@ -2045,6 +2085,8 @@ jobs:
- nightly-8-gpu-mi35x-glm5-mxfp4-rocm720
# 4-GPU MiniMax-M2.5 (MI35x)
- nightly-4-gpu-mi35x-minimax-m25-rocm720
# 4-GPU MiniMax-M3 MXFP8 (MI35x)
- nightly-4-gpu-mi35x-minimax-m3-rocm720
# 8-GPU MiniMax-M2.7 (MI30x only)
- nightly-8-gpu-minimax-m27-rocm720
# Diffusion (MI30x)
+42
View File
@@ -83,6 +83,8 @@ on:
- nightly-8-gpu-mi35x-glm5-mxfp4
# 4-GPU MiniMax-M2.5 (MI35x)
- nightly-4-gpu-mi35x-minimax-m25
# 4-GPU MiniMax-M3 MXFP8 (MI35x)
- nightly-4-gpu-mi35x-minimax-m3
# 8-GPU MiniMax-M2.7 (MI30x only)
- nightly-8-gpu-minimax-m27
# Diffusion (MI30x)
@@ -1744,6 +1746,44 @@ jobs:
echo "$(<github_summary.md )" >> $GITHUB_STEP_SUMMARY || true
exit ${TEST_EXIT_CODE:-0}
# ==============================================================================
# 4-GPU MiniMax-M3 MXFP8 (MI35x)
# ==============================================================================
nightly-4-gpu-mi35x-minimax-m3:
if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-4-gpu-mi35x-minimax-m3,'))
runs-on: linux-mi35x-gpu-8
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ inputs.ref || github.sha }}
- name: Ensure VRAM is clear
run: bash scripts/ci/amd/ensure_vram_clear.sh rocm
- name: Setup docker (ROCm 7.0)
run: |
touch github_summary.md
bash scripts/ci/amd/amd_ci_start_container.sh
env:
GITHUB_WORKSPACE: ${{ github.workspace }}
ENABLE_CACHE_HOST: "1"
- name: Install dependencies
run: bash scripts/ci/amd/amd_ci_install_dependency.sh --skip-test-time-deps
- name: Accuracy Test MI35x (4-GPU MiniMax-M3 MXFP8)
timeout-minutes: 120
run: |
> github_summary.md
bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \
-e SGLANG_USE_AITER=1 \
-e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \
python3 run_suite.py --hw amd --suite nightly-amd-4-gpu-mi35x-minimax-m3-tp4 --nightly --timeout-per-file 5400 ${{ (github.event_name == 'schedule' || inputs.continue_on_error) && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$?
echo "$(<github_summary.md )" >> $GITHUB_STEP_SUMMARY || true
exit ${TEST_EXIT_CODE:-0}
# ==============================================================================
# 8-GPU MiniMax-M2.7 (MI30x only)
# ==============================================================================
@@ -1895,6 +1935,8 @@ jobs:
- nightly-8-gpu-mi35x-glm5-mxfp4
# 4-GPU MiniMax-M2.5 (MI35x)
- nightly-4-gpu-mi35x-minimax-m25
# 4-GPU MiniMax-M3 MXFP8 (MI35x)
- nightly-4-gpu-mi35x-minimax-m3
# 8-GPU MiniMax-M2.7 (MI30x only)
- nightly-8-gpu-minimax-m27
# Diffusion (MI30x)