Skip to content

[atom-sglang-ci] Add qwen3.5-35b-ptpc-fp8 for atom sglang on mi308 #1561

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
zhuyuhua-v merged 1 commit into from
Jul 10, 2026
+38 −0
Merged

[atom-sglang-ci] Add qwen3.5-35b-ptpc-fp8 for atom sglang on mi308 #1561

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 16 additions & 0 deletions .github/benchmark/sglang_benchmark_models.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -479,6 +479,22 @@
"nightly_group": "B",
"env_vars": "qwen_common"
},
{
"display": "Qwen3.5-35B-A3B-PTPC-FP8 TP1 MI308",
"dashboard_model": "Qwen3.5-35B-A3B-PTPC-FP8 MI308",
"workload_label": "SGLang-OOB",
"source_path": "amd/Qwen3.5-35B-A3B-PTPC-FP8",
"path": "amd/Qwen3.5-35B-A3B-PTPC-FP8",
"prefix": "qwen3-5-35b-a3b-ptpc-fp8-tp1-mi308",
"extra_args": [
"--tensor-parallel-size 1",
"qwen_reasoning"
],
"bench_args": "",
"runner": "atom-mi308-8gpu-plugins-benchmark",
"nightly_group": "B",
"env_vars": "qwen_common"
},
{
"display": "Qwen3-32B-FP8 TP1 MI308",
"dashboard_model": "Qwen3-32B-FP8 MI308",
Expand Down
12 changes: 12 additions & 0 deletions .github/benchmark/sglang_models_accuracy.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -344,6 +344,18 @@
"accuracy_baseline_model": "Qwen/Qwen3.5-35B-A3B-FP8",
"_baseline_note": "Threshold aligned with the SGLANG accuracy validation workflow target for gsm8k."
},
{
"model_name": "MI308 Qwen3.5-35B-A3B-PTPC-FP8 TP1",
"model_path": "amd/Qwen3.5-35B-A3B-PTPC-FP8",
"extraArgs": "--tensor-parallel-size 1 --mem-fraction-static 0.9 --reasoning-parser qwen3 --disable-radix-cache",
"env_vars": "SGLANG_DEFAULT_SERVER_ARGS=\nSGLANG_EXTERNAL_MODEL_PACKAGE=atom.plugin.sglang.models\nATOM_ENABLE_QK_NORM_ROPE_CACHE_QUANT_FUSION=0",
"runner": "atom-mi308-8gpu-plugins-accuracy",
"test_level": "nightly",
"accuracy_threshold": 0.76,
"accuracy_baseline": null,
"accuracy_baseline_model": "amd/Qwen3.5-35B-A3B-PTPC-FP8",
"_baseline_note": "Threshold aligned with the SGLANG accuracy validation workflow target for gsm8k."
},
{
"model_name": "MI308 Qwen3-32B-FP8 TP1",
"model_path": "Qwen/Qwen3-32B-FP8",
Expand Down
9 changes: 9 additions & 0 deletions .github/workflows/atom-sglang-accuracy-validation.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,6 +50,7 @@ on:
- MI308 Qwen3.5-397B-A17B-FP8 TP4
- MI308 Qwen3.5-397B-A17B-FP8 TP8
- MI308 Qwen3.5-35B-A3B-FP8 TP1
- MI308 Qwen3.5-35B-A3B-PTPC-FP8 TP1
- MI308 Qwen3-32B-FP8 TP1
- MI308 Qwen3-32B-FP8 TP8
- MI308 GLM-5.1-FP8 TP8
Expand Down Expand Up @@ -384,6 +385,14 @@ jobs:
"env_vars": "SGLANG_DEFAULT_SERVER_ARGS=\nSGLANG_EXTERNAL_MODEL_PACKAGE=atom.plugin.sglang.models\nATOM_ENABLE_QK_NORM_ROPE_CACHE_QUANT_FUSION=0",
"runner": "atom-mi308-8gpu-plugins-accuracy",
},
{
"model_name": "MI308 Qwen3.5-35B-A3B-PTPC-FP8 TP1",
"model_path": "amd/Qwen3.5-35B-A3B-PTPC-FP8",
"extra_args": "--tensor-parallel-size 1 --mem-fraction-static 0.9 --reasoning-parser qwen3 --disable-radix-cache",
"accuracy_test_threshold": 0.76,
"env_vars": "SGLANG_DEFAULT_SERVER_ARGS=\nSGLANG_EXTERNAL_MODEL_PACKAGE=atom.plugin.sglang.models\nATOM_ENABLE_QK_NORM_ROPE_CACHE_QUANT_FUSION=0",
"runner": "atom-mi308-8gpu-plugins-accuracy",
},
{
"model_name": "MI308 Qwen3-32B-FP8 TP1",
"model_path": "Qwen/Qwen3-32B-FP8",
Expand Down
1 change: 1 addition & 0 deletions .github/workflows/atom-sglang-benchmark.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -46,6 +46,7 @@ on:
- "qwen3-5-397b-a17b-fp8-tp4-mi308 (1024x1024/8192x1024: [4,8,16,32,64])"
- "qwen3-5-397b-a17b-fp8-tp8-mi308 (1024x1024/8192x1024: [4,8,16,32,64])"
- "qwen3-5-35b-a3b-fp8-tp1-mi308 (1024x1024/8192x1024: [4,8,16,32,64])"
- "qwen3-5-35b-a3b-ptpc-fp8-tp1-mi308 (1024x1024/8192x1024: [4,8,16,32,64])"
- "qwen3-32b-fp8-tp1-mi308 (1024x1024/8192x1024: [4,8,16,32,64])"
- "qwen3-32b-fp8-tp8-mi308 (1024x1024/8192x1024: [4,8,16,32,64])"
- "deepseek-v4-flash-fp8-mi308 (1024x1024/8192x1024: [4,8,16,32,64])"
Expand Down
Loading