【CI】add nightly cases: MiniMax-M2.5-W8A8 Qwen3.5-27B-w8a8 Qwen3.5-397B-A1… (#7968)

### What this PR does / why we need it?
This PR Qwen3.5-27B ;MiniMax-M2.5-w8a8 ;Qwen3.5-397B-w8a8-mtp acc/perf 3
cases on A3, we need test them daily.

- vLLM version: v0.18.0
- vLLM main:
35141a7eed

Signed-off-by: guxin108 <1252896542@qq.com>

.github/workflows/misc/model_list.json

@@ -245,6 +245,9 @@
       "xlangai/OpenCUA-7B",
       "Eco-Tech/GLM-5-w4a8",
       "Eco-Tech/GLM-4.7-W8A8-floatmtp",
-      "MiniMax/MiniMax-M2.5"
+      "MiniMax/MiniMax-M2.5",
+      "Eco-Tech/Qwen3.5-27B-w8a8-mtp",
+      "Eco-Tech/MiniMax-M2.5-w8a8-QuaRot",
+      "Eco-Tech/Qwen3.5-397B-A17B-w8a8-mtp"
     ]
   }

.github/workflows/schedule_nightly_test_a3.yaml

@@ -223,6 +223,15 @@ jobs:
           - name: deepseek-r1-0528-w8a8-prefix-cache
             os: linux-aarch64-a3-16
             config_file_path: Prefix-Cache-DeepSeek-R1-0528-W8A8.yaml
+          - name: Qwen3.5-27B-w8a8-A3
+            os: linux-aarch64-a3-2
+            config_file_path: Qwen3.5-27B-w8a8-A3.yaml
+          - name: MiniMax-M2.5-w8a8
+            os: linux-aarch64-a3-16
+            config_file_path: MiniMax-M2.5-W8A8-A3.yaml
+          - name: Qwen3.5-397B-A17B-w8a8-mtp
+            os: linux-aarch64-a3-16
+            config_file_path: Qwen3.5-397B-A17B-W8A8-mtp-A3.yaml
     uses: ./.github/workflows/_e2e_nightly_single_node.yaml
     with:
       runner: ${{ matrix.test_config.os }}

tests/e2e/nightly/single_node/models/configs/MiniMax-M2.5-W8A8-A3.yaml

@@ -0,0 +1,62 @@
+# ==========================================
+# ACTUAL TEST CASES
+# ==========================================
+
+test_cases:
+  - name: "MiniMax-M2.5-w8a8"
+    model: "Eco-Tech/MiniMax-M2.5-w8a8-QuaRot"
+    envs:
+      HCCL_BUFFSIZE: "512"
+      HCCL_OP_EXPANSION_MODE: "AIV"
+      PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
+      OMP_NUM_THREADS: "1"
+      TASK_QUEUE_ENABLE: "1"
+      VLLM-ASCEND_ENABLE_FUSED_MC2: "2"
+      VLLM_ASCEND_ENABLE_FLASHCOMM1: "1"
+      SERVER_PORT: "DEFAULT_PORT"
+    server_cmd:
+      - "--no-enable-prefix-caching"
+      - "--tensor-parallel-size"
+      - "4"
+      - "--data-parallel-size"
+      - "4"
+      - "--port"
+      - "$SERVER_PORT"
+      - "--enable-expert-parallel"
+      - "--max-num-seqs"
+      - "128"
+      - "--max-model-len"
+      - "153600"
+      - "--max-num-batched-tokens"
+      - "16384"
+      - "--trust-remote-code"
+      - "--gpu-memory-utilization"
+      - "0.8"
+      - "--quantization"
+      - "ascend"
+      - "--compilation-config"
+      - '{"cudagraph_mode":"FULL_DECODE_ONLY"}'
+      - "--additional-config"
+      - '{"enable_cpu_binding":true}'
+      - "--async-scheduling"
+    benchmarks:
+      acc:
+        case_type: accuracy
+        dataset_path: vllm-ascend/gsm8k-lite
+        request_conf: vllm_api_general_chat
+        dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_chat_prompt
+        max_out_len: 32768
+        batch_size: 32
+        baseline: 95
+        threshold: 5
+      perf:
+        case_type: performance
+        dataset_path: vllm-ascend/GSM8K-in3500-bs2800
+        request_conf: vllm_api_stream_chat
+        dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_str_perf
+        num_prompts: 80
+        max_out_len: 1500
+        batch_size: 20
+        request_rate: 0
+        baseline: 730.0832
+        threshold: 0.97

tests/e2e/nightly/single_node/models/configs/Qwen3.5-27B-w8a8-A3.yaml

@@ -0,0 +1,64 @@
+# ==========================================
+# ACTUAL TEST CASES
+# ==========================================
+
+test_cases:
+  - name: "Qwen3.5-27B-w8a8"
+    model: "Eco-Tech/Qwen3.5-27B-w8a8-mtp"
+    envs:
+      VLLM_USE_MODELSCOPE: "true"
+      PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
+      HCCL_BUFFSIZE: "512"
+      OMP_PROC_BIND: "false"
+      OMP_NUM_THREADS: "1"
+      TASK_QUEUE_ENABLE: "1"
+      SERVER_PORT: "DEFAULT_PORT"
+    server_cmd:
+      - "--tensor-parallel-size"
+      - "2"
+      - "--data-parallel-size"
+      - "1"
+      - "--port"
+      - "$SERVER_PORT"
+      - "--seed"
+      - "1024"
+      - "--max-num-seqs"
+      - "32"
+      - "--quantization"
+      - "ascend"
+      - "--max-model-len"
+      - "133000"
+      - "--max-num-batched-tokens"
+      - "8096"
+      - "--trust-remote-code"
+      - "--gpu-memory-utilization"
+      - "0.9"
+      - "--no-enable-prefix-caching"
+      - "--additional-config"
+      - '{"enable_cpu_binding":true}'
+      - "--speculative_config"
+      - '{"method": "qwen3_5_mtp", "num_speculative_tokens": 3,"enforce_eager": true}'
+      - "--compilation-config"
+      - '{"cudagraph_mode":"FULL_DECODE_ONLY"}'
+      - "--async-scheduling"
+    benchmarks:
+      acc:
+        case_type: accuracy
+        dataset_path: vllm-ascend/gsm8k-lite
+        request_conf: vllm_api_general_chat
+        dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_chat_prompt
+        max_out_len: 32768
+        batch_size: 32
+        baseline: 95
+        threshold: 5
+      perf:
+        case_type: performance
+        dataset_path: vllm-ascend/GSM8K-in3500-bs8000-qwen3
+        request_conf: vllm_api_stream_chat
+        dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_str_perf
+        num_prompts: 80
+        max_out_len: 1500
+        batch_size: 20
+        request_rate: 0
+        baseline: 664.6449
+        threshold: 0.97

tests/e2e/nightly/single_node/models/configs/Qwen3.5-397B-A17B-W8A8-mtp-A3.yaml

@@ -0,0 +1,67 @@
+# ==========================================
+# ACTUAL TEST CASES
+# ==========================================
+
+test_cases:
+  - name: "Qwen3.5-397B-A17B-w8a8-mtp"
+    model: "Eco-Tech/Qwen3.5-397B-A17B-w8a8-mtp"
+    envs:
+      VLLM_USE_MODELSCOPE: "true"
+      PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
+      HCCL_BUFFSIZE: "512"
+      OMP_PROC_BIND: "false"
+      OMP_NUM_THREADS: "1"
+      TASK_QUEUE_ENABLE: "1"
+      SERVER_PORT: "DEFAULT_PORT"
+      VLLM_ASCEND_ENABLE_FUSED_MC2: "1"
+      VLLM_ASCEND_ENABLE_FLASHCOMM1: "1"
+    server_cmd:
+      - "--tensor-parallel-size"
+      - "16"
+      - "--data-parallel-size"
+      - "1"
+      - "--enable-expert-parallel"
+      - "--port"
+      - "$SERVER_PORT"
+      - "--seed"
+      - "1024"
+      - "--max-num-seqs"
+      - "32"
+      - "--quantization"
+      - "ascend"
+      - "--max-model-len"
+      - "133000"
+      - "--max-num-batched-tokens"
+      - "8096"
+      - "--trust-remote-code"
+      - "--gpu-memory-utilization"
+      - "0.9"
+      - "--no-enable-prefix-caching"
+      - "--additional-config"
+      - '{"enable_cpu_binding":true,"multistream_overlap_shared_expert":true}'
+      - "--speculative_config"
+      - '{"method": "qwen3_5_mtp","num_speculative_tokens": 3,"enforce_eager": true}'
+      - "--compilation-config"
+      - '{"cudagraph_mode":"FULL_DECODE_ONLY"}'
+      - "--async-scheduling"
+    benchmarks:
+      acc:
+        case_type: accuracy
+        dataset_path: vllm-ascend/gsm8k-lite
+        request_conf: vllm_api_general_chat
+        dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_chat_prompt
+        max_out_len: 32768
+        batch_size: 32
+        baseline: 95
+        threshold: 5
+      perf:
+        case_type: performance
+        dataset_path: vllm-ascend/GSM8K-in3500-bs8000-qwen3
+        request_conf: vllm_api_stream_chat
+        dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_str_perf
+        num_prompts: 80
+        max_out_len: 1500
+        batch_size: 20
+        request_rate: 0
+        baseline: 768.6894
+        threshold: 0.97