From 086c093347118d0448df6ac7e80a7b4dd1cdc4f8 Mon Sep 17 00:00:00 2001
From: starmountain1997 <77533802+starmountain1997@users.noreply.github.com>
Date: Wed, 7 Jan 2026 10:02:02 +0800
Subject: [PATCH] [CI] Add DeepSeek-V3.2-W8A8 nightly ci test (#5371)

# What this PR does / why we need it?

Add DeepSeek-V3.2-W8A8 dual-node nightly CI test and update A3 nightly
test configuration:

1. Add DeepSeek-V3.2-W8A8 dual-node test:
tests/e2e/nightly/multi_node/config/DeepSeek-V3_2-W8A8-A3-dual-nodes.yaml
    - 2 nodes, 16 NPUs per node (32 NPUs total)
- Configuration: 2P+1D (data-parallel-size=4, tensor-parallel-size=8,
data-parallel-size-local=2)
    - Includes performance and accuracy benchmarks with GSM8K dataset
  2. Update A3 nightly workflow: .github/workflows/nightly_test_a3.yaml
- Added DeepSeek-V3.2-W8A8 dual-node test to the A3 nightly test matrix
    - Test name: multi-node-dpsk3.2-2node
3. Improve test scripts: Updated
.github/workflows/_e2e_nightly_multi_node.yaml and related scripts for
better multi-node testing support

test on A3 instances
  - Performance baseline: 1 (threshold: 0.97)
  - Accuracy baseline: 95% (threshold: 5%)
- Test dataset: GSM8K with 512 prompts for performance, gsm8k-lite for
accuracy
---------
Signed-off-by: guozr <guozr1997@hotmail.com>
Co-authored-by: guozr <guozr1997@hotmail.com>
---
 .github/workflows/nightly_test_a3.yaml        |  3 +
 .../DeepSeek-V3_2-W8A8-A3-dual-nodes.yaml     | 86 +++++++++++++++++++
 2 files changed, 89 insertions(+)
 create mode 100644 tests/e2e/nightly/multi_node/config/DeepSeek-V3_2-W8A8-A3-dual-nodes.yaml

diff --git a/.github/workflows/nightly_test_a3.yaml b/.github/workflows/nightly_test_a3.yaml
index ff583d32..8c4c5b4d 100644
--- a/.github/workflows/nightly_test_a3.yaml
+++ b/.github/workflows/nightly_test_a3.yaml
@@ -68,6 +68,9 @@ jobs:
           - name: multi-node-qwenw8a8-2node-eplb
             config_file_path: Qwen3-235B-W8A8-EPLB.yaml
             size: 2
+          - name: multi-node-dpsk3.2-2node
+            config_file_path: DeepSeek-V3_2-W8A8-A3-dual-nodes.yaml
+            size: 2
           - name: multi-node-deepseek-r1-w8a8-longseq
             config_file_path: DeepSeek-R1-W8A8-longseq.yaml
             size: 2
diff --git a/tests/e2e/nightly/multi_node/config/DeepSeek-V3_2-W8A8-A3-dual-nodes.yaml b/tests/e2e/nightly/multi_node/config/DeepSeek-V3_2-W8A8-A3-dual-nodes.yaml
new file mode 100644
index 00000000..0fc4fac2
--- /dev/null
+++ b/tests/e2e/nightly/multi_node/config/DeepSeek-V3_2-W8A8-A3-dual-nodes.yaml
@@ -0,0 +1,86 @@
+test_name: "test DeepSeek-V3.2-W8A8 on A3"
+model: "vllm-ascend/DeepSeek-V3.2-W8A8"
+num_nodes: 2
+npu_per_node: 16
+env_common:
+  HCCL_OP_EXPANSION_MODE: "AIV"
+
+  VLLM_USE_MODELSCOPE: true
+  HCCL_BUFFSIZE: 1024
+  SERVER_PORT: 8080
+  OMP_PROC_BIND: false
+  OMP_NUM_THREADS: 1
+  VLLM_ASCEND_ENABLE_MLAPO: 1
+  PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
+  VLLM_ASCEND_ENABLE_FLASHCOMM1: 1
+  ASCEND_A3_EBA_ENABLE: 1
+
+
+deployment:
+  -
+    server_cmd: >
+      vllm serve vllm-ascend/DeepSeek-V3.2-W8A8
+      --host 0.0.0.0
+      --port $SERVER_PORT
+      --data-parallel-size 4
+      --data-parallel-size-local 2
+      --data-parallel-address $LOCAL_IP
+      --data-parallel-rpc-port 13399
+      --tensor-parallel-size 8
+      --quantization ascend
+      --seed 1024
+      --enable-expert-parallel
+      --max-num-seqs 16
+      --max-model-len 8192
+      --max-num-batched-tokens 4096
+      --no-enable-prefix-caching
+      --gpu-memory-utilization 0.85
+      --trust-remote-code
+      --speculative-config '{"num_speculative_tokens": 2, "method":"deepseek_mtp"}'
+      --tokenizer-mode deepseek_v32
+      --reasoning-parser deepseek_v3
+      --api-server-count 4
+
+  -
+    server_cmd: >
+      vllm serve vllm-ascend/DeepSeek-V3.2-W8A8
+      --headless
+      --data-parallel-size 4
+      --data-parallel-rpc-port 13399
+      --data-parallel-size-local 2
+      --data-parallel-start-rank 2
+      --data-parallel-address $MASTER_IP
+      --tensor-parallel-size 8
+      --quantization ascend
+      --seed 1024
+      --enable-expert-parallel
+      --max-num-seqs 16
+      --max-model-len 8192
+      --max-num-batched-tokens 4096
+      --no-enable-prefix-caching
+      --gpu-memory-utilization 0.85
+      --trust-remote-code
+      --speculative-config '{"num_speculative_tokens": 2, "method":"deepseek_mtp"}'
+      --tokenizer-mode deepseek_v32
+      --reasoning-parser deepseek_v3
+benchmarks:
+  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: 512
+    max_out_len: 3000
+    batch_size: 512
+    request_rate: 11.2
+    baseline: 594.915
+    threshold: 0.97
+  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: 4096
+    batch_size: 64
+    baseline: 95
+    threshold: 5