[AMD CI] re-enable testcases missed when migrating ci test files (#16535)

Co-authored-by: michael-amd <michael.zhang@amd.com>
Co-authored-by: yctseng0211 <yctseng@amd.com>

.github/workflows/pr-test-amd.yml

@@ -200,7 +200,7 @@ jobs:
       fail-fast: false
       matrix:
         runner: [linux-mi325-gpu-1]
-        part: [0, 1, 2, 3]
+        part: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11]
     runs-on: ${{matrix.runner}}
     steps:
       - name: Checkout code
@@ -222,7 +222,7 @@ jobs:
       - name: Run test
         timeout-minutes: 30
         run: |
-          bash scripts/ci/amd_ci_exec.sh -w "/sglang-checkout/test" python3 run_suite.py --hw amd --suite stage-b-test-small-1-gpu --auto-partition-id ${{ matrix.part }} --auto-partition-size 4
+          bash scripts/ci/amd_ci_exec.sh -w "/sglang-checkout/test" python3 run_suite.py --hw amd --suite stage-b-test-small-1-gpu --auto-partition-id ${{ matrix.part }} --auto-partition-size 12
 
   stage-b-test-large-2-gpu-amd:
     needs: [check-changes, stage-a-test-1-amd]
@@ -519,7 +519,7 @@ jobs:
       fail-fast: false
       matrix:
         runner: [linux-mi325-gpu-1]
-        part: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11]
+        part: [0, 1]
     runs-on: ${{matrix.runner}}
     steps:
       - name: Checkout code
@@ -541,7 +541,7 @@ jobs:
       - name: Run test
         timeout-minutes: 30
         run: |
-          bash scripts/ci/amd_ci_exec.sh python3 run_suite.py --suite per-commit-amd --auto-partition-id ${{ matrix.part }} --auto-partition-size 12
+          bash scripts/ci/amd_ci_exec.sh python3 run_suite.py --suite per-commit-amd --auto-partition-id ${{ matrix.part }} --auto-partition-size 2
 
   unit-test-backend-1-gpu-amd-mi35x:
     needs: [check-changes, stage-a-test-1-amd]
@@ -695,9 +695,9 @@ jobs:
         run: bash scripts/ci/amd_ci_install_dependency.sh
 
       - name: Run test
-        timeout-minutes: 30
+        timeout-minutes: 60
         run: |
-          bash scripts/ci/amd_ci_exec.sh python3 run_suite.py --suite per-commit-8-gpu-amd-mi35x --timeout-per-file 1800
+          bash scripts/ci/amd_ci_exec.sh python3 run_suite.py --suite per-commit-8-gpu-amd-mi35x --timeout-per-file 3600
 
   performance-test-1-gpu-part-1-amd:
     needs: [check-changes, stage-a-test-1-amd]

test/registered/attention/test_create_kvindices.py

@@ -4,11 +4,12 @@ import numpy as np
 import torch
 
 from sglang.srt.layers.attention.utils import create_flashinfer_kv_indices_triton
-from sglang.test.ci.ci_register import register_cuda_ci
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import CustomTestCase
 
 # Triton kernel unit test for KV indices creation
 register_cuda_ci(est_time=10, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=10, suite="stage-b-test-small-1-gpu")
 
 
 class TestCreateKvIndices(CustomTestCase):

test/registered/attention/test_radix_attention.py

@@ -1,7 +1,7 @@
 import unittest
 
 from sglang.srt.environ import envs
-from sglang.test.ci.ci_register import register_cuda_ci
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.kits.radix_cache_server_kit import run_radix_attention_test
 from sglang.test.test_utils import (
     DEFAULT_SMALL_MODEL_NAME_FOR_TEST,
@@ -15,6 +15,7 @@ from sglang.test.test_utils import (
 
 # RadixAttention server integration tests
 register_cuda_ci(est_time=100, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=100, suite="stage-b-test-small-1-gpu")
 
 
 class TestRadixCacheFCFS(CustomTestCase):

test/registered/attention/test_torch_native_attention_backend.py

@@ -7,7 +7,7 @@ import unittest
 from types import SimpleNamespace
 
 from sglang.srt.utils import kill_process_tree
-from sglang.test.ci.ci_register import register_cuda_ci
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.run_eval import run_eval
 from sglang.test.test_utils import (
     DEFAULT_MODEL_NAME_FOR_TEST,
@@ -19,6 +19,7 @@ from sglang.test.test_utils import (
 
 # Torch native attention backend integration test with MMLU eval
 register_cuda_ci(est_time=150, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=150, suite="stage-b-test-small-1-gpu")
 
 
 class TestTorchNativeAttnBackend(CustomTestCase):

test/registered/attention/test_triton_attention_backend.py

@@ -7,7 +7,7 @@ import unittest
 from types import SimpleNamespace
 
 from sglang.srt.utils import kill_process_tree
-from sglang.test.ci.ci_register import register_cuda_ci
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.run_eval import run_eval
 from sglang.test.test_utils import (
     DEFAULT_MODEL_NAME_FOR_TEST,
@@ -21,6 +21,7 @@ from sglang.test.test_utils import (
 
 # Triton attention backend integration test with latency benchmark and MMLU eval
 register_cuda_ci(est_time=200, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=1110, suite="stage-b-test-small-1-gpu")
 
 
 class TestTritonAttnBackend(CustomTestCase):

test/registered/attention/test_triton_attention_kernels.py

@@ -19,11 +19,16 @@ from sglang.srt.layers.attention.triton_ops.prefill_attention import (
     context_attention_fwd,
 )
 from sglang.srt.utils import get_device
-from sglang.test.ci.ci_register import register_cuda_ci
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import CustomTestCase
 
 # Triton attention kernel unit tests (decode, extend, prefill)
 register_cuda_ci(est_time=30, suite="stage-b-test-small-1-gpu")
+register_amd_ci(
+    est_time=30,
+    suite="stage-b-test-small-1-gpu",
+    disabled="test was never enabled for AMD CI, needs validation",
+)
 
 
 def extend_attention_fwd_torch(

test/registered/attention/test_triton_sliding_window.py

@@ -4,7 +4,7 @@ from types import SimpleNamespace
 import requests
 
 from sglang.srt.utils import kill_process_tree
-from sglang.test.ci.ci_register import register_cuda_ci
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.run_eval import run_eval
 from sglang.test.test_utils import (
     DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
@@ -16,6 +16,7 @@ from sglang.test.test_utils import (
 
 # Sliding window attention with Triton backend (Gemma-3 model)
 register_cuda_ci(est_time=100, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=100, suite="stage-b-test-small-1-gpu")
 
 
 class TestSlidingWindowAttentionTriton(CustomTestCase):

test/registered/backends/test_torch_compile.py

@@ -17,7 +17,7 @@ from sglang.test.test_utils import (
 )
 
 register_cuda_ci(est_time=190, suite="stage-b-test-small-1-gpu")
-register_amd_ci(est_time=993, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=1100, suite="stage-b-test-small-1-gpu")
 
 
 class TestTorchCompile(CustomTestCase):

test/registered/lora/test_lora.py

@@ -16,7 +16,7 @@ import multiprocessing as mp
 import os
 import unittest
 
-from sglang.test.ci.ci_register import register_cuda_ci
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.lora_utils import (
     ALL_OTHER_MULTI_LORA_MODELS,
     CI_MULTI_LORA_MODELS,
@@ -25,6 +25,7 @@ from sglang.test.lora_utils import (
 from sglang.test.test_utils import CustomTestCase, is_in_ci
 
 register_cuda_ci(est_time=82, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=82, suite="stage-b-test-small-1-gpu")
 
 
 class TestLoRA(CustomTestCase):

test/registered/lora/test_lora_backend.py

@@ -17,7 +17,7 @@ import os
 import unittest
 from typing import List
 
-from sglang.test.ci.ci_register import register_cuda_ci
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.lora_utils import (
     ALL_OTHER_LORA_MODELS,
     BACKENDS,
@@ -30,6 +30,11 @@ from sglang.test.lora_utils import (
 from sglang.test.test_utils import CustomTestCase, is_in_ci
 
 register_cuda_ci(est_time=200, suite="stage-b-test-small-1-gpu")
+register_amd_ci(
+    est_time=200,
+    suite="stage-b-test-small-1-gpu",
+    disabled="see https://github.com/sgl-project/sglang/issues/13107",
+)
 
 
 class TestLoRABackend(CustomTestCase):

test/registered/lora/test_lora_eviction.py

@@ -19,11 +19,12 @@ from typing import Dict, List, Tuple
 
 import torch
 
-from sglang.test.ci.ci_register import register_cuda_ci
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.runners import SRTRunner
 from sglang.test.test_utils import CustomTestCase
 
 register_cuda_ci(est_time=224, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=224, suite="stage-b-test-small-1-gpu")
 
 PROMPTS = [
     "AI is a field of computer science focused on",

test/registered/lora/test_lora_tp.py

@@ -17,7 +17,7 @@ import os
 import unittest
 from typing import List
 
-from sglang.test.ci.ci_register import register_cuda_ci
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.lora_utils import (
     ALL_OTHER_LORA_MODELS,
     CI_LORA_MODELS,
@@ -29,6 +29,11 @@ from sglang.test.lora_utils import (
 from sglang.test.test_utils import CustomTestCase, is_in_ci
 
 register_cuda_ci(est_time=116, suite="stage-b-test-large-2-gpu")
+register_amd_ci(
+    est_time=116,
+    suite="stage-b-test-large-2-gpu-amd",
+    disabled="see https://github.com/sgl-project/sglang/issues/13107",
+)
 
 
 class TestLoRATP(CustomTestCase):

test/registered/lora/test_multi_lora_backend.py

@@ -16,7 +16,7 @@ import multiprocessing as mp
 import os
 import unittest
 
-from sglang.test.ci.ci_register import register_cuda_ci
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.lora_utils import (
     ALL_OTHER_MULTI_LORA_MODELS,
     CI_MULTI_LORA_MODELS,
@@ -25,6 +25,7 @@ from sglang.test.lora_utils import (
 from sglang.test.test_utils import CustomTestCase, is_in_ci
 
 register_cuda_ci(est_time=60, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=60, suite="stage-b-test-small-1-gpu")
 
 # All prompts are used at once in a batch.
 PROMPTS = [

test/registered/models/test_qwen_models.py

@@ -2,7 +2,7 @@ from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 
 # Qwen model tests
 register_cuda_ci(est_time=90, suite="stage-b-test-small-1-gpu")
-register_amd_ci(est_time=82, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=130, suite="stage-b-test-small-1-gpu")
 
 import unittest
 from types import SimpleNamespace

test/registered/models/test_vlm_models.py

@@ -2,7 +2,7 @@ from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 
 # VLM (Vision Language Model) tests
 register_cuda_ci(est_time=270, suite="stage-b-test-small-1-gpu")
-register_amd_ci(est_time=387, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=420, suite="stage-b-test-small-1-gpu")
 
 import argparse
 import random

test/registered/openai_server/features/test_enable_thinking.py

@@ -22,7 +22,7 @@ from sglang.test.test_utils import (
 )
 
 register_cuda_ci(est_time=70, suite="stage-b-test-small-1-gpu")
-register_amd_ci(est_time=70, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=200, suite="stage-b-test-small-1-gpu")
 
 
 class TestEnableThinking(CustomTestCase):

test/registered/openai_server/features/test_json_mode.py

@@ -14,7 +14,7 @@ from sglang.test.test_utils import (
 )
 
 register_cuda_ci(est_time=109, suite="stage-b-test-small-1-gpu")
-register_amd_ci(est_time=120, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=180, suite="stage-b-test-small-1-gpu")
 
 
 class TestJSONModeMixin:

test/registered/sampling/test_penalty.py

@@ -10,7 +10,7 @@ from sglang.srt.utils import kill_process_tree
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 
 register_cuda_ci(est_time=82, suite="stage-b-test-small-1-gpu")
-register_amd_ci(est_time=180, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=82, suite="stage-b-test-small-1-gpu")
 from sglang.test.test_utils import (
     DEFAULT_SMALL_MODEL_NAME_FOR_TEST,
     DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,

test/srt/run_suite.py

@@ -113,7 +113,7 @@ suite_amd = {
         TestFile("test_deepseek_v3_mtp.py", 275),
     ],
     "per-commit-8-gpu-amd-mi35x": [
-        TestFile("test_deepseek_r1_mxfp4_8gpu.py", 1800),
+        TestFile("test_deepseek_r1_mxfp4_8gpu.py", 3600),
     ],
     "nightly-amd": [
         TestFile("nightly/test_gsm8k_eval_amd.py"),