[AMD CI] Migrate and Add More Testcases (#17116)

Co-authored-by: yctseng0211 <yctseng@amd.com>
2026-01-20 00:07:39 +08:00
parent 20b0523eca
commit 2ea02f0642
19 changed files with 311 additions and 67 deletions
--- a/.github/workflows/pr-test-amd.yml
+++ b/.github/workflows/pr-test-amd.yml
@@ -149,7 +149,10 @@ jobs:
          docker exec -w /sglang-checkout/sgl-kernel/tests ci_sglang python3 -m pytest test_topk.py
          docker exec -w /sglang-checkout/sgl-kernel/tests ci_sglang python3 -m pytest test_kvcacheio.py
          docker exec -w /sglang-checkout/sgl-kernel/tests/sgl_diffusion ci_sglang python3 -m pytest test_timestep_embedding.py
-
+          docker exec -w /sglang-checkout/sgl-kernel/tests ci_sglang python3 -m pytest test_moe_topk_sigmoid.py
+          docker exec -w /sglang-checkout/sgl-kernel/tests ci_sglang python3 -m pytest test_torch_defaults_reset.py
+          docker exec -w /sglang-checkout/sgl-kernel/tests ci_sglang python3 -m pytest test_amd_deterministic_custom_allreduce.py
+          docker exec -w /sglang-checkout/sgl-kernel/tests ci_sglang python3 -m pytest test_amd_nccl_allreduce_determinism.py
  # =============================================== primary ====================================================

  stage-a-test-1-amd:
@@ -190,7 +193,7 @@ jobs:
      - name: Run test
        timeout-minutes: 10
        run: |
-          bash scripts/ci/amd_ci_exec.sh -w "/sglang-checkout/test" python3 run_suite.py --hw amd --suite stage-a-test-1
+          bash scripts/ci/amd_ci_exec.sh -w "/sglang-checkout/test" python3 run_suite.py --hw amd --suite stage-a-test-1-amd

  stage-b-test-small-1-gpu-amd:
    needs: [check-changes, stage-a-test-1-amd]
@@ -548,52 +551,13 @@ jobs:
          echo "=== Post-test System Memory Status ==="
          free -h

-  unit-test-backend-1-gpu-amd:
-    needs: [check-changes, stage-a-test-1-amd]
+
+  stage-c-test-large-8-gpu-amd:
+    needs: [check-changes, call-gate, stage-b-test-small-1-gpu-amd, stage-b-test-large-2-gpu-amd]
    if: |
      always() &&
      (
-        (inputs.target_stage == 'unit-test-backend-1-gpu-amd') ||
-        (
-          !inputs.target_stage &&
-          (!failure() && !cancelled()) &&
-          ((needs.check-changes.outputs.main_package == 'true') || (needs.check-changes.outputs.sgl_kernel == 'true'))
-        )
-      )
-    strategy:
-      fail-fast: false
-      matrix:
-        runner: [linux-mi325-gpu-1]
-        part: [0, 1]
-    runs-on: ${{matrix.runner}}
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-        with:
-          ref: ${{ inputs.pr_head_sha || inputs.ref || github.sha }}
-
-      - name: Ensure VRAM is clear
-        run: bash scripts/ensure_vram_clear.sh rocm
-
-      - name: Start CI container
-        run: bash scripts/ci/amd_ci_start_container.sh
-        env:
-          GITHUB_WORKSPACE: ${{ github.workspace }}
-
-      - name: Install dependencies
-        run: bash scripts/ci/amd_ci_install_dependency.sh
-
-      - 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 2
-
-  unit-test-backend-8-gpu-amd:
-    needs: [check-changes, stage-a-test-1-amd]
-    if: |
-      always() &&
-      (
-        (inputs.target_stage == 'unit-test-backend-8-gpu-amd') ||
+        (inputs.target_stage == 'stage-c-test-large-8-gpu-amd') ||
        (
          !inputs.target_stage &&
          (!failure() && !cancelled()) &&
@@ -634,7 +598,7 @@ jobs:
      - name: Run test
        timeout-minutes: 60
        run: |
-          bash scripts/ci/amd_ci_exec.sh python3 run_suite.py --suite per-commit-8-gpu-amd --auto-partition-id ${{ matrix.part }} --auto-partition-size 2 --timeout-per-file 3600
+          bash scripts/ci/amd_ci_exec.sh -w "/sglang-checkout/test" python3 run_suite.py --hw amd --suite stage-c-test-large-8-gpu-amd --auto-partition-id ${{ matrix.part }} --auto-partition-size 2 --timeout-per-file 3600

  stage-c-test-large-8-gpu-amd-mi35x:
    needs: [check-changes, call-gate, stage-b-test-small-1-gpu-amd, stage-b-test-large-2-gpu-amd]
@@ -721,6 +685,12 @@ jobs:
        run: |
          bash scripts/ci/amd_ci_exec.sh -w /sglang-checkout/test/registered/perf python3 -m unittest test_bench_serving_1gpu_part1.TestBenchServing1GPUPart1.test_online_latency_default

+      - name: Benchmark online latency (LoRA)
+        timeout-minutes: 10
+        run: |
+          bash scripts/ci/amd_ci_exec.sh python3 -m unittest test_bench_serving.TestBenchServing.test_lora_online_latency
+          bash scripts/ci/amd_ci_exec.sh python3 -m unittest test_bench_serving.TestBenchServing.test_lora_online_latency_with_concurrent_adapter_updates
+
      - name: Benchmark offline throughput
        timeout-minutes: 15
        run: |
@@ -780,6 +750,70 @@ jobs:
        run: |
          bash scripts/ci/amd_ci_exec.sh -w /sglang-checkout/test/registered/perf python3 -m unittest test_bench_serving_1gpu_large.TestBenchServing1GPULarge.test_offline_throughput_default_fp8

+      - name: Benchmark VLM offline throughput
+        timeout-minutes: 10
+        run: |
+          bash scripts/ci/amd_ci_exec.sh python3 -m unittest test_bench_serving.TestBenchServing.test_vlm_offline_throughput
+
+      - name: Benchmark VLM online latency
+        timeout-minutes: 10
+        run: |
+          bash scripts/ci/amd_ci_exec.sh python3 -m unittest test_bench_serving.TestBenchServing.test_vlm_online_latency
+
+  performance-test-1-gpu-part-3-amd:
+    needs: [check-changes, stage-a-test-1-amd]
+    if: |
+      always() &&
+      (
+        (inputs.target_stage == 'performance-test-1-gpu-part-3-amd') ||
+        (
+          !inputs.target_stage &&
+          (!failure() && !cancelled()) &&
+          ((needs.check-changes.outputs.main_package == 'true') || (needs.check-changes.outputs.sgl_kernel == 'true'))
+        )
+      )
+    strategy:
+      fail-fast: false
+      matrix:
+        runner: [linux-mi325-gpu-1]
+    runs-on: ${{matrix.runner}}
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+        with:
+          ref: ${{ inputs.pr_head_sha || inputs.ref || github.sha }}
+
+      - name: Ensure VRAM is clear
+        run: bash scripts/ensure_vram_clear.sh rocm
+
+      - name: Start CI container
+        run: bash scripts/ci/amd_ci_start_container.sh
+        env:
+          GITHUB_WORKSPACE: ${{ github.workspace }}
+
+      - name: Install dependencies
+        run: bash scripts/ci/amd_ci_install_dependency.sh
+
+      - name: Benchmark Scores online latency and throughput
+        timeout-minutes: 10
+        run: |
+          bash scripts/ci/amd_ci_exec.sh python3 -m unittest test_bench_serving.TestBenchServing.test_score_api_latency_throughput
+
+      - name: Benchmark Scores online latency and throughput (batch size scaling)
+        timeout-minutes: 10
+        run: |
+          bash scripts/ci/amd_ci_exec.sh python3 -m unittest test_bench_serving.TestBenchServing.test_score_api_batch_scaling
+
+      - name: Benchmark Embeddings online latency and throughput
+        timeout-minutes: 10
+        run: |
+          bash scripts/ci/amd_ci_exec.sh python3 -m unittest test_bench_serving.TestBenchServing.test_embeddings_api_latency_throughput
+
+      - name: Benchmark Embeddings online latency and throughput (batch size scaling)
+        timeout-minutes: 10
+        run: |
+          bash scripts/ci/amd_ci_exec.sh python3 -m unittest test_bench_serving.TestBenchServing.test_embeddings_api_batch_scaling
+
  performance-test-2-gpu-amd:
    needs: [check-changes, stage-a-test-1-amd]
    if: |
@@ -942,11 +976,11 @@ jobs:
        stage-b-test-small-1-gpu-amd,
        stage-b-test-small-1-gpu-amd-mi35x,
        stage-b-test-large-2-gpu-amd,
-        unit-test-backend-1-gpu-amd,
-        unit-test-backend-8-gpu-amd,
+        stage-c-test-large-8-gpu-amd,
        stage-c-test-large-8-gpu-amd-mi35x,
        performance-test-1-gpu-part-1-amd,
        performance-test-1-gpu-part-2-amd,
+        performance-test-1-gpu-part-3-amd,
        performance-test-2-gpu-amd,
        accuracy-test-1-gpu-amd,
        accuracy-test-2-gpu-amd,
--- a/test/registered/amd/test_deepseek_r1_mxfp4_8gpu.py
+++ b/test/registered/amd/test_deepseek_r1_mxfp4_8gpu.py
@@ -19,7 +19,7 @@ from sglang.test.test_utils import (
 register_amd_ci(est_time=3600, suite="stage-c-test-large-8-gpu-amd-mi35x")

 DEEPSEEK_R1_MODEL_PATH = "amd/DeepSeek-R1-MXFP4-Preview"
-SERVER_LAUNCH_TIMEOUT = 1200
+SERVER_LAUNCH_TIMEOUT = 1800


 class TestDeepseekR1MXFP4(CustomTestCase):
--- a/test/registered/amd/test_deepseek_v3_basic.py
+++ b/test/registered/amd/test_deepseek_v3_basic.py
@@ -0,0 +1,84 @@
+import unittest
+from types import SimpleNamespace
+
+from sglang.srt.utils import kill_process_tree
+from sglang.test.ci.ci_register import register_amd_ci
+from sglang.test.few_shot_gsm8k import run_eval as run_eval_few_shot_gsm8k
+from sglang.test.send_one import BenchArgs, send_one_prompt
+from sglang.test.test_utils import (
+    DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
+    DEFAULT_URL_FOR_TEST,
+    CustomTestCase,
+    is_in_amd_ci,
+    is_in_ci,
+    popen_launch_server,
+    write_github_step_summary,
+)
+
+register_amd_ci(est_time=952, suite="stage-c-test-large-8-gpu-amd")
+
+FULL_DEEPSEEK_V3_MODEL_PATH = "deepseek-ai/DeepSeek-V3-0324"
+
+
+class TestDeepseekV3Basic(CustomTestCase):
+    @classmethod
+    def setUpClass(cls):
+        cls.model = FULL_DEEPSEEK_V3_MODEL_PATH
+        cls.base_url = DEFAULT_URL_FOR_TEST
+        other_args = [
+            "--trust-remote-code",
+            "--tp",
+            "8",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true, "num_threads": 64}',
+        ]
+        cls.process = popen_launch_server(
+            cls.model,
+            cls.base_url,
+            timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH * 5,
+            other_args=other_args,
+        )
+
+    @classmethod
+    def tearDownClass(cls):
+        kill_process_tree(cls.process.pid)
+
+    def test_a_gsm8k(
+        self,
+    ):  # Append an "a" to make this test run first (alphabetically) to warm up the server
+        args = SimpleNamespace(
+            num_shots=8,
+            data_path=None,
+            num_questions=1400,
+            parallel=1400,
+            max_new_tokens=512,
+            host="http://127.0.0.1",
+            port=int(self.base_url.split(":")[-1]),
+        )
+        metrics = run_eval_few_shot_gsm8k(args)
+        print(f"{metrics=}")
+
+        if is_in_ci():
+            write_github_step_summary(
+                f"### test_gsm8k (deepseek-v3)\n" f'{metrics["accuracy"]=:.3f}\n'
+            )
+            self.assertGreater(metrics["accuracy"], 0.935)
+
+    def test_bs_1_speed(self):
+        args = BenchArgs(port=int(self.base_url.split(":")[-1]), max_new_tokens=2048)
+        acc_length, speed = send_one_prompt(args)
+
+        print(f"{speed=:.2f}")
+
+        if is_in_ci():
+            write_github_step_summary(
+                f"### test_bs_1_speed (deepseek-v3)\n" f"{speed=:.2f} token/s\n"
+            )
+            if is_in_amd_ci():
+                self.assertGreater(speed, 12)
+            else:
+                self.assertGreater(speed, 75)
+
+
+if __name__ == "__main__":
+    unittest.main()
--- a/test/registered/amd/test_deepseek_v3_mtp.py
+++ b/test/registered/amd/test_deepseek_v3_mtp.py
@@ -0,0 +1,116 @@
+import unittest
+from types import SimpleNamespace
+
+import requests
+
+from sglang.srt.utils import kill_process_tree
+from sglang.test.ci.ci_register import register_amd_ci
+from sglang.test.few_shot_gsm8k import run_eval as run_eval_few_shot_gsm8k
+from sglang.test.send_one import BenchArgs, send_one_prompt
+from sglang.test.test_utils import (
+    DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
+    DEFAULT_URL_FOR_TEST,
+    CustomTestCase,
+    is_in_amd_ci,
+    is_in_ci,
+    popen_launch_server,
+    write_github_step_summary,
+)
+
+register_amd_ci(est_time=980, suite="stage-c-test-large-8-gpu-amd")
+
+FULL_DEEPSEEK_V3_MODEL_PATH = "deepseek-ai/DeepSeek-V3-0324"
+
+
+class TestDeepseekV3MTP(CustomTestCase):
+    @classmethod
+    def setUpClass(cls):
+        cls.model = FULL_DEEPSEEK_V3_MODEL_PATH
+        cls.base_url = DEFAULT_URL_FOR_TEST
+        other_args = [
+            "--tp",
+            "8",
+            "--trust-remote-code",
+            "--speculative-algorithm",
+            "EAGLE",
+            "--speculative-num-steps",
+            "3",
+            "--speculative-eagle-topk",
+            "1",
+            "--speculative-num-draft-tokens",
+            "4",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true, "num_threads": 64}',
+        ]
+        if not is_in_amd_ci():
+            other_args += ["--mem-frac", "0.7"]
+        cls.process = popen_launch_server(
+            cls.model,
+            cls.base_url,
+            timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH * 5,
+            other_args=other_args,
+        )
+
+    @classmethod
+    def tearDownClass(cls):
+        kill_process_tree(cls.process.pid)
+
+    def test_a_gsm8k(
+        self,
+    ):  # Append an "a" to make this test run first (alphabetically) to warm up the server
+        requests.get(self.base_url + "/flush_cache")
+
+        args = SimpleNamespace(
+            num_shots=5,
+            data_path=None,
+            num_questions=200,
+            max_new_tokens=512,
+            parallel=128,
+            host="http://127.0.0.1",
+            port=int(self.base_url.split(":")[-1]),
+        )
+        metrics = run_eval_few_shot_gsm8k(args)
+        print(f"{metrics=}")
+
+        server_info = requests.get(self.base_url + "/get_server_info")
+        avg_spec_accept_length = server_info.json()["internal_states"][0][
+            "avg_spec_accept_length"
+        ]
+        print(f"{avg_spec_accept_length=}")
+
+        if is_in_ci():
+            write_github_step_summary(
+                f"### test_gsm8k (deepseek-v3 mtp)\n"
+                f'{metrics["accuracy"]=:.3f}\n'
+                f"{avg_spec_accept_length=:.2f}\n"
+            )
+            self.assertGreater(metrics["accuracy"], 0.935)
+            if is_in_amd_ci():
+                self.assertGreater(avg_spec_accept_length, 2.8)
+            else:
+                self.assertGreater(avg_spec_accept_length, 2.9)
+
+    def test_bs_1_speed(self):
+        args = BenchArgs(port=int(self.base_url.split(":")[-1]), max_new_tokens=2048)
+        acc_length, speed = send_one_prompt(args)
+
+        print(f"{acc_length=:.2f} {speed=:.2f}")
+
+        if is_in_ci():
+            write_github_step_summary(
+                f"### test_bs_1_speed (deepseek-v3 mtp)\n"
+                f"{acc_length=:.2f}\n"
+                f"{speed=:.2f} token/s\n"
+            )
+            if is_in_amd_ci():
+                self.assertGreater(acc_length, 2.8)
+            else:
+                self.assertGreater(acc_length, 2.9)
+            if is_in_amd_ci():
+                self.assertGreater(speed, 15)
+            else:
+                self.assertGreater(speed, 130)
+
+
+if __name__ == "__main__":
+    unittest.main()
--- a/test/registered/attention/test_wave_attention_kernels.py
+++ b/test/registered/attention/test_wave_attention_kernels.py
@@ -24,7 +24,7 @@ from sglang.srt.layers.attention.wave_ops.prefill_attention import (
 from sglang.test.ci.ci_register import register_amd_ci

 # Wave attention kernel unit tests (AMD only - requires wave_lang)
-register_amd_ci(est_time=60, suite="stage-a-test-1")
+register_amd_ci(est_time=60, suite="stage-a-test-1-amd")


 class TestWaveAttention(unittest.TestCase):
--- a/test/registered/core/test_deterministic.py
+++ b/test/registered/core/test_deterministic.py
@@ -9,15 +9,18 @@ test into unit tests so that's easily reproducible in CI.

 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.test_deterministic_utils import (
    COMMON_SERVER_ARGS,
    TestDeterministicBase,
 )
+from sglang.test.test_utils import is_in_amd_ci

 register_cuda_ci(est_time=278, suite="stage-b-test-large-1-gpu")
+register_amd_ci(est_time=278, suite="stage-b-test-small-1-gpu-amd")


+@unittest.skipIf(is_in_amd_ci(), "Skip for AMD CI.")
 class TestFlashinferDeterministic(TestDeterministicBase):
    # Test with flashinfer attention backend
    @classmethod
@@ -32,6 +35,7 @@ class TestFlashinferDeterministic(TestDeterministicBase):
        return args


+@unittest.skipIf(is_in_amd_ci(), "Skip for AMD CI.")
 class TestFa3Deterministic(TestDeterministicBase):
    # Test with fa3 attention backend
    @classmethod
--- a/test/registered/hicache/test_hicache_storage_3fs_backend.py
+++ b/test/registered/hicache/test_hicache_storage_3fs_backend.py
@@ -10,10 +10,11 @@ import unittest

 from test_hicache_storage_file_backend import HiCacheStorageBaseMixin

-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

 register_cuda_ci(est_time=200, suite="stage-b-test-large-2-gpu")
+register_amd_ci(est_time=300, suite="stage-b-test-large-2-gpu")


 class HiCacheStorage3FSBackendBaseMixin(HiCacheStorageBaseMixin):
--- a/test/registered/hicache/test_hicache_storage_file_backend.py
+++ b/test/registered/hicache/test_hicache_storage_file_backend.py
@@ -18,7 +18,7 @@ import requests

 from sglang.bench_serving import get_tokenizer
 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.few_shot_gsm8k import run_eval as run_eval_few_shot_gsm8k
 from sglang.test.test_utils import (
    DEFAULT_MLA_MODEL_NAME_FOR_TEST,
@@ -31,6 +31,7 @@ from sglang.test.test_utils import (
 )

 register_cuda_ci(est_time=200, suite="stage-b-test-large-2-gpu")
+register_amd_ci(est_time=526, suite="stage-b-test-large-2-gpu-amd")


 class HiCacheStorageBaseMixin:
--- a/test/registered/openai_server/basic/test_serving_rerank.py
+++ b/test/registered/openai_server/basic/test_serving_rerank.py
@@ -7,7 +7,7 @@ from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 # Keep consistent with other openai_server/basic unit tests.
 register_cuda_ci(est_time=10, suite="stage-b-test-large-1-gpu")
-register_amd_ci(est_time=10, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=10, suite="stage-b-test-small-1-gpu-amd")

 try:
    from sglang.srt.entrypoints.openai.serving_rerank import (
--- a/test/registered/quant/test_awq_dequant.py
+++ b/test/registered/quant/test_awq_dequant.py
@@ -19,7 +19,7 @@ from sglang.srt.layers.quantization.awq_triton import (
 from sglang.test.ci.ci_register import register_amd_ci
 from sglang.test.test_utils import CustomTestCase

-register_amd_ci(est_time=2, suite="stage-a-test-1")
+register_amd_ci(est_time=2, suite="stage-a-test-1-amd")

 device = "cuda"

--- a/test/registered/quant/test_fused_rms_fp8_group_quant.py
+++ b/test/registered/quant/test_fused_rms_fp8_group_quant.py
@@ -7,7 +7,7 @@ import torch.nn.functional as F
 from sglang.test.ci.ci_register import register_amd_ci
 from sglang.test.test_utils import CustomTestCase

-register_amd_ci(est_time=10, suite="stage-a-test-1")
+register_amd_ci(est_time=10, suite="stage-a-test-1-amd")


 def _fp8_available() -> bool:
--- a/test/registered/quant/test_int4fp8_moe.py
+++ b/test/registered/quant/test_int4fp8_moe.py
@@ -1,6 +1,7 @@
 from types import SimpleNamespace

 from sglang.srt.utils import kill_process_tree
+from sglang.test.ci.ci_register import register_amd_ci
 from sglang.test.few_shot_gsm8k import run_eval
 from sglang.test.test_utils import (
    DEFAULT_URL_FOR_TEST,
@@ -8,6 +9,8 @@ from sglang.test.test_utils import (
    popen_launch_server,
 )

+register_amd_ci(est_time=313, suite="stage-b-test-small-1-gpu-amd")
+

 class TestMixtralAccuracy(CustomTestCase):
    @classmethod
--- a/test/registered/rl/test_lora_load_from_tensor.py
+++ b/test/registered/rl/test_lora_load_from_tensor.py
@@ -11,7 +11,7 @@ from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import CustomTestCase

 register_cuda_ci(est_time=90, suite="stage-b-test-small-1-gpu")
-register_amd_ci(est_time=90, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=90, suite="stage-b-test-small-1-gpu-amd")

 MODEL_PATH = "Qwen/Qwen3-0.6B"
 LORA_REPO = "charent/self_cognition_Alice"
--- a/test/registered/rotary/test_rope_rocm.py
+++ b/test/registered/rotary/test_rope_rocm.py
@@ -4,8 +4,11 @@ import torch

 from sglang.srt.layers.rotary_embedding import RotaryEmbedding
 from sglang.srt.utils import get_bool_env_var, is_hip
+from sglang.test.ci.ci_register import register_amd_ci
 from sglang.test.test_utils import CustomTestCase

+register_amd_ci(est_time=3, suite="stage-b-test-small-1-gpu-amd")
+
 torch.manual_seed(0)

 _is_hip = is_hip()
--- a/test/registered/test_srt_backend.py
+++ b/test/registered/test_srt_backend.py
@@ -20,7 +20,7 @@ from sglang.test.test_programs import (
 from sglang.test.test_utils import DEFAULT_MODEL_NAME_FOR_TEST, CustomTestCase

 register_cuda_ci(est_time=80, suite="stage-a-test-1")
-register_amd_ci(est_time=120, suite="stage-a-test-1")
+register_amd_ci(est_time=120, suite="stage-a-test-1-amd")


 class TestSRTBackend(CustomTestCase):
--- a/test/registered/utils/test_bench_typebaseddispatcher.py
+++ b/test/registered/utils/test_bench_typebaseddispatcher.py
@@ -1,8 +1,11 @@
 import timeit
 from typing import Any, Callable, List, Tuple, Type

+from sglang.test.ci.ci_register import register_amd_ci
 from sglang.utils import TypeBasedDispatcher

+register_amd_ci(est_time=10, suite="stage-b-test-small-1-gpu-amd")
+

 class TypeBasedDispatcherList:
    def __init__(self, mapping: List[Tuple[Type, Callable]]):
--- a/test/registered/utils/test_type_based_dispatcher.py
+++ b/test/registered/utils/test_type_based_dispatcher.py
@@ -8,8 +8,11 @@ import timeit
 import unittest

 from sglang.srt.managers.io_struct import SamplingParams
+from sglang.test.ci.ci_register import register_amd_ci
 from sglang.utils import TypeBasedDispatcher

+register_amd_ci(est_time=10, suite="stage-b-test-small-1-gpu-amd")
+

 class TestTypeBasedDispatcher(unittest.TestCase):
    """Unit tests for TypeBasedDispatcher e2e performance."""
--- a/test/run_suite.py
+++ b/test/run_suite.py
@@ -19,7 +19,7 @@ HW_MAPPING = {
 PER_COMMIT_SUITES = {
    HWBackend.CPU: ["default", "stage-a-cpu-only"],
    HWBackend.AMD: [
-        "stage-a-test-1",
+        "stage-a-test-1-amd",
        "stage-b-test-small-1-gpu-amd",
        "stage-b-test-small-1-gpu-amd-mi35x",
        "stage-b-test-large-2-gpu-amd",
--- a/test/srt/run_suite.py
+++ b/test/srt/run_suite.py
@@ -80,10 +80,7 @@ suite_amd = {
        # TestFile("lora/test_lora_backend.py", 99), # Disabled temporarily, see https://github.com/sgl-project/sglang/issues/13107
        # TestFile("lora/test_lora_cuda_graph.py", 250), # Disabled temporarily, see https://github.com/sgl-project/sglang/issues/13107
        # TestFile("lora/test_lora_qwen3.py", 97), # Disabled temporarily, see https://github.com/sgl-project/sglang/issues/13107
-        TestFile("test_bench_typebaseddispatcher.py", 10),
-        TestFile("test_rope_rocm.py", 3),
        # TestFile("test_torch_compile_moe.py", 210), # Disabled temporarily, see https://github.com/sgl-project/sglang/issues/13107
-        TestFile("test_type_based_dispatcher.py", 10),
        # Disabled temporarily
        # TestFile("test_vlm_input_format.py", 300),
        # TestFile("openai_server/features/test_openai_server_hidden_states.py", 240),
@@ -92,15 +89,10 @@ suite_amd = {
        # TestFile("test_vision_chunked_prefill.py", 175), # Disabled temporarily and track in #7701
        # TestFile("test_wave_attention_backend.py", 150), # Disabled temporarily, see https://github.com/sgl-project/sglang/issues/11127
        # The time estimation for `test_int4fp8_moe.py` assumes `mistralai/Mixtral-8x7B-Instruct-v0.1` is already cached (running on 1xMI300X).
-        TestFile("test_int4fp8_moe.py", 313),
    ],
    "per-commit-4-gpu-amd": [
        TestFile("test_pp_single_node.py", 150),
    ],
-    "per-commit-8-gpu-amd": [
-        TestFile("test_deepseek_v3_basic.py", 275),
-        TestFile("test_deepseek_v3_mtp.py", 275),
-    ],
    # NOTE: AMD nightly suites (nightly-amd, nightly-amd-vlm, nightly-amd-8-gpu)
    # have been migrated to test/registered/amd/nightly/ and are now managed
    # by test/run_suite.py using the registry system.