Skip to content

[AMD CI] re-enable testcases missed when migrating attention backend tests #16476

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

Closed
bingxche wants to merge 2 commits into from
Closed

[AMD CI] re-enable testcases missed when migrating attention backend tests #16476

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
06e6181
[AMD CI] re-enable testcases missed when migrating attention backend …
bingxche Jan 5, 2026
49965c6
Merge branch 'main' into fix-amd-ci
bingxche Jan 5, 2026
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
3 changes: 2 additions & 1 deletion test/registered/attention/test_create_kvindices.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,11 +4,12 @@
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):
Expand Down
3 changes: 2 additions & 1 deletion test/registered/attention/test_radix_attention.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -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,
Expand All @@ -15,6 +15,7 @@

# 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):
Expand Down
3 changes: 2 additions & 1 deletion test/registered/attention/test_torch_native_attention_backend.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@
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,
Expand All @@ -19,6 +19,7 @@

# 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):
Expand Down
3 changes: 2 additions & 1 deletion test/registered/attention/test_triton_attention_backend.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@
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,
Expand All @@ -21,6 +21,7 @@

# 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=200, suite="stage-b-test-small-1-gpu")


class TestTritonAttnBackend(CustomTestCase):
Expand Down
3 changes: 2 additions & 1 deletion test/registered/attention/test_triton_attention_kernels.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,11 +19,12 @@
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=" ")
Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist bot Jan 5, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

For better clarity and maintainability, it would be helpful to provide a more descriptive reason for disabling this test on AMD CI. A single space is not very informative for future developers who might wonder why this test is disabled. Providing a clear reason, for example "Triton kernels not yet supported on AMD", improves code readability.

Suggested change
register_amd_ci(est_time=30, suite="stage-b-test-small-1-gpu", disabled=" ")
register_amd_ci(est_time=30, suite="stage-b-test-small-1-gpu", disabled="Triton kernels not yet supported on AMD")



def extend_attention_fwd_torch(
Expand Down
3 changes: 2 additions & 1 deletion test/registered/attention/test_triton_sliding_window.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@
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,
Expand All @@ -16,6 +16,7 @@

# 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):
Expand Down
Loading