minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[AMD][CI] Adjust MI300 score API performance thresholds (#36290)

Browse Source
This commit is contained in:
Bingxu Chen
2026-08-25 09:41:36 -07:00
committed by GitHub
parent e9c9df6a52
commit 96a73c4e15
1 changed files with 9 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/registered/perf/test_bench_serving_1gpu_part2.py
+9 -3
View File
@@ -94,9 +94,15 @@ class TestBenchServing1GPUPart2(CustomTestCase):
) )
self.assertEqual(res["successful_requests"], res["total_requests"]) self.assertEqual(res["successful_requests"], res["total_requests"])
self.assertLess(res["avg_latency_ms"], 48) # relax for mi300x
self.assertLess(res["p95_latency_ms"], 50) if is_in_amd_ci():
self.assertGreater(res["throughput"], 20) self.assertLess(res["avg_latency_ms"], 60)
self.assertLess(res["p95_latency_ms"], 65)
self.assertGreater(res["throughput"], 16)
else:
self.assertLess(res["avg_latency_ms"], 48)
self.assertLess(res["p95_latency_ms"], 50)
self.assertGreater(res["throughput"], 20)
def test_score_api_batch_scaling(self): def test_score_api_batch_scaling(self):
"""Test score API performance with different batch sizes""" """Test score API performance with different batch sizes"""