[Bugfix][HiSparse] Never allocate GPU pages without host backing (#59036)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com> Signed-off-by: Matthew Bonanni <mbonanni@redhat.com> Co-authored-by: Matthew Bonanni <mbonanni@redhat.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
co-authored by
Matthew Bonanni
Claude Opus 5.5
parent
e006d761a5
commit
90e13fc757
@@ -420,6 +420,19 @@ def test_hisparse_rejects_disabled_hybrid_kv_cache_manager(monkeypatch):
|
||||
)
|
||||
|
||||
|
||||
def test_hisparse_rejects_disabled_full_isl_reservation(monkeypatch):
|
||||
monkeypatch.setattr(current_platform, "is_cuda", lambda: True)
|
||||
with pytest.raises(ValueError, match="requires --scheduler-reserve-full-isl"):
|
||||
VllmConfig(
|
||||
attention_config=AttentionConfig(hisparse_config=HiSparseConfig()),
|
||||
scheduler_config=SchedulerConfig(
|
||||
max_model_len=2048,
|
||||
is_encoder_decoder=False,
|
||||
scheduler_reserve_full_isl=False,
|
||||
),
|
||||
)
|
||||
|
||||
|
||||
def test_hisparse_rejects_non_cuda(monkeypatch):
|
||||
monkeypatch.setattr(current_platform, "is_cuda", lambda: False)
|
||||
with pytest.raises(ValueError, match="requires NVIDIA CUDA"):
|
||||
|
||||
Reference in New Issue
Block a user