biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Misc] Move GPUModelRunner.prepare_kernel_block_sizes to utils (#35400)

Browse Source 
Signed-off-by: NickLucche <nlucches@redhat.com>
This commit is contained in:
Nicolò Lucchesi
2026-02-27 04:42:51 +01:00
committed by GitHub
parent 06be53563b
commit cabdaa7619
3 changed files with 135 additions and 123 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
tests/v1/worker/test_gpu_model_runner.py
View File

@@ -38,7 +38,7 @@ from vllm.v1.kv_cache_interface import (
from vllm.v1.sample.metadata import SamplingMetadata
from vllm.v1.worker.gpu_input_batch import InputBatch
from vllm.v1.worker.gpu_model_runner import GPUModelRunner
from vllm.v1.worker.utils import AttentionGroup
from vllm.v1.worker.utils import AttentionGroup, select_common_block_size
BLOCK_SIZE = 16
NUM_BLOCKS = 10
@@ -209,7 +209,7 @@ def test_select_common_block_size_prefers_manager_block_size():
AttentionGroup(backend_b, [], [], _make_kv_cache_spec(), 0),
]
selected_size = GPUModelRunner.select_common_block_size(128, attn_groups)
selected_size = select_common_block_size(128, attn_groups)
assert selected_size == 128
@@ -221,7 +221,7 @@ def test_select_common_block_size_uses_largest_shared_int():
AttentionGroup(backend_b, [], [], _make_kv_cache_spec(), 0),
]
selected_size = GPUModelRunner.select_common_block_size(256, attn_groups)
selected_size = select_common_block_size(256, attn_groups)
assert selected_size == 64
@@ -234,7 +234,7 @@ def test_select_common_block_size_no_valid_option():
]
with pytest.raises(ValueError):
GPUModelRunner.select_common_block_size(48, attn_groups)
select_common_block_size(48, attn_groups)
def test_update_states_new_request(model_runner, dist_init):