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

[Hybrid]: Decouple Kernel Block Size from KV Page Size (#24486)

Browse Source 
Signed-off-by: lizhiyuan <uniartisan2017@gmail.com>
Signed-off-by: Zhiyuan Li <uniartisan2017@gmail.com>
This commit is contained in:
Zhiyuan Li
2025-10-09 14:43:39 +08:00
committed by GitHub
parent d17f0fbf30
commit d24cf322e1
18 changed files with 573 additions and 55 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
tests/v1/worker/test_gpu_input_batch.py
View File

@@ -241,6 +241,7 @@ def test_sampling_metadata_in_input_batch(device: str, batch_size: int):
pin_memory=is_pin_memory_available(),
vocab_size=1024,
block_sizes=[1],
kernel_block_sizes=[1],
)
reqs: list[CachedRequestState] = []
req_id_reqs = {}
@@ -335,6 +336,7 @@ def test_swap_states_in_input_batch(device: str, batch_size: int, swap_list: lis
pin_memory=is_pin_memory_available(),
vocab_size=1024,
block_sizes=[1],
kernel_block_sizes=[1],
)
ref_input_batch: InputBatch = InputBatch(
max_num_reqs=batch_size,
@@ -344,6 +346,7 @@ def test_swap_states_in_input_batch(device: str, batch_size: int, swap_list: lis
pin_memory=is_pin_memory_available(),
vocab_size=1024,
block_sizes=[1],
kernel_block_sizes=[1],
)
reqs: list[CachedRequestState] = []