[Refactor] Cleanup platform (#5566)
### What this PR does / why we need it?
1. add `COMPILATION_PASS_KEY` constant
2. clean up useless platform interface `empty_cache`, `synchronize`,
`mem_get_info`, `clear_npu_memory`
3. rename `CUSTOM_OP_REGISTERED` to `_CUSTOM_OP_REGISTERED`
4. remove uesless env `VLLM_ENABLE_CUDAGRAPH_GC`
NPUPlatform is the interface called by vLLM. Do not call it inner
vllm-ascend.
### Does this PR introduce _any_ user-facing change?
This PR is just a cleanup. All CI should pass.
### How was this patch tested?
- vLLM version: v0.13.0
- vLLM main:
7157596103
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
This commit is contained in:
@@ -879,7 +879,6 @@ class TestAscendMLAImpl(TestBase):
|
||||
B, H, D = 4, self.impl.num_heads, self.impl.v_head_dim # total: [4, 4, 8]
|
||||
test_cases = [(1, 1), (1, 2), (2, 1), (2, 2), (4, 4)]
|
||||
for test_case in test_cases:
|
||||
print(test_case)
|
||||
self.impl.dcp_size = test_case[0]
|
||||
self.impl.pcp_size = test_case[1]
|
||||
mock_dcp.world_size = test_case[0]
|
||||
|
||||
Reference in New Issue
Block a user