Skip to content

support kvcacheio#163

Merged
iforgetmyname merged 6 commits intosgl-project:mainfrom
husf1130:br_kvcacheio
Nov 5, 2025
Merged

support kvcacheio#163
iforgetmyname merged 6 commits intosgl-project:mainfrom
husf1130:br_kvcacheio

Conversation

@husf1130
Copy link
Copy Markdown

@husf1130 husf1130 commented Nov 5, 2025

support kvcacheio

@gemini-code-assist
Copy link
Copy Markdown
Contributor

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

Copy link
Copy Markdown
Collaborator

@xiaomingbao008 xiaomingbao008 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

please address the 5 comments

Comment thread python/sgl_kernel_npu/sgl_kernel_npu/kvcacheio.py Outdated
Comment thread python/sgl_kernel_npu/sgl_kernel_npu/kvcacheio.py
Comment thread python/sgl_kernel_npu/sgl_kernel_npu/kvcacheio.py Outdated
Comment thread python/sgl_kernel_npu/sgl_kernel_npu/kvcacheio.py Outdated
Comment thread python/sgl_kernel_npu/sgl_kernel_npu/kvcacheio.py
@husf1130
Copy link
Copy Markdown
Author

husf1130 commented Nov 5, 2025

UT Test Results:

python3 tests/python/sgl_kernel_npu/test_transfer_kv_dim_exchange.py
kv transfer D2H, v_empty=True, index_k_empty=True, 2d copy times is 30, total copy size is 251658240 bytes, total duration 21.995ms
.kv transfer H2D, v_empty=True, index_k_empty=True, 2d copy times is 30, total copy size is 251658240 bytes, total duration 42.014ms
.kv transfer D2H, v_empty=False, index_k_empty=True, 2d copy times is 60, total copy size is 503316480 bytes, total duration 102.387ms
.kv transfer H2D, v_empty=False, index_k_empty=True, 2d copy times is 60, total copy size is 503316480 bytes, total duration 95.783ms
.kv transfer D2H, v_empty=False, index_k_empty=False, 2d copy times is 90, total copy size is 754974720 bytes, total duration 142.581ms
.kv transfer H2D, v_empty=False, index_k_empty=False, 2d copy times is 90, total copy size is 754974720 bytes, total duration 140.975ms
.
----------------------------------------------------------------------
Ran 6 tests in 7.970s

OK

@husf1130 husf1130 closed this Nov 5, 2025
@husf1130 husf1130 reopened this Nov 5, 2025
Comment thread python/sgl_kernel_npu/sgl_kernel_npu/kvcacheio.py Outdated
Comment thread python/sgl_kernel_npu/sgl_kernel_npu/kvcacheio.py Outdated
@iforgetmyname iforgetmyname merged commit 02318a4 into sgl-project:main Nov 5, 2025
5 of 8 checks passed
@husf1130 husf1130 deleted the br_kvcacheio branch January 31, 2026 08:33
AndyKong2020 pushed a commit to AndyKong2020/sgl-kernel-npu that referenced this pull request Mar 24, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants