Skip to content

Fix DeepSeek DP Attention + torch compile #5367

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 4 commits into from
Apr 14, 2025
Merged

Fix DeepSeek DP Attention + torch compile #5367

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
607f6ab
more
fzyzcjy Apr 14, 2025
d50c6cb
fix and add ut
ispobock Apr 14, 2025
d5f50c3
fix
ispobock Apr 14, 2025
7282abe
lint
ispobock Apr 14, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 2 additions & 4 deletions python/sglang/srt/layers/dp_attention.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -192,8 +192,7 @@ def _dp_gather(

if local_tokens.shape[0] > 0 and (is_partial or get_attention_tp_rank() == 0):
assert (
global_tokens.untyped_storage().data_ptr()
!= local_tokens.untyped_storage().data_ptr()
local_tokens.untyped_storage() is not global_tokens.untyped_storage()
), "aliasing between global_tokens and local_tokens not allowed"
memcpy_triton(
global_tokens, local_tokens, 0, local_start_pos, local_num_tokens, False
Expand Down Expand Up @@ -243,8 +242,7 @@ def dp_scatter(
assert global_tokens.is_contiguous()
if local_tokens.shape[0] > 0:
assert (
local_tokens.untyped_storage().data_ptr()
!= global_tokens.untyped_storage().data_ptr()
local_tokens.untyped_storage() is not global_tokens.untyped_storage()
), "aliasing between local_tokens and global_tokens not allowed"
memcpy_triton(
local_tokens, global_tokens, 0, local_start_pos, local_num_tokens, True
Expand Down
5 changes: 3 additions & 2 deletions test/srt/parse_results.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,7 +1,8 @@
import json
import pandas as pd
import argparse
import json
import os

import pandas as pd
from tabulate import tabulate

# Parse command-line arguments
Expand Down
3 changes: 3 additions & 0 deletions test/srt/test_dp_attention.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,6 +28,9 @@ def setUpClass(cls):
"--enable-dp-attention",
"--dp",
"2",
"--enable-torch-compile",
"--torch-compile-max-bs",
"2",
],
)

Expand Down
Loading