Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
change
sglang
Commits
439f6580
"src/vscode:/vscode.git/clone" did not exist on "fdb1baa05c8da5b4ed3e7a62200f406dcb26ba79"
Unverified
Commit
439f6580
authored
Jan 06, 2025
by
Ke Bao
Committed by
GitHub
Jan 06, 2025
Browse files
Fix sgl-kernel cu118 compile issue (#2750)
parent
2f0d3864
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
4 additions
and
1 deletion
+4
-1
sgl-kernel/src/sgl-kernel/csrc/trt_reduce_internal.cu
sgl-kernel/src/sgl-kernel/csrc/trt_reduce_internal.cu
+4
-1
No files found.
sgl-kernel/src/sgl-kernel/csrc/trt_reduce_internal.cu
View file @
439f6580
...
@@ -302,8 +302,10 @@ static __global__ void __launch_bounds__(512, 1) twoShotAllReduceKernel(AllReduc
...
@@ -302,8 +302,10 @@ static __global__ void __launch_bounds__(512, 1) twoShotAllReduceKernel(AllReduc
buffers
[
ii
]
=
reinterpret_cast
<
T
*>
(
params
.
peer_comm_buffer_ptrs
[
rank
]);
buffers
[
ii
]
=
reinterpret_cast
<
T
*>
(
params
.
peer_comm_buffer_ptrs
[
rank
]);
}
}
#if (defined(__CUDACC_VER_MAJOR__) && (__CUDACC_VER_MAJOR__ >= 12))
#if (defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 900))
#if (defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 900))
cudaGridDependencySynchronize
();
cudaGridDependencySynchronize
();
#endif
#endif
#endif
block_barrier
(
params
.
peer_barrier_ptrs_in
,
params
.
barrier_flag
,
params
.
local_rank
,
RANKS_PER_NODE
,
tidx
,
bidx
,
block_barrier
(
params
.
peer_barrier_ptrs_in
,
params
.
barrier_flag
,
params
.
local_rank
,
RANKS_PER_NODE
,
tidx
,
bidx
,
...
@@ -350,10 +352,11 @@ static __global__ void __launch_bounds__(512, 1) twoShotAllReduceKernel(AllReduc
...
@@ -350,10 +352,11 @@ static __global__ void __launch_bounds__(512, 1) twoShotAllReduceKernel(AllReduc
*
reinterpret_cast
<
int4
*>
(
&
local_output_buffer
[
offset_rank
])
=
*
reinterpret_cast
<
int4
*>
(
&
buffers
[
ii
][
offset_rank
]);
*
reinterpret_cast
<
int4
*>
(
&
local_output_buffer
[
offset_rank
])
=
*
reinterpret_cast
<
int4
*>
(
&
buffers
[
ii
][
offset_rank
]);
}
}
}
}
#if (defined(__CUDACC_VER_MAJOR__) && (__CUDACC_VER_MAJOR__ >= 12))
#if (defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 900))
#if (defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 900))
cudaTriggerProgrammaticLaunchCompletion
();
cudaTriggerProgrammaticLaunchCompletion
();
#endif
#endif
#endif
}
}
////////////////////////////////////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////////////////////////////////////
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment