Skip to content

Fix errors of globalToShmemAsyncCopy.cu #263

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

Open
aioprli wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Fix errors of globalToShmemAsyncCopy.cu #263

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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: 3 additions & 3 deletions Samples/3_CUDA_Features/globalToShmemAsyncCopy/globalToShmemAsyncCopy.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -128,7 +128,7 @@ __global__ void MatrixMulAsyncCopyMultiStageLargeChunk(
a <= aEnd; a += aStep, b += bStep, ++i) {
// Load the matrices from device memory to shared memory; each thread loads
// one element of each matrix
for (; aStage <= a + aStep * maxPipelineStages;
for (; aStage < a + aStep * maxPipelineStages;
aStage += aStep, bStage += bStep, ++iStage) {
pipe.producer_acquire();
if (aStage <= aEnd && t4x < BLOCK_SIZE) {
Expand All @@ -137,7 +137,7 @@ __global__ void MatrixMulAsyncCopyMultiStageLargeChunk(
cuda::memcpy_async(&As[j][threadIdx.y][t4x],
&A[aStage + wA * threadIdx.y + t4x], shape4, pipe);
cuda::memcpy_async(&Bs[j][threadIdx.y][t4x],
&B[aStage + wA * threadIdx.y + t4x], shape4, pipe);
&B[bStage + wB * threadIdx.y + t4x], shape4, pipe);
}
pipe.producer_commit();
}
Expand Down Expand Up @@ -222,7 +222,7 @@ __global__ void MatrixMulAsyncCopyLargeChunk(float *__restrict__ C,

cuda::memcpy_async(&As[threadIdx.y][t4x], &A[a + wA * threadIdx.y + t4x],
shape4, pipe);
cuda::memcpy_async(&Bs[threadIdx.y][t4x], &B[a + wA * threadIdx.y + t4x],
cuda::memcpy_async(&Bs[threadIdx.y][t4x], &B[b + wB * threadIdx.y + t4x],
shape4, pipe);

pipe.producer_commit();
Expand Down