Skip to content

Fix fused token prune in low cuda arch #44391

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 2 commits into from
Jul 18, 2022
Merged

Fix fused token prune in low cuda arch #44391

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
2fb8bee
fix build error in low arch
RichardWooSJTU Jul 18, 2022
ae977a5
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into…
RichardWooSJTU Jul 18, 2022
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
4 changes: 4 additions & 0 deletions paddle/fluid/inference/tensorrt/plugin/fused_token_prune_op_plugin.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,10 +38,12 @@ __global__ void ElementwiseMask(const T* a,
const T* b,
T* res,
int num_elements) {
#if CUDA_ARCH_FP16_SUPPORTED(__CUDA_ARCH__)
auto tid = threadIdx.x + blockIdx.x * blockDim.x;
if (tid >= num_elements) return;
const T zero = 0;
res[tid] = b[tid] >= zero ? a[tid] : zero;
#endif
}

template <typename T>
Expand Down Expand Up @@ -121,6 +123,7 @@ __global__ void ReduceSum2(
template <>
__global__ void ReduceSum2<half>(
const half* src, half* dst, int bsz, int nb_head, int max_seq_len) {
#if CUDA_ARCH_FP16_SUPPORTED(__CUDA_ARCH__)
int tid = threadIdx.x;
int bid = blockIdx.x;
int num_blocks_per_head = ((max_seq_len / blockDim.x) * max_seq_len);
Expand Down Expand Up @@ -152,6 +155,7 @@ __global__ void ReduceSum2<half>(
static_cast<size_t>(bsz * max_seq_len),
static_cast<platform::float16>(res_half[0]));
}
#endif
}

template <typename T>
Expand Down