Skip to content
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

Jump to bottom

Use cudaOccupancyMaxPotentialBlockSize to calculate the block size. #5926

Merged
merged 2 commits into from
Jul 23, 2020
Merged

Use cudaOccupancyMaxPotentialBlockSize to calculate the block size. #5926

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
a48f3eb
Use `cudaOccupancyMaxPotentialBlockSize` to calculate the block size.
trivialfis Jul 22, 2020
63abec4
Windows build.
trivialfis Jul 23, 2020
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
9 changes: 7 additions & 2 deletions src/tree/gpu_hist/histogram.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -175,7 +175,11 @@ void BuildGradientHistogram(EllpackDeviceAccessor const& matrix,
}

// determine the launch configuration
unsigned block_threads = shared ? 1024 : 256;
int min_grid_size;
int block_threads = 1024;
dh::safe_cuda(cudaOccupancyMaxPotentialBlockSize(
&min_grid_size, &block_threads, kernel, smem_size, 0));

int num_groups = feature_groups.NumGroups();
int n_mps = 0;
dh::safe_cuda(cudaDeviceGetAttribute(&n_mps, cudaDevAttrMultiProcessorCount, device));
Expand All @@ -199,7 +203,8 @@ void BuildGradientHistogram(EllpackDeviceAccessor const& matrix,
grid_size = common::DivRoundUp(grid_size,
common::DivRoundUp(num_groups, num_groups_threshold));

dh::LaunchKernel {dim3(grid_size, num_groups), block_threads, smem_size} (
dh::LaunchKernel {
dim3(grid_size, num_groups), static_cast<uint32_t>(block_threads), smem_size} (
kernel,
matrix, feature_groups, d_ridx, histogram.data(), gpair.data(), rounding,
shared);
Expand Down