Reduce stream syncs in split tools #40
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| @@ -779,6 +766,7 @@ void copy_keys_if(split::SplitVector<T, split::split_unified_allocator<T>>& inpu | |||
| const size_t memory_for_pool = 8 * nBlocks * sizeof(uint32_t); | |||
| Cuda_mempool mPool(memory_for_pool, s); | |||
| auto len = copy_keys_if_raw(input, output.data(), rule, nBlocks, mPool, s); | |||
| SPLIT_CHECK_ERR(split_gpuStreamSynchronize(s)); | |||
There was a problem hiding this comment.
This one is not needed. Copy_keys_if_raw has a stream sync before returning.
| @@ -622,14 +622,12 @@ uint32_t copy_if_raw(split::SplitVector<T, split::split_unified_allocator<T>>& i | |||
| uint32_t* d_counts; | |||
There was a problem hiding this comment.
up above I see what I believe are unnecessary syncs:
on lines 595, 603, 611
| @@ -622,14 +622,12 @@ uint32_t copy_if_raw(split::SplitVector<T, split::split_unified_allocator<T>>& i | |||
| uint32_t* d_counts; | |||
| uint32_t* d_offsets; | |||
| d_counts = (uint32_t*)mPool.allocate(nBlocks * sizeof(uint32_t)); | |||
| SPLIT_CHECK_ERR(split_gpuStreamSynchronize(s)); | |||
| SPLIT_CHECK_ERR(split_gpuMemsetAsync(d_counts, 0, nBlocks * sizeof(uint32_t),s)); | |||
There was a problem hiding this comment.
d_counts here is passed to scan_reduce_raw as output, and in that kernel it gets directly written to, not incremented. Thus, the memset appears unnecessary, as long as the kernel actually writes to all elements. This same logic check should be done to the other memsets as well.
There was a problem hiding this comment.
To be addressed later as I am in a bit of a git wreck
|
This can be close as it is encapsulated in PR #48. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Reduces number of stream syncs in spit tools.