Raise default threads for BAM parsing 8->32, BLAS 8->16 #292
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
|
This is ready to go, but we need to measure the memory usage of parsing 32 large (10M ref seqs) BAM files in parallel. |
It was originally capped at 8 under the belief that reading 8 BAM files in parallel would saturate the disk, so there would be no benefit of going higher. However, my laptop can read at 4 GB/s, and decompress BAM files perhaps 40 times slower, so it's CPU bottlenecked even with 32 threads. This change is significant, because users have reported slow BAM file parsing. However, it will potentially quadruple the memory usage of the BAM parsing step. Will be benchmarked before merging. The BLAS change is simply because I think 8 CPUs is too conservative.
jakobnissen
force-pushed
the
nthreads
branch
from
757d936 to
60c22a9
Compare
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Outdated suggestions cannot be applied.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.
It was originally capped at 8 under the belief that reading 8 BAM files in parallel would saturate the disk, so there would be no benefit of going higher. However, my laptop can read at 4 GB/s, and decompress BAM files perhaps 40 times slower, so it's CPU bottlenecked even with 32 threads.
This change is significant, because users have reported slow BAM file parsing. However, it will potentially quadruple the memory usage of the BAM parsing step. Will be benchmarked before merging.
The BLAS change is simply because I think 8 CPUs is too conservative.