Branch: refs/heads/main
Revision: 04d32d4
Author: bmcdonald3
Link: Add new knucleotide benchmark with custom record and hash function by bmcdonald3 · Pull Request #18489 · chapel-lang/chapel · GitHub
Log Message:
Merge pull request #18489 from bmcdonald3/new-knuc-test
Add new knucleotide benchmark with custom record and hash function
[ reviewed by @ronawho ]
This new benchmark takes advantage of the user-defined hash from this PR
Here are the numbers that I was seeing:
| Current main | This PR | This PR 1.25 | Regular 1.25 |
|---|---|---|---|
| 5.162s | 3.962s | 8.483s | 8.417s |
So, we see a speedup on Chapel 1.26, but it is the exact same performance in 1.25, it just has some strange, unnecessary code for 1.25 that doesn't affect how the program will run.
Motivated by: https://github.com/Cray/chapel-private/issues/2365
Work this PR utilizes: Allow user-defined hash functions on user-defined records by bmcdonald3 · Pull Request #18343 · chapel-lang/chapel · GitHub
Modified Files:
A test/studies/shootout/k-nucleotide/benmcd/EXECOPTS
A test/studies/shootout/k-nucleotide/benmcd/PERFEXECOPTS
A test/studies/shootout/k-nucleotide/benmcd/PERFTIMEEXEC
A test/studies/shootout/k-nucleotide/benmcd/PREEXEC
A test/studies/shootout/k-nucleotide/benmcd/SKIPIF
A test/studies/shootout/k-nucleotide/benmcd/fasta.small
A test/studies/shootout/k-nucleotide/benmcd/knucleotide-user-hash.chpl
A test/studies/shootout/k-nucleotide/benmcd/knucleotide-user-hash.good
A test/studies/shootout/k-nucleotide/benmcd/knucleotide-user-hash.perfkeys
A test/studies/shootout/k-nucleotide/benmcd/knucleotide-user-hash.suppressif
M test/studies/shootout/k-nucleotide/knucleotide.graph
Compare: https://github.com/chapel-lang/chapel/compare/ea6b72c1c3f2...04d32d4e1d7b