Reduce allocations per operation to zero #195
Merged
Conversation
This file contains hidden or 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
Signed-off-by: Paulo Gomes <[email protected]>
The previous field holding the block function was causing two
allocations per operation. The logic only existed for testing
purposes, and therefore had no reason to be kept based on the
performance impact.
As a result, the performance specially for smaller payloads
increased up to 24% for AMD64. Likewise, the processing time
dropped by up to 33% in similar input sizes.
A short version of the benchstat can be seen below:
│ /tmp/before │ /tmp/after │
│ sec/op │ sec/op vs base
Hash8Bytes/sha1cd_native-16 247.3n ± 2% 185.7n ± 2% -24.89% (p=0.000 n=10)
Hash8Bytes/sha1cd_generic-16 273.1n ± 1% 214.8n ± 1% -21.36% (p=0.000 n=10)
Hash320Bytes/sha1cd_native-16 975.9n ± 1% 904.2n ± 1% -7.35% (p=0.000 n=10)
Hash320Bytes/sha1cd_generic-16 1.143µ ± 1% 1.084µ ± 0% -5.16% (p=0.000 n=10)
Hash1K/sha1cd_native-16 2.797µ ± 4% 2.507µ ± 1% -10.37% (p=0.000 n=10)
Hash1K/sha1cd_generic-16 3.286µ ± 6% 2.971µ ± 0% -9.57% (p=0.000 n=10)
Hash8K/sha1cd_native-16 19.89µ ± 1% 18.71µ ± 0% -5.95% (p=0.000 n=10)
Hash8K/sha1cd_generic-16 23.80µ ± 1% 22.33µ ± 2% -6.17% (p=0.000 n=10)
HashWithCollision/sha1cd_native-16 2.994µ ± 2% 2.818µ ± 1% -5.88% (p=0.000 n=10)
HashWithCollision/sha1cd_generic-16 3.242µ ± 2% 3.108µ ± 0% -4.15% (p=0.000 n=10)
│ /tmp/before │ /tmp/after │
│ allocs/op │ allocs/op vs base │
Hash8Bytes/sha1cd_native-16 2.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10)
Hash8Bytes/sha1cd_generic-16 2.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10)
Hash320Bytes/sha1cd_native-16 2.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10)
Hash320Bytes/sha1cd_generic-16 2.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10)
Hash1K/sha1cd_native-16 2.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10)
Hash1K/sha1cd_generic-16 2.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10)
Hash8K/sha1cd_native-16 2.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10)
Hash8K/sha1cd_generic-16 2.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10)
HashWithCollision/sha1cd_native-16 2.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10)
HashWithCollision/sha1cd_generic-16 2.000 ± 0% 0.000 ± 0% -100.00% (p=0.000 n=10)
│ /tmp/before │ /tmp/after │
│ B/s │ B/s vs base │
Hash8Bytes/sha1cd_native-16 30.86Mi ± 2% 41.07Mi ± 1% +33.12% (p=0.000 n=10)
Hash8Bytes/sha1cd_generic-16 27.93Mi ± 1% 35.52Mi ± 1% +27.16% (p=0.000 n=10)
Hash320Bytes/sha1cd_native-16 312.7Mi ± 1% 337.5Mi ± 1% +7.93% (p=0.000 n=10)
Hash320Bytes/sha1cd_generic-16 267.0Mi ± 1% 281.6Mi ± 0% +5.45% (p=0.000 n=10)
Hash1K/sha1cd_native-16 349.2Mi ± 4% 389.5Mi ± 1% +11.57% (p=0.000 n=10)
Hash1K/sha1cd_generic-16 297.3Mi ± 6% 328.7Mi ± 0% +10.57% (p=0.000 n=10)
Hash8K/sha1cd_native-16 392.7Mi ± 1% 417.6Mi ± 0% +6.33% (p=0.000 n=10)
Hash8K/sha1cd_generic-16 328.3Mi ± 1% 349.9Mi ± 2% +6.58% (p=0.000 n=10)
HashWithCollision/sha1cd_native-16 203.9Mi ± 2% 216.7Mi ± 1% +6.27% (p=0.000 n=10)
HashWithCollision/sha1cd_generic-16 188.2Mi ± 2% 196.4Mi ± 0% +4.34% (p=0.000 n=10)
Signed-off-by: Paulo Gomes <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
The previous field holding the block function was causing two allocations per operation. The logic only existed for testing
purposes, and therefore had no reason to be kept based on the performance impact.
As a result, the performance specially for smaller payloads increased up to 24% for AMD64. Likewise, the processing time
dropped by up to 33% in similar input sizes.
The benchstat results for AMD64 can be seen below: