-
Notifications
You must be signed in to change notification settings - Fork 248
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.
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
Performance issue on SHA3-Keccak256 since v0.10.7 #468
Comments
Could you run your benchmark with |
I'm very surprised to hear about any performance regressions on x86/x86_64, as all of the relevant changes I can think of were ARM-related |
@Slixe what would be really helpful is if you could |
I've tried with the keccak v0.1.2 on sha3 v0.10.7 and still the same. I've also tried with lower version of digest too, same result. I've tried also on rust toolchain 1.69.0 (latest), 1.68, 1.65, 1.64, same for these versions... Performances are still much better on sha3 v0.10.6, and I don't understand why. |
According to the This change is what reduced my performances: 9b218cf#diff-90cc61bbc907b3ec5bb54982793be9757502097a40a475873b6bab46d5874889L19 If I put |
Interesting /cc @aewag |
Hmm, my first guess: maybe it is because the Could you change the call within
This should result in the same speed as with using |
No, still not, it's improving only by ~1-5% but still not as much as I had before the update. |
Thanks. Just looked again into it. The issue lies within the generic approach of keccak_p. If I just wrote a fix and tested it. My approach would be to add functions for |
Following are the benchmarks evaluated with current
with p1600 routines added in
@tarcieri If the draft PR in the sponges directory looks good to you, I would add, after a |
Fixed in v0.10.8: #474 |
Hey there,
After an update to the last version for this crate, I've found a strange problem, my performance have decreased to half of what I get with v0.10.6.
I'm using a Ryzen 3700X with 32GB on Debian.
In case you believe its my fault, here is what I saw:
To see the performance issue you can run
cargo run --release --bin xelis_miner -- --benchmark --iterations=1000000
on this commit:
xelis-project/xelis-blockchain@7a31b0b
And on the previous commit to compare.
Thanks
The text was updated successfully, but these errors were encountered: