arnaucube
/
ark-r1cs-std-cherry-picked
mirror of https://github.com/arnaucube/ark-r1cs-std-cherry-picked.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
Browse Source

--emit=asm seemed to be hurting performance significantly... So lets not recommend it for now.

master
jon-chuang 4 years ago
parent
a919e8658c
commit
535f93d378
1 changed files with 2 additions and 1 deletions
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. +2
    -1
      README.md

+ 2
- 1
README.md
View File

@ -78,8 +78,9 @@ cargo +nightly bench
Compiling with `adcxq`, `adoxq` and `mulxq` instructions can lead to a 30-70% speedup. These are available on most `x86_64` platforms (Broadwell onwards for Intel and Ryzen onwards for AMD). Run the following command: Compiling with `adcxq`, `adoxq` and `mulxq` instructions can lead to a 30-70% speedup. These are available on most `x86_64` platforms (Broadwell onwards for Intel and Ryzen onwards for AMD). Run the following command:
```bash ```bash
RUSTFLAGS="--emit=asm -C target-feature=+bmi2,+adx" cargo +nightly test/build/bench --features asm
RUSTFLAGS="-C target-feature=+bmi2,+adx" cargo +nightly test/build/bench --features asm
``` ```
Tip: If optimising for performance, your mileage may vary with passing `--emit=asm` to `RUSTFLAGS`.
To bench `algebra-benches` with greater accuracy, especially for functions with execution times on the order of nanoseconds, use the `n_fold` feature to run selected functions 1000x per iteration. To run with multiple features, make sure to double quote the features. To bench `algebra-benches` with greater accuracy, especially for functions with execution times on the order of nanoseconds, use the `n_fold` feature to run selected functions 1000x per iteration. To run with multiple features, make sure to double quote the features.
```bash ```bash

Write Preview
Loading…
Cancel
Save