Initial powdr acceleration #3
Conversation
| let mut data = Array::default(); | ||
| proptest!(ProptestConfig::with_cases( | ||
| if cfg!(all(target_os = "zkvm", target_arch = "riscv32")) { | ||
| 1 |
There was a problem hiding this comment.
@qwang98 why did you change this from 1000 to 1 in your PR?
There was a problem hiding this comment.
PR #2 is mix of Risc0 field implementation and Powdr modmul syscall, so this part is actually from the Risc0 commit here: https://github.com/risc0/RustCrypto-elliptic-curves/blob/risczero/k256/src/arithmetic/hash2curve.rs#L418-L424. Other ported over elements are also from this same commit.
leonardoalt
force-pushed
the
powdr-leo-steve
branch
2 times, most recently
from
ce5952d
to
87d2805
Compare
| @@ -796,6 +803,22 @@ version = "1.9.0" | |||
| source = "registry+https://github.com/rust-lang/crates.io-index" | |||
| checksum = "cc9c68a3f6da06753e9335d63e27f6b9754dd1920d941135b7ea8224f141adb2" | |||
|
|
|||
| [[package]] | |||
| name = "powdr-riscv-runtime" | |||
| version = "0.1.0-alpha.2" | |||
There was a problem hiding this comment.
this is the current version of that package because we didn't update it with the releases. But we actually merged 0.1.2 today, so a cargo update should change this too
There was a problem hiding this comment.
yea this is actually correct because we use the tag v0.1.1 which had this wrong version
k256/src/arithmetic/field.rs
Outdated
| if #[cfg(target_pointer_width = "32")] { | ||
|
|
||
| if #[cfg(all(target_os = "zkvm", target_arch = "riscv32"))] { | ||
| mod field_8x32_risc0; |
There was a problem hiding this comment.
rn I actually prefer just field_8x32
| #![allow(unsafe_code)] | ||
|
|
||
| // This file is the same as the original below except for minor snippets. | ||
| // https://github.com/risc0/RustCrypto-elliptic-curves/blob/risczero/k256/src/arithmetic/field/field_8x32_risc0.rs |
There was a problem hiding this comment.
change our version to _powdr.rs?
There was a problem hiding this comment.
yea ok. I mean, IMO we could actually only call it 8x32 and keep this attribution, and also not call it powdr
|
review fixed |
|
fixed failing native test |
leonardoalt
force-pushed
the
powdr-leo-steve
branch
from
c07c388
to
f0a0282
Compare
This PR
I think we should just merge this PR to our master and keep working on top.