You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
From reading the trace log, it doesn't look like this is setting the is_halted to true at the cycle count for the split. I expect to see this message: log::info!("SPLIT[{cycle}]> pc: 0x{pc:08x}");
This PR fixes a bug occurring when calling iteratively sha2, ending up
calling the `expand` function. The fix increases the const `SHA_CYCLES`
from 72 to 73.
It also adds a unit test for that case.
Fixes#820
One can get the following error when running a long computation iterating calls to sha256:
This was observed on 0.17.0.
Steps to reproduce:
cd risczero
cargo b -r
target/release/host --out dummy.csv iter-sha2
The guest code is there: https://github.com/igarnier/zk-benchmarking/blob/iter-sha2-failure/risczero/methods/guest/src/bin/iter_sha2.rs
The error was triggered for
num_iter = 10_000
(this is an upper bound, didn't try to bisect).The text was updated successfully, but these errors were encountered: