Victor M. Alvarez

Results 206 comments of Victor M. Alvarez

Something strange is going on here, can you tell me more about your setup? Like `rustc` version and which exact commit of YARA-X you are compiling? Also, if you have...

I've upgraded from rustc 1.80 to 1.81 and there's a huge performance drop, my numbers are now similar to yours: ``` [2024-09-26T08:21:16Z INFO yara_x::compiler] WASM module build time: 159.544566313s [2024-09-26T08:21:20Z...

This is the flamegraph for when compiled with rustc 1.80: ![image](https://github.com/user-attachments/assets/119f6988-58b3-446c-8028-10053f781434) And this other with rustc 1.81: ![image](https://github.com/user-attachments/assets/5441c1f0-9c5f-4c12-9479-fba2c36dc39b) There's a huge performance difference in the [merge_bundle](https://github.com/bytecodealliance/regalloc2/blob/e684ee565d5b55e5d7bde6efa50996d5505f038c/src/ion/merge.rs#L24) function. The `merge_bundle` function...

There were some changes in sorting algorithms in the Rust standard library between versions 1.80 and 1.81. https://github.com/rust-lang/rust/pull/124032/ It was also mentioned in the release notes: https://releases.rs/docs/1.81.0/#libraries At least in...

I will open an issue in the wasmtime project about this regression between rust 1.80 and 1.81 and see what they say. I tried with the latest version of wasmtime...

The issues with rustc 1.81 should be solved once this fix is merged and propagated to `wasmtime` and `yara-x`: https://github.com/bytecodealliance/regalloc2/pull/204

I'm afraid this is expected. Any other program (like `cat`) that tries to read those files exhibit the same behavior, they get blocked waiting for data to be read. This...

@ruppde is your dataset public? I would like to investigate why it's 25% slower.

I've been experimenting with the `daachorse` crate and the results are promising. I see a 20%-25% increase in raw scanning speed. However the API is not fully compatible with the...

> > I've been experimenting with the `daachorse` crate and the results are promising. I see a 20%-25% increase in raw scanning speed. However the API is not fully compatible...