David Rubin
David Rubin
Upstream: https://github.com/llvm/llvm-project/issues/99639
It turns out this was fixed in LLVM just a month ago, so it should be gone when we update to LLVM 19.
I am trying to understand a use case for this. Could you explain it a bit more?
You've described what you're proposing, but when would you need this? What is a real-world usecase?
> > I'm unsure if this is an issue that needs fixing. > > The issue is #21655 , where correct code for a generic data structure in status-quo would...
I am quite certain that `std.mem.reverse` will get SIMD optimized. It would use something like `vperm*`. 1. Could you please post benchmarks on this PR that showcase how it's faster...
> If you'd like I can send the benchmarks directly here, or you can see them at this link https://github.com/JonathanHallstrom/array_reversing/tree/main/bench_results. They are a little bit out of date but I...
> You're quite right that constant size buffers get optimized, I'm relying on that in my implementation. It probably would have been good if I clarified that I meant I'm...
> ~Just noticed that my change seems [to make constant size array case worse](https://zig.godbolt.org/).~ Accidentally used the wrong version of the code. Heres the up to date version, [no change...
> > It doesn't really need to be a library. Just running it a thousand times and using a cycle counter with a `clflush` in between runs would be just...