Author here, AMA.
Any thoughts about SIMD-related crates?
Rust SIMD on the GPU
41–50 of 130 posts
Re: Rust SIMD on the GPU
#42I love how ever example of portable SIMD isn't portable. They specifies a constant SIMD width so it's non-portable. Well, not performance portable, but why are we using SIMD again?
Go's implementation is vector size independant https://pkg.go.dev/simd@master
Re: Rust SIMD on the GPU
#43My heard hurts - i was stupid enough to think that SIMD was a CPU only thing - I don't understand why it would be ported to GPU - huge kudos to managing to surprise me
GPU "cores" are basically what a CPU would call SIMD lanes. So a GPU with 1024 'CUDA cores' might be structured as 16 relatively independent pieces that a CPU might call a core, each with a 64 wide SIMD unit.
Re: Rust SIMD on the GPU
#44Re: Rust SIMD on the GPU
#45The author mentions Rust's portable SIMD library [0]. The only issue with portable SIMD is it's only available on nightly. I used it in my FFT crate, but we had to switch to the fearless_simd crate in order to get a portable SIMD solution that works on stable [1]. [0] https://doc.rust-lang.org/std/simd/index.html [1] https://github.com/linebender/fearless_simd
Pretty common for Rust to cook things in nightly for a very long time; I wouldn't consider it a bad thing, tbh.
Re: Rust SIMD on the GPU
#46My heard hurts - i was stupid enough to think that SIMD was a CPU only thing - I don't understand why it would be ported to GPU - huge kudos to managing to surprise me
GPU "cores" are basically what a CPU would call SIMD lanes. So a GPU with 1024 'CUDA cores' might be structured as 16 relatively independent pieces that a CPU might call a core, each with a 64 wide SIMD unit.
32 wide - only AMD has a 64 wide mode
Re: Rust SIMD on the GPU
#47I love how ever example of portable SIMD isn't portable. They specifies a constant SIMD width so it's non-portable. Well, not performance portable, but why are we using SIMD again?
Why should it be portable? Honest question.
SIMD seems to me, to be very platform specific. Maybe there are times one SIMD unit is not anothers' SIMD unit?
Re: Rust SIMD on the GPU
#48I love how ever example of portable SIMD isn't portable. They specifies a constant SIMD width so it's non-portable. Well, not performance portable, but why are we using SIMD again?
Why should it be portable? Honest question. SIMD seems to me, to be very platform specific. Maybe there are times one SIMD unit is not anothers' SIMD unit?
The create is called portable_simd.
There is no reason a portable_simd relu_dot implemention should need to specify the SIMD width.
But the design and documentation of portable_simd makes the fixed size syntactically easy/the default and the width agnostic code harder.
Re: Rust SIMD on the GPU
#49Love the pendantic mode setting on the website
Re: Rust SIMD on the GPU
#50Author here, AMA.
[dead]