New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Extracting all lanes from a v128 #1529

Open

verbessern opened this issue Sep 16, 2024 · 1 comment

Labels

simd

verbessern commented Sep 16, 2024 •

edited

Loading

At the moment (as far as I see) the only way to extract more then one lane from a v128 value is to do the following:

(local $t v128)
...
  local.tee $t
  f32x4.extract_lane 0
  local.get $t
  f32x4.extract_lane 1
  local.get $t
  f32x4.extract_lane 2
  local.get $t
  f32x4.extract_lane 3
...

This seems quite inefficient, and clearly has a large storage footprint. I'm wondering whether there is a need of [f32x4,f64x2,...].extract_all.

The text was updated successfully, but these errors were encountered:

Member

sunfishcode commented Dec 6, 2024

Some things that would be useful to motivate this issue are:

How often does this sequence occur in realistic applications?
What CPU architectures have SIMD instructions that could be used to optimize extract_all better than 4 separate extracts?

sunfishcode added the simd label

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment