Use PackedBaseField in avx fft

[spapinistarkware]

This change is 

[graphite-app]

Your org has enabled the Graphite merge queue for merging into dev

Add the label “merge-queue” to the PR and Graphite will automatically add it to the merge queue when it’s ready to merge.

You must have a Graphite account in order to use the merge queue. Sign up using this link.

[spapinistarkware]

• Use PackedBaseField in avx fft #455 👈
• dev

This stack of pull requests is managed by Graphite. Learn more about stacking.

Join @spapinistarkware and the rest of your teammates on Graphite

[ilyalesokhin-starkware]

src/core/backend/avx512/m31.rs line 41 at r2 (raw file):

    }

    pub fn permute_with(self, idx: __m512i, other: Self) -> Self {

document.

condiser
adding
interleave_lower_halfs
and
interleave_upper_halfs

functions.

Code quote:

permute_with

[ilyalesokhin-starkware]

Reviewed 2 of 4 files at r1, all commit messages.
Reviewable status: 2 of 4 files reviewed, 1 unresolved discussion (waiting on @spapinistarkware)

[spapinistarkware]

Reviewable status: 0 of 4 files reviewed, 1 unresolved discussion (waiting on @ilyalesokhin-starkware)

---

src/core/backend/avx512/m31.rs line 41 at r2 (raw file):

Previously, ilyalesokhin-starkware wrote…

document.

condiser
adding
interleave_lower_halfs
and
interleave_upper_halfs

functions.

Done.

[ilyalesokhin-starkware]

src/core/backend/avx512/m31.rs line 41 at r2 (raw file):

Previously, spapinistarkware (Shahar Papini) wrote…

Done.

you didn't document permute_with, tough I guess it is no longer used.

[spapinistarkware]

Reviewable status: 0 of 4 files reviewed, all discussions resolved (waiting on @ilyalesokhin-starkware)

---

src/core/backend/avx512/m31.rs line 41 at r2 (raw file):

Previously, ilyalesokhin-starkware wrote…

you didn't document permute_with, tough I guess it is no longer used.

Sorry, meant to delete it.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.04%. Comparing base (7ce4ce7) to head (078632b).

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #455      +/-   ##
==========================================
- Coverage   96.05%   96.04%   -0.02%     
==========================================
  Files          55       55              
  Lines        8536     8512      -24     
  Branches     8536     8512      -24     
==========================================
- Hits         8199     8175      -24     
  Misses        289      289              
  Partials       48       48              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ilyalesokhin-starkware]

Reviewed 4 of 4 files at r3, 4 of 4 files at r4.
Reviewable status: all files reviewed (commit messages unreviewed), all discussions resolved (waiting on @spapinistarkware)

[graphite-app]

Merge activity

• Mar 12, 7:53 AM EDT: spapinistarkware added this pull request to the Graphite merge queue.
• Mar 12, 7:59 AM EDT: spapinistarkware merged this pull request with the Graphite merge queue.

[ilyalesokhin-starkware]

src/core/backend/avx512/fft/ifft.rs line 369 at r3 (raw file):

    //  0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1
    let t = _mm512_set1_epi64(std::mem::transmute(twiddle3_dbl));
    // Apply the permutation, resulting in indexing a:bcid.

where is the permutation?

Code quote:

  // Apply the permutation, resulting in indexing a:bcid.

[spapinistarkware]

Reviewable status: all files reviewed (commit messages unreviewed), 1 unresolved discussion

---

src/core/backend/avx512/fft/ifft.rs line 369 at r3 (raw file):

Previously, ilyalesokhin-starkware wrote…

where is the permutation?

Look above, it's

// At each layer we apply the following permutation to the index:
//   i:abcd => d:iabc

[ilyalesokhin-starkware]

src/core/backend/avx512/fft/ifft.rs line 336 at r5 (raw file):

    // We need to permute the 512-bit registers to get the right order for the butterflies.
    // Denote the index of the 16 M31 elements in register i as i:abcd.
    // At each layer we apply the following permutation to the index:

Suggestion:

 // At each layer, we interleave the registers which applies the following permutation to the index: