Save twiddles

[spapinistarkware]

This change is 

[spapinistarkware]

• ColumnOps #459
• Use precomputed twiddles in avx #443
• Precompute twiddles #401
• Remove FRI PolyOps dependency #399
• avx slow eval #398
• avx poly extend #397
• AVX fft eval on larger domains #396
• Reverse order of evaluations #395
• Incorporate fft in AVX backend #394
• AVX regular fft #390
• Separate ifft #389
• Support ifft for itnermediate sizes #383
• AVX fft benchmark #382
• ifft_full #381
• Save twiddles #380 👈
• ifft_lower #379
• avx ifft3 #378
• avx vecwise butterflies #377
• avx vecwise ibutterfly #368
• avx ibutterfly #367
• avx butterfly #366
• avx poly #365
• 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]

Reviewable status: 0 of 1 files reviewed, 1 unresolved discussion (waiting on @spapinistarkware)

---

src/core/backend/avx512/fft.rs line 276 at r2 (raw file):

/// # Safety
/// This function is safe.
pub unsafe fn vecwise_ibutterflies(

this is only used on the first 4 layers? you can't use it in the middle of the FFT?

Code quote:

vecwise_ibutterflies

[spapinistarkware]

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

---

src/core/backend/avx512/fft.rs line 276 at r2 (raw file):

Previously, ilyalesokhin-starkware wrote…

this is only used on the first 4 layers? you can't use it in the middle of the FFT?

In this implementation, no.
It's good for ffts up to 2^28 elements, which I think is good for now.
igher than that, we would also need to transpose inside the vec, which I currently don't do.

[ilyalesokhin-starkware]

Reviewed 1 of 1 files at r3, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @spapinistarkware)

[ilyalesokhin-starkware]

src/core/backend/avx512/fft.rs line 274 at r4 (raw file):

    };
    let t = _mm512_permutexvar_epi32(INDICES_FROM_T1, t);
    let t = _mm512_xor_epi32(t, NEGATION_MASK);

what is this?

Code quote:

    const INDICES_FROM_T1: __m512i = unsafe {
        core::mem::transmute([
            0b0001, 0b0001, 0b0000, 0b0000, 0b0011, 0b0011, 0b0010, 0b0010, 0b0101, 0b0101, 0b0100,
            0b0100, 0b0111, 0b0111, 0b0110, 0b0110,
        ])
    };
    const NEGATION_MASK: __m512i = unsafe {
        core::mem::transmute([0i32, -2, -2, 0, 0, -2, -2, 0, 0, -2, -2, 0, 0, -2, -2, 0])
    };
    let t = _mm512_permutexvar_epi32(INDICES_FROM_T1, t);
    let t = _mm512_xor_epi32(t, NEGATION_MASK);

[spapinistarkware]

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

---

src/core/backend/avx512/fft.rs line 274 at r4 (raw file):

Previously, ilyalesokhin-starkware wrote…

what is this?

// See the comments in its body for more info.

[ilyalesokhin-starkware]

Reviewed all commit messages.
Reviewable status: 0 of 1 files reviewed, 2 unresolved discussions (waiting on @spapinistarkware)

---

src/core/backend/avx512/fft.rs line 328 at r5 (raw file):

    };
    let t = _mm512_permutexvar_epi32(INDICES_FROM_T1, t);
    let t = _mm512_xor_epi32(t, NEGATION_MASK);

can you move this to a function?

compute_layer0_twiddles or something like that?

Code quote:

    const INDICES_FROM_T1: __m512i = unsafe {
        core::mem::transmute([
            0b0001, 0b0001, 0b0000, 0b0000, 0b0011, 0b0011, 0b0010, 0b0010, 0b0101, 0b0101, 0b0100,
            0b0100, 0b0111, 0b0111, 0b0110, 0b0110,
        ])
    };
    const NEGATION_MASK: __m512i = unsafe {
        core::mem::transmute([0i32, -2, -2, 0, 0, -2, -2, 0, 0, -2, -2, 0, 0, -2, -2, 0])
    };
    let t = _mm512_permutexvar_epi32(INDICES_FROM_T1, t);
    let t = _mm512_xor_epi32(t, NEGATION_MASK);

[spapinistarkware]

Reviewable status: 0 of 1 files reviewed, 2 unresolved discussions (waiting on @ilyalesokhin-starkware)

---

src/core/backend/avx512/fft.rs line 328 at r5 (raw file):

Previously, ilyalesokhin-starkware wrote…

can you move this to a function?

compute_layer0_twiddles or something like that?

Done.

[ilyalesokhin-starkware]

Reviewed 1 of 1 files at r6, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @spapinistarkware)

[graphite-app]

Merge activity

• Mar 4, 6:34 AM EST: spapinistarkware added this pull request to the Graphite merge queue.
• Mar 4, 6:38 AM EST: spapinistarkware merged this pull request with the Graphite merge queue.