feat(starknet_os): inline ark_poly bit_reversal algorithm

[dorimedini-starkware]

feat(starknet_os): inline ark_poly bit_reversal algorithm

feat(starknet_os): refactor bit_reversal for our needs

Signed-off-by: Dori Medini dori@starkware.co

[reviewable-StarkWare]

This change is 

[github-actions]

Artifacts upload workflows:

• Started at 2025-03-20T10:17:24Z
• Started at 2025-03-20T10:28:33Z
• Started at 2025-03-20T10:35:29Z
• Started at 2025-03-20T11:39:55Z
• Started at 2025-03-20T14:50:01Z
• Started at 2025-03-23T09:38:41Z
• Started at 2025-03-23T11:14:59Z
• Started at 2025-03-23T15:57:18Z
• Started at 2025-03-23T16:29:04Z
• Started at 2025-03-24T14:01:36Z
• Started at 2025-03-25T09:49:22Z

[dorimedini-starkware]

• feat(starknet_os): implement write_split_result #5163
• chore(starknet_os): move write_split_result #5162
• feat(starknet_os): integrate the split function #5161
• feat(starknet_os): convert BigUints from decimal strings only once #5127
• feat(starknet_os): use ark_poly FFT implementation #4985
• feat(starknet_os): inline ark_poly bit_reversal algorithm #5110 👈 (View in Graphite)
• test(starknet_os): fft regression #4984
• feat(starknet_os): use BigUint instead of BigInt in KZG area #4922
• main

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

[aner-starkware]

crates/starknet_os/src/hints/hint_implementation/kzg/utils.rs line 176 at r6 (raw file):

        }
        r
    }

What's going on here? (I mean, bit manipulation is hard to follow, so maybe consider an explanation\reference on why it works?)

Code quote:

    fn bitreverse(mut n: usize) -> usize {
        let mut r = 0;
        for _ in 0..WIDTH {
            r = (r << 1) | (n & 1);
            n >>= 1;
        }
        r
    }

[aner-starkware]

Reviewed 1 of 3 files at r1, 1 of 2 files at r3, 2 of 2 files at r6, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @dorimedini-starkware and @Yoni-Starkware)

[aner-starkware]

Reviewed 2 of 2 files at r7, 2 of 2 files at r8, all commit messages.
Reviewable status: all files reviewed, 2 unresolved discussions (waiting on @Yoni-Starkware)

---

crates/starknet_os/src/hints/hint_implementation/kzg/utils.rs line 169 at r8 (raw file):

    }

    /// Reverses the bits of `n`, assuming it has a width (in bits) of `WIDTH`.

Maybe?

Suggestion:

; assumes `n` is at most `WIDTH` bits.

---

crates/starknet_os/src/hints/hint_implementation/kzg/utils.rs line 184 at r8 (raw file):

        // As bitreverse(bitreverse(i)) = i, by swapping only when i < reversed_i we avoid swapping
        // the same element twice. When the loop terminates, each index will have swapped locations
        // with it's reversed, as required.

It's a bit verbose; consider shortening.

Code quote:

        // As bitreverse(bitreverse(i)) = i, by swapping only when i < reversed_i we avoid swapping
        // the same element twice. When the loop terminates, each index will have swapped locations
        // with it's reversed, as required.

[dorimedini-starkware]

Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @aner-starkware and @Yoni-Starkware)

---

crates/starknet_os/src/hints/hint_implementation/kzg/utils.rs line 169 at r8 (raw file):

Previously, aner-starkware wrote…

Maybe?

no, this is not an upper bound; if n=7 then bitreverse(n, 3) == 7 but bitreverse(n, 4) == 14. n can be "smaller than the width".

---

crates/starknet_os/src/hints/hint_implementation/kzg/utils.rs line 184 at r8 (raw file):

Previously, aner-starkware wrote…

It's a bit verbose; consider shortening.

what would you suggest?

[aner-starkware]

Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @dorimedini-starkware and @Yoni-Starkware)

---

crates/starknet_os/src/hints/hint_implementation/kzg/utils.rs line 169 at r8 (raw file):

Previously, dorimedini-starkware wrote…

no, this is not an upper bound; if n=7 then bitreverse(n, 3) == 7 but bitreverse(n, 4) == 14. n can be "smaller than the width".

Ah, right, so maybe "where n is represented by WIDTH bits". Is WIDTH the best name, btw?

---

crates/starknet_os/src/hints/hint_implementation/kzg/utils.rs line 184 at r8 (raw file):

Previously, dorimedini-starkware wrote…

what would you suggest?

Maybe "Applies the bit-reversal permutation on all elements. Swaps only when i < reversed_i to avoid swapping the same element twice."

[aner-starkware]

Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @dorimedini-starkware and @Yoni-Starkware)

[dorimedini-starkware]

Reviewable status: 2 of 4 files reviewed, all discussions resolved (waiting on @aner-starkware and @Yoni-Starkware)

---

config/sequencer/presets/consolidated_node/application_configs/node.json line 306 at r9 (raw file):

  "versioned_constants_overrides.max_recursion_depth": 50,
  "versioned_constants_overrides.validate_max_n_steps": 1000000
}

I have no idea why the latest gt sync cause so many conflicts, but this JSON was missing a trailing newline... unrelated to my PR but no damage

[aner-starkware]

Reviewed 2 of 2 files at r9.
Reviewable status: all files reviewed (commit messages unreviewed), 1 unresolved discussion (waiting on @Yoni-Starkware)

---

-- commits line 10 at r9:
What's this?

Code quote:

New commits in r9 on 3/25/2025 at 11:48 AM by rotem-starkware:
- dabe98d: feat(starknet_os): implement load deprecated class inner hint (#5108)

  * feat(starknet_os): copy get_deprecated_contract_class_struct from moonsong's repo

  * feat(starknet_os): implement LoadCairoObject trait for DeprecatedContractClass

  * feat(starknet_os): copy load_deprecated_class_inner hint implementation from moonsong's repo

  * feat(starknet_os): implement load deprecated class inner hint

[dorimedini-starkware]

Reviewable status: all files reviewed (commit messages unreviewed), 1 unresolved discussion (waiting on @aner-starkware and @Yoni-Starkware)

---

-- commits line 10 at r9:

Previously, aner-starkware wrote…

What's this?

gt sync issue... commits will be squashed anyway, as long as the diff is fine it's fine.
not sure what happened here

[dorimedini-starkware]

Reviewable status: all files reviewed (commit messages unreviewed), 1 unresolved discussion (waiting on @aner-starkware and @Yoni-Starkware)

---

-- commits line 10 at r9:

Previously, dorimedini-starkware wrote…

gt sync issue... commits will be squashed anyway, as long as the diff is fine it's fine.
not sure what happened here

the commit message will be the PR title

[aner-starkware]

Reviewed all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @Yoni-Starkware)