perf/inline blake3 round by gabriel-barrett · Pull Request #557 · argumentcomputer/ix

gabriel-barrett · 2026-08-13T15:47:07Z

Inline blake3_compress_inner_j. This has the effect of increasing the total width, though not at the same proportion as what we gain on efficiency, for the lookup itself is costly, especially when the input arguments are big.

Eventually we will have to decrease the proof size of the recursive verifier, but there are ways around it see #551

gabriel-barrett · 2026-08-14T15:52:31Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T15:52:48Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T15:56:36Z

`!benchmark` — main vs `69ce736`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `4e89546` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	35.764 s	31.074 s	-13.1% (1.15× faster) 🟢	107.68 GiB	81.50 GiB	-24.3% (1.32× smaller) 🟢	5.30 MiB	6.27 MiB	+18.3% (1.18× larger) ⚠️	33.1 ms	38.1 ms	+15.3% (1.15× slower) ⚠️	5.961 s	4.727 s	-20.7% (1.26× faster) 🟢	180.38B	143.49B	-20.5% (1.26× fewer) 🟢	15.713 s	14.745 s	-6.2% (1.07× faster) 🟢	11.21 MiB	12.14 MiB	+8.3% (1.08× larger) ⚠️	69.0 ms	75.7 ms	+9.7% (1.10× slower) ⚠️	32.74 GiB	29.83 GiB	-8.9% (1.10× smaller) 🟢
`Vector.append`	32.628 s	27.362 s	-16.1% (1.19× faster) 🟢	95.00 GiB	73.59 GiB	-22.5% (1.29× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	32.4 ms	39.9 ms	+23.0% (1.23× slower) ⚠️	5.158 s	4.255 s	-17.5% (1.21× faster) 🟢	157.02B	128.00B	-18.5% (1.23× fewer) 🟢	3.885 s	3.795 s	-2.3%	10.42 MiB	11.34 MiB	+8.9% (1.09× larger) ⚠️	65.7 ms	69.1 ms	+5.1% (1.05× slower) ⚠️	6.80 GiB	6.89 GiB	+1.4%
`Nat.add_comm`	21.106 s	17.709 s	-16.1% (1.19× faster) 🟢	58.08 GiB	44.38 GiB	-23.6% (1.31× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	34.6 ms	38.9 ms	+12.3% (1.12× slower) ⚠️	3.752 s	3.105 s	-17.2% (1.21× faster) 🟢	113.85B	92.75B	-18.5% (1.23× fewer) 🟢	1.024 s	1.034 s	+1.0%	8.93 MiB	9.86 MiB	+10.4% (1.10× larger) ⚠️	52.6 ms	60.8 ms	+15.7% (1.16× slower) ⚠️	3.79 GiB	4.06 GiB	+7.0% (1.07× larger) ⚠️

Workflow logs

argument-ci-bot · 2026-08-14T15:56:40Z

`!benchmark` — main vs `69ce736`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `4e89546` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	39.589 s	38.272 s	-3.3% 🟢	38.950	40.290	+3.4% 🟢	73.66 GiB	70.70 GiB	-4.0% 🟢	9.861 s	9.625 s	-2.4%	131.2 ms	146.6 ms	+11.7% (1.12× slower) ⚠️	21.59 MiB	23.38 MiB	+8.3% (1.08× larger) ⚠️	136.02B	130.81B	-3.8% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	38.143 s	36.140 s	-5.2% (1.06× faster) 🟢	70.550	74.460	+5.5% (1.06× faster) 🟢	73.09 GiB	67.14 GiB	-8.1% (1.09× smaller) 🟢	9.759 s	9.374 s	-3.9% 🟢	136.4 ms	145.1 ms	+6.4% (1.06× slower) ⚠️	21.69 MiB	23.47 MiB	+8.2% (1.08× larger) ⚠️	138.80B	126.89B	-8.6% (1.09× fewer) 🟢
`Char.ofOrdinal_le_of_le`	29.836 s	29.502 s	-1.1%	88.690	89.690	+1.1%	58.53 GiB	58.49 GiB	-0.1%	6.903 s	6.630 s	-4.0% 🟢	137.5 ms	201.3 ms	+46.4% (1.46× slower) ⚠️	21.64 MiB	23.42 MiB	+8.3% (1.08× larger) ⚠️	99.42B	90.63B	-8.8% (1.10× fewer) 🟢
`Vector.extract_append._proof_2`	21.860 s	21.323 s	-2.5%	59.560	61.060	+2.5%	39.70 GiB	38.22 GiB	-3.7% 🟢	5.365 s	5.263 s	-1.9%	130.4 ms	148.2 ms	+13.7% (1.14× slower) ⚠️	21.30 MiB	23.08 MiB	+8.4% (1.08× larger) ⚠️	77.08B	73.31B	-4.9% (1.05× fewer) 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.638 s	16.652 s	-5.6% (1.06× faster) 🟢	102.740	108.810	+5.9% (1.06× faster) 🟢	34.77 GiB	31.77 GiB	-8.6% (1.09× smaller) 🟢	3.634 s	3.537 s	-2.7%	138.0 ms	149.8 ms	+8.5% (1.09× slower) ⚠️	21.47 MiB	23.26 MiB	+8.3% (1.08× larger) ⚠️	54.85B	49.57B	-9.6% (1.11× fewer) 🟢
`String.split`	16.598 s	15.578 s	-6.1% (1.07× faster) 🟢	106.340	113.300	+6.5% (1.07× faster) 🟢	32.71 GiB	29.74 GiB	-9.1% (1.10× smaller) 🟢	3.376 s	3.252 s	-3.7% 🟢	133.0 ms	141.9 ms	+6.7% (1.07× slower) ⚠️	21.68 MiB	23.46 MiB	+8.2% (1.08× larger) ⚠️	50.25B	45.59B	-9.3% (1.10× fewer) 🟢
`List.mergeSort`	12.202 s	11.714 s	-4.0% 🟢	118.670	123.610	+4.2% 🟢	23.77 GiB	22.24 GiB	-6.4% (1.07× smaller) 🟢	2.438 s	2.333 s	-4.3% 🟢	129.5 ms	147.4 ms	+13.7% (1.14× slower) ⚠️	21.53 MiB	23.32 MiB	+8.3% (1.08× larger) ⚠️	36.10B	32.44B	-10.1% (1.11× fewer) 🟢
`Vector.append`	3.968 s	3.886 s	-2.1%	122.230	124.810	+2.1%	6.69 GiB	6.60 GiB	-1.3%	684.6 ms	653.3 ms	-4.6% 🟢	132.6 ms	134.9 ms	+1.7%	20.13 MiB	21.92 MiB	+8.9% (1.09× larger) ⚠️	8.02B	7.24B	-9.7% (1.11× fewer) 🟢
`Nat.gcd_comm`	3.235 s	3.104 s	-4.1% 🟢	120.550	125.640	+4.2% 🟢	5.07 GiB	5.78 GiB	+14.0% (1.14× larger) ⚠️	524.7 ms	507.6 ms	-3.3% 🟢	123.7 ms	133.3 ms	+7.8% (1.08× slower) ⚠️	19.90 MiB	21.68 MiB	+9.0% (1.09× larger) ⚠️	5.25B	4.60B	-12.4% (1.14× fewer) 🟢
`String.append`	2.399 s	2.299 s	-4.2% 🟢	126.740	132.250	+4.3% 🟢	4.62 GiB	4.64 GiB	+0.3%	403.9 ms	393.2 ms	-2.7%	123.5 ms	127.6 ms	+3.3% ⚠️	19.17 MiB	20.96 MiB	+9.3% (1.09× larger) ⚠️	2.90B	2.55B	-12.1% (1.14× fewer) 🟢
`Int.gcd`	1.991 s	2.001 s	+0.5%	104.480	103.950	-0.5%	4.50 GiB	4.53 GiB	+0.7%	355.0 ms	343.3 ms	-3.3% 🟢	118.2 ms	129.5 ms	+9.5% (1.10× slower) ⚠️	18.73 MiB	20.51 MiB	+9.5% (1.10× larger) ⚠️	1.87B	1.64B	-11.9% (1.14× fewer) 🟢
`Nat.sub_le_of_le_add`	1.815 s	1.788 s	-1.5%	93.650	95.090	+1.5%	4.60 GiB	4.62 GiB	+0.3%	343.9 ms	333.7 ms	-2.9%	113.8 ms	128.4 ms	+12.8% (1.13× slower) ⚠️	19.08 MiB	20.87 MiB	+9.4% (1.09× larger) ⚠️	1.60B	1.42B	-11.5% (1.13× fewer) 🟢
`Nat.add_comm`	1.011 s	1.050 s	+3.9% ⚠️	41.550	40	-3.7% ⚠️	3.54 GiB	4.17 GiB	+17.7% (1.18× larger) ⚠️	255.5 ms	251.9 ms	-1.4%	106.7 ms	116.6 ms	+9.3% (1.09× slower) ⚠️	17.25 MiB	19.04 MiB	+10.4% (1.10× larger) ⚠️	271.37M	249.57M	-8.0% (1.09× fewer) 🟢

Workflow logs

Rebased onto the fused xor-rotation branch: the seven inline calls now compose with the two-operand u32_xor_rotr7/12 forms. Codegen regenerated and all 71 kernel FFT pins + the shard aggregate re-measured on top of the new base (median -5.4%, range -11.4%..-0.2%; shard 6.08B -> 5.21B, -14.4%). aiur (cargo + prove + cross), ixvm (pins + parity), multi-stark and recursive-verifier suites pass; fmt, clippy, deny clean.

gabriel-barrett · 2026-08-14T17:51:01Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T17:51:11Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T17:55:05Z

`!benchmark` — main vs `b6a4f22`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	34.284 s	31.716 s	-7.5% (1.08× faster) 🟢	102.25 GiB	81.04 GiB	-20.7% (1.26× smaller) 🟢	5.27 MiB	6.27 MiB	+18.9% (1.19× larger) ⚠️	34.7 ms	37.9 ms	+9.2% (1.09× slower) ⚠️	5.645 s	4.839 s	-14.3% (1.17× faster) 🟢	173.38B	143.49B	-17.2% (1.21× fewer) 🟢	15.805 s	15.068 s	-4.7% 🟢	11.19 MiB	12.14 MiB	+8.5% (1.09× larger) ⚠️	68.0 ms	75.8 ms	+11.6% (1.12× slower) ⚠️	32.25 GiB	29.75 GiB	-7.7% (1.08× smaller) 🟢
`Vector.append`	31.433 s	28.063 s	-10.7% (1.12× faster) 🟢	90.02 GiB	73.45 GiB	-18.4% (1.23× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	32.5 ms	37.4 ms	+15.3% (1.15× slower) ⚠️	4.991 s	4.283 s	-14.2% (1.17× faster) 🟢	152.01B	128.00B	-15.8% (1.19× fewer) 🟢	3.853 s	3.825 s	-0.7%	10.39 MiB	11.34 MiB	+9.2% (1.09× larger) ⚠️	65.2 ms	68.8 ms	+5.5% (1.06× slower) ⚠️	7.23 GiB	6.34 GiB	-12.3% (1.14× smaller) 🟢
`Nat.add_comm`	20.507 s	17.985 s	-12.3% (1.14× faster) 🟢	55.58 GiB	44.60 GiB	-19.8% (1.25× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	34.9 ms	38.4 ms	+10.2% (1.10× slower) ⚠️	3.591 s	3.070 s	-14.5% (1.17× faster) 🟢	107.43B	92.75B	-13.7% (1.16× fewer) 🟢	1.034 s	1.031 s	-0.4%	8.90 MiB	9.86 MiB	+10.7% (1.11× larger) ⚠️	53.3 ms	61.2 ms	+14.9% (1.15× slower) ⚠️	3.98 GiB	3.56 GiB	-10.7% (1.12× smaller) 🟢

Workflow logs

argument-ci-bot · 2026-08-14T17:55:08Z

`!benchmark` — main vs `b6a4f22`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	37.550 s	37.412 s	-0.4%	41.070	41.220	+0.4%	73.20 GiB	70.64 GiB	-3.5% 🟢	9.577 s	9.391 s	-1.9%	138.9 ms	151.1 ms	+8.8% (1.09× slower) ⚠️	21.54 MiB	23.38 MiB	+8.5% (1.09× larger) ⚠️	135.19B	130.81B	-3.2% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	37.157 s	34.546 s	-7.0% (1.08× faster) 🟢	72.420	77.900	+7.6% (1.08× faster) 🟢	72.13 GiB	67.22 GiB	-6.8% (1.07× smaller) 🟢	9.613 s	9.122 s	-5.1% (1.05× faster) 🟢	197.0 ms	148.7 ms	-24.5% (1.33× faster) 🟢	21.63 MiB	23.47 MiB	+8.5% (1.09× larger) ⚠️	136.87B	126.89B	-7.3% (1.08× fewer) 🟢
`Char.ofOrdinal_le_of_le`	28.515 s	28.454 s	-0.2%	92.790	92.990	+0.2%	58.03 GiB	58.51 GiB	+0.8%	6.602 s	6.421 s	-2.8%	131.8 ms	150.6 ms	+14.2% (1.14× slower) ⚠️	21.58 MiB	23.42 MiB	+8.5% (1.09× larger) ⚠️	98.00B	90.63B	-7.5% (1.08× fewer) 🟢
`Vector.extract_append._proof_2`	21.362 s	20.460 s	-4.2% 🟢	60.950	63.640	+4.4% 🟢	39.42 GiB	38.21 GiB	-3.1% 🟢	5.182 s	5.110 s	-1.4%	128.2 ms	152.2 ms	+18.7% (1.19× slower) ⚠️	21.24 MiB	23.08 MiB	+8.7% (1.09× larger) ⚠️	76.48B	73.31B	-4.1% 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.040 s	15.965 s	-6.3% (1.07× faster) 🟢	106.340	113.500	+6.7% (1.07× faster) 🟢	34.22 GiB	31.82 GiB	-7.0% (1.08× smaller) 🟢	3.560 s	3.361 s	-5.6% (1.06× faster) 🟢	134.9 ms	143.8 ms	+6.6% (1.07× slower) ⚠️	21.42 MiB	23.26 MiB	+8.6% (1.09× larger) ⚠️	54.00B	49.57B	-8.2% (1.09× fewer) 🟢
`String.split`	16.080 s	15.024 s	-6.6% (1.07× faster) 🟢	109.770	117.480	+7.0% (1.07× faster) 🟢	32.26 GiB	29.72 GiB	-7.9% (1.09× smaller) 🟢	3.243 s	3.158 s	-2.6%	132.5 ms	141.8 ms	+7.0% (1.07× slower) ⚠️	21.62 MiB	23.46 MiB	+8.5% (1.09× larger) ⚠️	49.51B	45.59B	-7.9% (1.09× fewer) 🟢
`List.mergeSort`	11.944 s	11.350 s	-5.0% (1.05× faster) 🟢	121.230	127.570	+5.2% (1.05× faster) 🟢	23.52 GiB	22.26 GiB	-5.4% (1.06× smaller) 🟢	2.435 s	2.308 s	-5.2% (1.05× faster) 🟢	133.7 ms	142.9 ms	+6.9% (1.07× slower) ⚠️	21.48 MiB	23.32 MiB	+8.6% (1.09× larger) ⚠️	35.52B	32.44B	-8.7% (1.09× fewer) 🟢
`Vector.append`	3.982 s	3.778 s	-5.1% (1.05× faster) 🟢	121.780	128.380	+5.4% (1.05× faster) 🟢	7.11 GiB	6.06 GiB	-14.8% (1.17× smaller) 🟢	666.1 ms	647.1 ms	-2.8%	135.2 ms	131.0 ms	-3.1% 🟢	20.08 MiB	21.92 MiB	+9.2% (1.09× larger) ⚠️	7.90B	7.24B	-8.3% (1.09× fewer) 🟢
`Nat.gcd_comm`	3.194 s	3.051 s	-4.5% 🟢	122.110	127.830	+4.7% 🟢	4.92 GiB	4.81 GiB	-2.2%	529.7 ms	504.2 ms	-4.8% (1.05× faster) 🟢	122.8 ms	131.9 ms	+7.4% (1.07× slower) ⚠️	19.84 MiB	21.68 MiB	+9.3% (1.09× larger) ⚠️	5.15B	4.60B	-10.7% (1.12× fewer) 🟢
`String.append`	2.339 s	2.245 s	-4.0% 🟢	129.970	135.390	+4.2% 🟢	4.63 GiB	4.79 GiB	+3.5% ⚠️	400.3 ms	382.6 ms	-4.4% 🟢	120.4 ms	127.3 ms	+5.7% (1.06× slower) ⚠️	19.12 MiB	20.96 MiB	+9.6% (1.10× larger) ⚠️	2.85B	2.55B	-10.5% (1.12× fewer) 🟢
`Int.gcd`	1.991 s	1.956 s	-1.8%	104.490	106.350	+1.8%	5.40 GiB	4.53 GiB	-16.2% (1.19× smaller) 🟢	342.9 ms	340.7 ms	-0.6%	113.8 ms	127.5 ms	+12.0% (1.12× slower) ⚠️	18.67 MiB	20.51 MiB	+9.9% (1.10× larger) ⚠️	1.83B	1.64B	-10.3% (1.12× fewer) 🟢
`Nat.sub_le_of_le_add`	1.788 s	1.741 s	-2.7%	95.060	97.660	+2.7%	4.42 GiB	4.61 GiB	+4.3% ⚠️	337.3 ms	333.3 ms	-1.2%	113.0 ms	128.9 ms	+14.1% (1.14× slower) ⚠️	19.03 MiB	20.87 MiB	+9.7% (1.10× larger) ⚠️	1.58B	1.42B	-10.0% (1.11× fewer) 🟢
`Nat.add_comm`	1.019 s	1.084 s	+6.3% (1.06× slower) ⚠️	41.200	38.740	-6.0% (1.06× slower) ⚠️	4.17 GiB	3.49 GiB	-16.4% (1.20× smaller) 🟢	245.4 ms	246.5 ms	+0.4%	107.8 ms	115.9 ms	+7.4% (1.07× slower) ⚠️	17.20 MiB	19.04 MiB	+10.7% (1.11× larger) ⚠️	268.28M	249.57M	-7.0% (1.07× fewer) 🟢

Workflow logs

gabriel-barrett · 2026-08-13T15:47:07Z

Inline blake3_compress_inner_j. This has the effect of increasing the total width, though not at the same proportion as what we gain on efficiency, for the lookup itself is costly, especially when the input arguments are big.

Eventually we will have to decrease the proof size of the recursive verifier, but there are ways around it see #551

gabriel-barrett · 2026-08-14T15:52:31Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T15:52:48Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T15:56:36Z

`!benchmark` — main vs `69ce736`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `4e89546` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	35.764 s	31.074 s	-13.1% (1.15× faster) 🟢	107.68 GiB	81.50 GiB	-24.3% (1.32× smaller) 🟢	5.30 MiB	6.27 MiB	+18.3% (1.18× larger) ⚠️	33.1 ms	38.1 ms	+15.3% (1.15× slower) ⚠️	5.961 s	4.727 s	-20.7% (1.26× faster) 🟢	180.38B	143.49B	-20.5% (1.26× fewer) 🟢	15.713 s	14.745 s	-6.2% (1.07× faster) 🟢	11.21 MiB	12.14 MiB	+8.3% (1.08× larger) ⚠️	69.0 ms	75.7 ms	+9.7% (1.10× slower) ⚠️	32.74 GiB	29.83 GiB	-8.9% (1.10× smaller) 🟢
`Vector.append`	32.628 s	27.362 s	-16.1% (1.19× faster) 🟢	95.00 GiB	73.59 GiB	-22.5% (1.29× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	32.4 ms	39.9 ms	+23.0% (1.23× slower) ⚠️	5.158 s	4.255 s	-17.5% (1.21× faster) 🟢	157.02B	128.00B	-18.5% (1.23× fewer) 🟢	3.885 s	3.795 s	-2.3%	10.42 MiB	11.34 MiB	+8.9% (1.09× larger) ⚠️	65.7 ms	69.1 ms	+5.1% (1.05× slower) ⚠️	6.80 GiB	6.89 GiB	+1.4%
`Nat.add_comm`	21.106 s	17.709 s	-16.1% (1.19× faster) 🟢	58.08 GiB	44.38 GiB	-23.6% (1.31× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	34.6 ms	38.9 ms	+12.3% (1.12× slower) ⚠️	3.752 s	3.105 s	-17.2% (1.21× faster) 🟢	113.85B	92.75B	-18.5% (1.23× fewer) 🟢	1.024 s	1.034 s	+1.0%	8.93 MiB	9.86 MiB	+10.4% (1.10× larger) ⚠️	52.6 ms	60.8 ms	+15.7% (1.16× slower) ⚠️	3.79 GiB	4.06 GiB	+7.0% (1.07× larger) ⚠️

Workflow logs

argument-ci-bot · 2026-08-14T15:56:40Z

`!benchmark` — main vs `69ce736`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `4e89546` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	39.589 s	38.272 s	-3.3% 🟢	38.950	40.290	+3.4% 🟢	73.66 GiB	70.70 GiB	-4.0% 🟢	9.861 s	9.625 s	-2.4%	131.2 ms	146.6 ms	+11.7% (1.12× slower) ⚠️	21.59 MiB	23.38 MiB	+8.3% (1.08× larger) ⚠️	136.02B	130.81B	-3.8% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	38.143 s	36.140 s	-5.2% (1.06× faster) 🟢	70.550	74.460	+5.5% (1.06× faster) 🟢	73.09 GiB	67.14 GiB	-8.1% (1.09× smaller) 🟢	9.759 s	9.374 s	-3.9% 🟢	136.4 ms	145.1 ms	+6.4% (1.06× slower) ⚠️	21.69 MiB	23.47 MiB	+8.2% (1.08× larger) ⚠️	138.80B	126.89B	-8.6% (1.09× fewer) 🟢
`Char.ofOrdinal_le_of_le`	29.836 s	29.502 s	-1.1%	88.690	89.690	+1.1%	58.53 GiB	58.49 GiB	-0.1%	6.903 s	6.630 s	-4.0% 🟢	137.5 ms	201.3 ms	+46.4% (1.46× slower) ⚠️	21.64 MiB	23.42 MiB	+8.3% (1.08× larger) ⚠️	99.42B	90.63B	-8.8% (1.10× fewer) 🟢
`Vector.extract_append._proof_2`	21.860 s	21.323 s	-2.5%	59.560	61.060	+2.5%	39.70 GiB	38.22 GiB	-3.7% 🟢	5.365 s	5.263 s	-1.9%	130.4 ms	148.2 ms	+13.7% (1.14× slower) ⚠️	21.30 MiB	23.08 MiB	+8.4% (1.08× larger) ⚠️	77.08B	73.31B	-4.9% (1.05× fewer) 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.638 s	16.652 s	-5.6% (1.06× faster) 🟢	102.740	108.810	+5.9% (1.06× faster) 🟢	34.77 GiB	31.77 GiB	-8.6% (1.09× smaller) 🟢	3.634 s	3.537 s	-2.7%	138.0 ms	149.8 ms	+8.5% (1.09× slower) ⚠️	21.47 MiB	23.26 MiB	+8.3% (1.08× larger) ⚠️	54.85B	49.57B	-9.6% (1.11× fewer) 🟢
`String.split`	16.598 s	15.578 s	-6.1% (1.07× faster) 🟢	106.340	113.300	+6.5% (1.07× faster) 🟢	32.71 GiB	29.74 GiB	-9.1% (1.10× smaller) 🟢	3.376 s	3.252 s	-3.7% 🟢	133.0 ms	141.9 ms	+6.7% (1.07× slower) ⚠️	21.68 MiB	23.46 MiB	+8.2% (1.08× larger) ⚠️	50.25B	45.59B	-9.3% (1.10× fewer) 🟢
`List.mergeSort`	12.202 s	11.714 s	-4.0% 🟢	118.670	123.610	+4.2% 🟢	23.77 GiB	22.24 GiB	-6.4% (1.07× smaller) 🟢	2.438 s	2.333 s	-4.3% 🟢	129.5 ms	147.4 ms	+13.7% (1.14× slower) ⚠️	21.53 MiB	23.32 MiB	+8.3% (1.08× larger) ⚠️	36.10B	32.44B	-10.1% (1.11× fewer) 🟢
`Vector.append`	3.968 s	3.886 s	-2.1%	122.230	124.810	+2.1%	6.69 GiB	6.60 GiB	-1.3%	684.6 ms	653.3 ms	-4.6% 🟢	132.6 ms	134.9 ms	+1.7%	20.13 MiB	21.92 MiB	+8.9% (1.09× larger) ⚠️	8.02B	7.24B	-9.7% (1.11× fewer) 🟢
`Nat.gcd_comm`	3.235 s	3.104 s	-4.1% 🟢	120.550	125.640	+4.2% 🟢	5.07 GiB	5.78 GiB	+14.0% (1.14× larger) ⚠️	524.7 ms	507.6 ms	-3.3% 🟢	123.7 ms	133.3 ms	+7.8% (1.08× slower) ⚠️	19.90 MiB	21.68 MiB	+9.0% (1.09× larger) ⚠️	5.25B	4.60B	-12.4% (1.14× fewer) 🟢
`String.append`	2.399 s	2.299 s	-4.2% 🟢	126.740	132.250	+4.3% 🟢	4.62 GiB	4.64 GiB	+0.3%	403.9 ms	393.2 ms	-2.7%	123.5 ms	127.6 ms	+3.3% ⚠️	19.17 MiB	20.96 MiB	+9.3% (1.09× larger) ⚠️	2.90B	2.55B	-12.1% (1.14× fewer) 🟢
`Int.gcd`	1.991 s	2.001 s	+0.5%	104.480	103.950	-0.5%	4.50 GiB	4.53 GiB	+0.7%	355.0 ms	343.3 ms	-3.3% 🟢	118.2 ms	129.5 ms	+9.5% (1.10× slower) ⚠️	18.73 MiB	20.51 MiB	+9.5% (1.10× larger) ⚠️	1.87B	1.64B	-11.9% (1.14× fewer) 🟢
`Nat.sub_le_of_le_add`	1.815 s	1.788 s	-1.5%	93.650	95.090	+1.5%	4.60 GiB	4.62 GiB	+0.3%	343.9 ms	333.7 ms	-2.9%	113.8 ms	128.4 ms	+12.8% (1.13× slower) ⚠️	19.08 MiB	20.87 MiB	+9.4% (1.09× larger) ⚠️	1.60B	1.42B	-11.5% (1.13× fewer) 🟢
`Nat.add_comm`	1.011 s	1.050 s	+3.9% ⚠️	41.550	40	-3.7% ⚠️	3.54 GiB	4.17 GiB	+17.7% (1.18× larger) ⚠️	255.5 ms	251.9 ms	-1.4%	106.7 ms	116.6 ms	+9.3% (1.09× slower) ⚠️	17.25 MiB	19.04 MiB	+10.4% (1.10× larger) ⚠️	271.37M	249.57M	-8.0% (1.09× fewer) 🟢

Workflow logs

Rebased onto the fused xor-rotation branch: the seven inline calls now compose with the two-operand u32_xor_rotr7/12 forms. Codegen regenerated and all 71 kernel FFT pins + the shard aggregate re-measured on top of the new base (median -5.4%, range -11.4%..-0.2%; shard 6.08B -> 5.21B, -14.4%). aiur (cargo + prove + cross), ixvm (pins + parity), multi-stark and recursive-verifier suites pass; fmt, clippy, deny clean.

gabriel-barrett · 2026-08-14T17:51:01Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T17:51:11Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T17:55:05Z

`!benchmark` — main vs `b6a4f22`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	34.284 s	31.716 s	-7.5% (1.08× faster) 🟢	102.25 GiB	81.04 GiB	-20.7% (1.26× smaller) 🟢	5.27 MiB	6.27 MiB	+18.9% (1.19× larger) ⚠️	34.7 ms	37.9 ms	+9.2% (1.09× slower) ⚠️	5.645 s	4.839 s	-14.3% (1.17× faster) 🟢	173.38B	143.49B	-17.2% (1.21× fewer) 🟢	15.805 s	15.068 s	-4.7% 🟢	11.19 MiB	12.14 MiB	+8.5% (1.09× larger) ⚠️	68.0 ms	75.8 ms	+11.6% (1.12× slower) ⚠️	32.25 GiB	29.75 GiB	-7.7% (1.08× smaller) 🟢
`Vector.append`	31.433 s	28.063 s	-10.7% (1.12× faster) 🟢	90.02 GiB	73.45 GiB	-18.4% (1.23× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	32.5 ms	37.4 ms	+15.3% (1.15× slower) ⚠️	4.991 s	4.283 s	-14.2% (1.17× faster) 🟢	152.01B	128.00B	-15.8% (1.19× fewer) 🟢	3.853 s	3.825 s	-0.7%	10.39 MiB	11.34 MiB	+9.2% (1.09× larger) ⚠️	65.2 ms	68.8 ms	+5.5% (1.06× slower) ⚠️	7.23 GiB	6.34 GiB	-12.3% (1.14× smaller) 🟢
`Nat.add_comm`	20.507 s	17.985 s	-12.3% (1.14× faster) 🟢	55.58 GiB	44.60 GiB	-19.8% (1.25× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	34.9 ms	38.4 ms	+10.2% (1.10× slower) ⚠️	3.591 s	3.070 s	-14.5% (1.17× faster) 🟢	107.43B	92.75B	-13.7% (1.16× fewer) 🟢	1.034 s	1.031 s	-0.4%	8.90 MiB	9.86 MiB	+10.7% (1.11× larger) ⚠️	53.3 ms	61.2 ms	+14.9% (1.15× slower) ⚠️	3.98 GiB	3.56 GiB	-10.7% (1.12× smaller) 🟢

Workflow logs

argument-ci-bot · 2026-08-14T17:55:08Z

`!benchmark` — main vs `b6a4f22`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	37.550 s	37.412 s	-0.4%	41.070	41.220	+0.4%	73.20 GiB	70.64 GiB	-3.5% 🟢	9.577 s	9.391 s	-1.9%	138.9 ms	151.1 ms	+8.8% (1.09× slower) ⚠️	21.54 MiB	23.38 MiB	+8.5% (1.09× larger) ⚠️	135.19B	130.81B	-3.2% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	37.157 s	34.546 s	-7.0% (1.08× faster) 🟢	72.420	77.900	+7.6% (1.08× faster) 🟢	72.13 GiB	67.22 GiB	-6.8% (1.07× smaller) 🟢	9.613 s	9.122 s	-5.1% (1.05× faster) 🟢	197.0 ms	148.7 ms	-24.5% (1.33× faster) 🟢	21.63 MiB	23.47 MiB	+8.5% (1.09× larger) ⚠️	136.87B	126.89B	-7.3% (1.08× fewer) 🟢
`Char.ofOrdinal_le_of_le`	28.515 s	28.454 s	-0.2%	92.790	92.990	+0.2%	58.03 GiB	58.51 GiB	+0.8%	6.602 s	6.421 s	-2.8%	131.8 ms	150.6 ms	+14.2% (1.14× slower) ⚠️	21.58 MiB	23.42 MiB	+8.5% (1.09× larger) ⚠️	98.00B	90.63B	-7.5% (1.08× fewer) 🟢
`Vector.extract_append._proof_2`	21.362 s	20.460 s	-4.2% 🟢	60.950	63.640	+4.4% 🟢	39.42 GiB	38.21 GiB	-3.1% 🟢	5.182 s	5.110 s	-1.4%	128.2 ms	152.2 ms	+18.7% (1.19× slower) ⚠️	21.24 MiB	23.08 MiB	+8.7% (1.09× larger) ⚠️	76.48B	73.31B	-4.1% 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.040 s	15.965 s	-6.3% (1.07× faster) 🟢	106.340	113.500	+6.7% (1.07× faster) 🟢	34.22 GiB	31.82 GiB	-7.0% (1.08× smaller) 🟢	3.560 s	3.361 s	-5.6% (1.06× faster) 🟢	134.9 ms	143.8 ms	+6.6% (1.07× slower) ⚠️	21.42 MiB	23.26 MiB	+8.6% (1.09× larger) ⚠️	54.00B	49.57B	-8.2% (1.09× fewer) 🟢
`String.split`	16.080 s	15.024 s	-6.6% (1.07× faster) 🟢	109.770	117.480	+7.0% (1.07× faster) 🟢	32.26 GiB	29.72 GiB	-7.9% (1.09× smaller) 🟢	3.243 s	3.158 s	-2.6%	132.5 ms	141.8 ms	+7.0% (1.07× slower) ⚠️	21.62 MiB	23.46 MiB	+8.5% (1.09× larger) ⚠️	49.51B	45.59B	-7.9% (1.09× fewer) 🟢
`List.mergeSort`	11.944 s	11.350 s	-5.0% (1.05× faster) 🟢	121.230	127.570	+5.2% (1.05× faster) 🟢	23.52 GiB	22.26 GiB	-5.4% (1.06× smaller) 🟢	2.435 s	2.308 s	-5.2% (1.05× faster) 🟢	133.7 ms	142.9 ms	+6.9% (1.07× slower) ⚠️	21.48 MiB	23.32 MiB	+8.6% (1.09× larger) ⚠️	35.52B	32.44B	-8.7% (1.09× fewer) 🟢
`Vector.append`	3.982 s	3.778 s	-5.1% (1.05× faster) 🟢	121.780	128.380	+5.4% (1.05× faster) 🟢	7.11 GiB	6.06 GiB	-14.8% (1.17× smaller) 🟢	666.1 ms	647.1 ms	-2.8%	135.2 ms	131.0 ms	-3.1% 🟢	20.08 MiB	21.92 MiB	+9.2% (1.09× larger) ⚠️	7.90B	7.24B	-8.3% (1.09× fewer) 🟢
`Nat.gcd_comm`	3.194 s	3.051 s	-4.5% 🟢	122.110	127.830	+4.7% 🟢	4.92 GiB	4.81 GiB	-2.2%	529.7 ms	504.2 ms	-4.8% (1.05× faster) 🟢	122.8 ms	131.9 ms	+7.4% (1.07× slower) ⚠️	19.84 MiB	21.68 MiB	+9.3% (1.09× larger) ⚠️	5.15B	4.60B	-10.7% (1.12× fewer) 🟢
`String.append`	2.339 s	2.245 s	-4.0% 🟢	129.970	135.390	+4.2% 🟢	4.63 GiB	4.79 GiB	+3.5% ⚠️	400.3 ms	382.6 ms	-4.4% 🟢	120.4 ms	127.3 ms	+5.7% (1.06× slower) ⚠️	19.12 MiB	20.96 MiB	+9.6% (1.10× larger) ⚠️	2.85B	2.55B	-10.5% (1.12× fewer) 🟢
`Int.gcd`	1.991 s	1.956 s	-1.8%	104.490	106.350	+1.8%	5.40 GiB	4.53 GiB	-16.2% (1.19× smaller) 🟢	342.9 ms	340.7 ms	-0.6%	113.8 ms	127.5 ms	+12.0% (1.12× slower) ⚠️	18.67 MiB	20.51 MiB	+9.9% (1.10× larger) ⚠️	1.83B	1.64B	-10.3% (1.12× fewer) 🟢
`Nat.sub_le_of_le_add`	1.788 s	1.741 s	-2.7%	95.060	97.660	+2.7%	4.42 GiB	4.61 GiB	+4.3% ⚠️	337.3 ms	333.3 ms	-1.2%	113.0 ms	128.9 ms	+14.1% (1.14× slower) ⚠️	19.03 MiB	20.87 MiB	+9.7% (1.10× larger) ⚠️	1.58B	1.42B	-10.0% (1.11× fewer) 🟢
`Nat.add_comm`	1.019 s	1.084 s	+6.3% (1.06× slower) ⚠️	41.200	38.740	-6.0% (1.06× slower) ⚠️	4.17 GiB	3.49 GiB	-16.4% (1.20× smaller) 🟢	245.4 ms	246.5 ms	+0.4%	107.8 ms	115.9 ms	+7.4% (1.07× slower) ⚠️	17.20 MiB	19.04 MiB	+10.7% (1.11× larger) ⚠️	268.28M	249.57M	-7.0% (1.07× fewer) 🟢

Workflow logs

gabriel-barrett · 2026-08-13T15:47:07Z

Inline blake3_compress_inner_j. This has the effect of increasing the total width, though not at the same proportion as what we gain on efficiency, for the lookup itself is costly, especially when the input arguments are big.

Eventually we will have to decrease the proof size of the recursive verifier, but there are ways around it see #551

gabriel-barrett · 2026-08-14T15:52:31Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T15:52:48Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T15:56:36Z

`!benchmark` — main vs `69ce736`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `4e89546` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	35.764 s	31.074 s	-13.1% (1.15× faster) 🟢	107.68 GiB	81.50 GiB	-24.3% (1.32× smaller) 🟢	5.30 MiB	6.27 MiB	+18.3% (1.18× larger) ⚠️	33.1 ms	38.1 ms	+15.3% (1.15× slower) ⚠️	5.961 s	4.727 s	-20.7% (1.26× faster) 🟢	180.38B	143.49B	-20.5% (1.26× fewer) 🟢	15.713 s	14.745 s	-6.2% (1.07× faster) 🟢	11.21 MiB	12.14 MiB	+8.3% (1.08× larger) ⚠️	69.0 ms	75.7 ms	+9.7% (1.10× slower) ⚠️	32.74 GiB	29.83 GiB	-8.9% (1.10× smaller) 🟢
`Vector.append`	32.628 s	27.362 s	-16.1% (1.19× faster) 🟢	95.00 GiB	73.59 GiB	-22.5% (1.29× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	32.4 ms	39.9 ms	+23.0% (1.23× slower) ⚠️	5.158 s	4.255 s	-17.5% (1.21× faster) 🟢	157.02B	128.00B	-18.5% (1.23× fewer) 🟢	3.885 s	3.795 s	-2.3%	10.42 MiB	11.34 MiB	+8.9% (1.09× larger) ⚠️	65.7 ms	69.1 ms	+5.1% (1.05× slower) ⚠️	6.80 GiB	6.89 GiB	+1.4%
`Nat.add_comm`	21.106 s	17.709 s	-16.1% (1.19× faster) 🟢	58.08 GiB	44.38 GiB	-23.6% (1.31× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	34.6 ms	38.9 ms	+12.3% (1.12× slower) ⚠️	3.752 s	3.105 s	-17.2% (1.21× faster) 🟢	113.85B	92.75B	-18.5% (1.23× fewer) 🟢	1.024 s	1.034 s	+1.0%	8.93 MiB	9.86 MiB	+10.4% (1.10× larger) ⚠️	52.6 ms	60.8 ms	+15.7% (1.16× slower) ⚠️	3.79 GiB	4.06 GiB	+7.0% (1.07× larger) ⚠️

Workflow logs

argument-ci-bot · 2026-08-14T15:56:40Z

`!benchmark` — main vs `69ce736`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `4e89546` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	39.589 s	38.272 s	-3.3% 🟢	38.950	40.290	+3.4% 🟢	73.66 GiB	70.70 GiB	-4.0% 🟢	9.861 s	9.625 s	-2.4%	131.2 ms	146.6 ms	+11.7% (1.12× slower) ⚠️	21.59 MiB	23.38 MiB	+8.3% (1.08× larger) ⚠️	136.02B	130.81B	-3.8% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	38.143 s	36.140 s	-5.2% (1.06× faster) 🟢	70.550	74.460	+5.5% (1.06× faster) 🟢	73.09 GiB	67.14 GiB	-8.1% (1.09× smaller) 🟢	9.759 s	9.374 s	-3.9% 🟢	136.4 ms	145.1 ms	+6.4% (1.06× slower) ⚠️	21.69 MiB	23.47 MiB	+8.2% (1.08× larger) ⚠️	138.80B	126.89B	-8.6% (1.09× fewer) 🟢
`Char.ofOrdinal_le_of_le`	29.836 s	29.502 s	-1.1%	88.690	89.690	+1.1%	58.53 GiB	58.49 GiB	-0.1%	6.903 s	6.630 s	-4.0% 🟢	137.5 ms	201.3 ms	+46.4% (1.46× slower) ⚠️	21.64 MiB	23.42 MiB	+8.3% (1.08× larger) ⚠️	99.42B	90.63B	-8.8% (1.10× fewer) 🟢
`Vector.extract_append._proof_2`	21.860 s	21.323 s	-2.5%	59.560	61.060	+2.5%	39.70 GiB	38.22 GiB	-3.7% 🟢	5.365 s	5.263 s	-1.9%	130.4 ms	148.2 ms	+13.7% (1.14× slower) ⚠️	21.30 MiB	23.08 MiB	+8.4% (1.08× larger) ⚠️	77.08B	73.31B	-4.9% (1.05× fewer) 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.638 s	16.652 s	-5.6% (1.06× faster) 🟢	102.740	108.810	+5.9% (1.06× faster) 🟢	34.77 GiB	31.77 GiB	-8.6% (1.09× smaller) 🟢	3.634 s	3.537 s	-2.7%	138.0 ms	149.8 ms	+8.5% (1.09× slower) ⚠️	21.47 MiB	23.26 MiB	+8.3% (1.08× larger) ⚠️	54.85B	49.57B	-9.6% (1.11× fewer) 🟢
`String.split`	16.598 s	15.578 s	-6.1% (1.07× faster) 🟢	106.340	113.300	+6.5% (1.07× faster) 🟢	32.71 GiB	29.74 GiB	-9.1% (1.10× smaller) 🟢	3.376 s	3.252 s	-3.7% 🟢	133.0 ms	141.9 ms	+6.7% (1.07× slower) ⚠️	21.68 MiB	23.46 MiB	+8.2% (1.08× larger) ⚠️	50.25B	45.59B	-9.3% (1.10× fewer) 🟢
`List.mergeSort`	12.202 s	11.714 s	-4.0% 🟢	118.670	123.610	+4.2% 🟢	23.77 GiB	22.24 GiB	-6.4% (1.07× smaller) 🟢	2.438 s	2.333 s	-4.3% 🟢	129.5 ms	147.4 ms	+13.7% (1.14× slower) ⚠️	21.53 MiB	23.32 MiB	+8.3% (1.08× larger) ⚠️	36.10B	32.44B	-10.1% (1.11× fewer) 🟢
`Vector.append`	3.968 s	3.886 s	-2.1%	122.230	124.810	+2.1%	6.69 GiB	6.60 GiB	-1.3%	684.6 ms	653.3 ms	-4.6% 🟢	132.6 ms	134.9 ms	+1.7%	20.13 MiB	21.92 MiB	+8.9% (1.09× larger) ⚠️	8.02B	7.24B	-9.7% (1.11× fewer) 🟢
`Nat.gcd_comm`	3.235 s	3.104 s	-4.1% 🟢	120.550	125.640	+4.2% 🟢	5.07 GiB	5.78 GiB	+14.0% (1.14× larger) ⚠️	524.7 ms	507.6 ms	-3.3% 🟢	123.7 ms	133.3 ms	+7.8% (1.08× slower) ⚠️	19.90 MiB	21.68 MiB	+9.0% (1.09× larger) ⚠️	5.25B	4.60B	-12.4% (1.14× fewer) 🟢
`String.append`	2.399 s	2.299 s	-4.2% 🟢	126.740	132.250	+4.3% 🟢	4.62 GiB	4.64 GiB	+0.3%	403.9 ms	393.2 ms	-2.7%	123.5 ms	127.6 ms	+3.3% ⚠️	19.17 MiB	20.96 MiB	+9.3% (1.09× larger) ⚠️	2.90B	2.55B	-12.1% (1.14× fewer) 🟢
`Int.gcd`	1.991 s	2.001 s	+0.5%	104.480	103.950	-0.5%	4.50 GiB	4.53 GiB	+0.7%	355.0 ms	343.3 ms	-3.3% 🟢	118.2 ms	129.5 ms	+9.5% (1.10× slower) ⚠️	18.73 MiB	20.51 MiB	+9.5% (1.10× larger) ⚠️	1.87B	1.64B	-11.9% (1.14× fewer) 🟢
`Nat.sub_le_of_le_add`	1.815 s	1.788 s	-1.5%	93.650	95.090	+1.5%	4.60 GiB	4.62 GiB	+0.3%	343.9 ms	333.7 ms	-2.9%	113.8 ms	128.4 ms	+12.8% (1.13× slower) ⚠️	19.08 MiB	20.87 MiB	+9.4% (1.09× larger) ⚠️	1.60B	1.42B	-11.5% (1.13× fewer) 🟢
`Nat.add_comm`	1.011 s	1.050 s	+3.9% ⚠️	41.550	40	-3.7% ⚠️	3.54 GiB	4.17 GiB	+17.7% (1.18× larger) ⚠️	255.5 ms	251.9 ms	-1.4%	106.7 ms	116.6 ms	+9.3% (1.09× slower) ⚠️	17.25 MiB	19.04 MiB	+10.4% (1.10× larger) ⚠️	271.37M	249.57M	-8.0% (1.09× fewer) 🟢

Workflow logs

Rebased onto the fused xor-rotation branch: the seven inline calls now compose with the two-operand u32_xor_rotr7/12 forms. Codegen regenerated and all 71 kernel FFT pins + the shard aggregate re-measured on top of the new base (median -5.4%, range -11.4%..-0.2%; shard 6.08B -> 5.21B, -14.4%). aiur (cargo + prove + cross), ixvm (pins + parity), multi-stark and recursive-verifier suites pass; fmt, clippy, deny clean.

gabriel-barrett · 2026-08-14T17:51:01Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T17:51:11Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T17:55:05Z

`!benchmark` — main vs `b6a4f22`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	34.284 s	31.716 s	-7.5% (1.08× faster) 🟢	102.25 GiB	81.04 GiB	-20.7% (1.26× smaller) 🟢	5.27 MiB	6.27 MiB	+18.9% (1.19× larger) ⚠️	34.7 ms	37.9 ms	+9.2% (1.09× slower) ⚠️	5.645 s	4.839 s	-14.3% (1.17× faster) 🟢	173.38B	143.49B	-17.2% (1.21× fewer) 🟢	15.805 s	15.068 s	-4.7% 🟢	11.19 MiB	12.14 MiB	+8.5% (1.09× larger) ⚠️	68.0 ms	75.8 ms	+11.6% (1.12× slower) ⚠️	32.25 GiB	29.75 GiB	-7.7% (1.08× smaller) 🟢
`Vector.append`	31.433 s	28.063 s	-10.7% (1.12× faster) 🟢	90.02 GiB	73.45 GiB	-18.4% (1.23× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	32.5 ms	37.4 ms	+15.3% (1.15× slower) ⚠️	4.991 s	4.283 s	-14.2% (1.17× faster) 🟢	152.01B	128.00B	-15.8% (1.19× fewer) 🟢	3.853 s	3.825 s	-0.7%	10.39 MiB	11.34 MiB	+9.2% (1.09× larger) ⚠️	65.2 ms	68.8 ms	+5.5% (1.06× slower) ⚠️	7.23 GiB	6.34 GiB	-12.3% (1.14× smaller) 🟢
`Nat.add_comm`	20.507 s	17.985 s	-12.3% (1.14× faster) 🟢	55.58 GiB	44.60 GiB	-19.8% (1.25× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	34.9 ms	38.4 ms	+10.2% (1.10× slower) ⚠️	3.591 s	3.070 s	-14.5% (1.17× faster) 🟢	107.43B	92.75B	-13.7% (1.16× fewer) 🟢	1.034 s	1.031 s	-0.4%	8.90 MiB	9.86 MiB	+10.7% (1.11× larger) ⚠️	53.3 ms	61.2 ms	+14.9% (1.15× slower) ⚠️	3.98 GiB	3.56 GiB	-10.7% (1.12× smaller) 🟢

Workflow logs

argument-ci-bot · 2026-08-14T17:55:08Z

`!benchmark` — main vs `b6a4f22`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	37.550 s	37.412 s	-0.4%	41.070	41.220	+0.4%	73.20 GiB	70.64 GiB	-3.5% 🟢	9.577 s	9.391 s	-1.9%	138.9 ms	151.1 ms	+8.8% (1.09× slower) ⚠️	21.54 MiB	23.38 MiB	+8.5% (1.09× larger) ⚠️	135.19B	130.81B	-3.2% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	37.157 s	34.546 s	-7.0% (1.08× faster) 🟢	72.420	77.900	+7.6% (1.08× faster) 🟢	72.13 GiB	67.22 GiB	-6.8% (1.07× smaller) 🟢	9.613 s	9.122 s	-5.1% (1.05× faster) 🟢	197.0 ms	148.7 ms	-24.5% (1.33× faster) 🟢	21.63 MiB	23.47 MiB	+8.5% (1.09× larger) ⚠️	136.87B	126.89B	-7.3% (1.08× fewer) 🟢
`Char.ofOrdinal_le_of_le`	28.515 s	28.454 s	-0.2%	92.790	92.990	+0.2%	58.03 GiB	58.51 GiB	+0.8%	6.602 s	6.421 s	-2.8%	131.8 ms	150.6 ms	+14.2% (1.14× slower) ⚠️	21.58 MiB	23.42 MiB	+8.5% (1.09× larger) ⚠️	98.00B	90.63B	-7.5% (1.08× fewer) 🟢
`Vector.extract_append._proof_2`	21.362 s	20.460 s	-4.2% 🟢	60.950	63.640	+4.4% 🟢	39.42 GiB	38.21 GiB	-3.1% 🟢	5.182 s	5.110 s	-1.4%	128.2 ms	152.2 ms	+18.7% (1.19× slower) ⚠️	21.24 MiB	23.08 MiB	+8.7% (1.09× larger) ⚠️	76.48B	73.31B	-4.1% 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.040 s	15.965 s	-6.3% (1.07× faster) 🟢	106.340	113.500	+6.7% (1.07× faster) 🟢	34.22 GiB	31.82 GiB	-7.0% (1.08× smaller) 🟢	3.560 s	3.361 s	-5.6% (1.06× faster) 🟢	134.9 ms	143.8 ms	+6.6% (1.07× slower) ⚠️	21.42 MiB	23.26 MiB	+8.6% (1.09× larger) ⚠️	54.00B	49.57B	-8.2% (1.09× fewer) 🟢
`String.split`	16.080 s	15.024 s	-6.6% (1.07× faster) 🟢	109.770	117.480	+7.0% (1.07× faster) 🟢	32.26 GiB	29.72 GiB	-7.9% (1.09× smaller) 🟢	3.243 s	3.158 s	-2.6%	132.5 ms	141.8 ms	+7.0% (1.07× slower) ⚠️	21.62 MiB	23.46 MiB	+8.5% (1.09× larger) ⚠️	49.51B	45.59B	-7.9% (1.09× fewer) 🟢
`List.mergeSort`	11.944 s	11.350 s	-5.0% (1.05× faster) 🟢	121.230	127.570	+5.2% (1.05× faster) 🟢	23.52 GiB	22.26 GiB	-5.4% (1.06× smaller) 🟢	2.435 s	2.308 s	-5.2% (1.05× faster) 🟢	133.7 ms	142.9 ms	+6.9% (1.07× slower) ⚠️	21.48 MiB	23.32 MiB	+8.6% (1.09× larger) ⚠️	35.52B	32.44B	-8.7% (1.09× fewer) 🟢
`Vector.append`	3.982 s	3.778 s	-5.1% (1.05× faster) 🟢	121.780	128.380	+5.4% (1.05× faster) 🟢	7.11 GiB	6.06 GiB	-14.8% (1.17× smaller) 🟢	666.1 ms	647.1 ms	-2.8%	135.2 ms	131.0 ms	-3.1% 🟢	20.08 MiB	21.92 MiB	+9.2% (1.09× larger) ⚠️	7.90B	7.24B	-8.3% (1.09× fewer) 🟢
`Nat.gcd_comm`	3.194 s	3.051 s	-4.5% 🟢	122.110	127.830	+4.7% 🟢	4.92 GiB	4.81 GiB	-2.2%	529.7 ms	504.2 ms	-4.8% (1.05× faster) 🟢	122.8 ms	131.9 ms	+7.4% (1.07× slower) ⚠️	19.84 MiB	21.68 MiB	+9.3% (1.09× larger) ⚠️	5.15B	4.60B	-10.7% (1.12× fewer) 🟢
`String.append`	2.339 s	2.245 s	-4.0% 🟢	129.970	135.390	+4.2% 🟢	4.63 GiB	4.79 GiB	+3.5% ⚠️	400.3 ms	382.6 ms	-4.4% 🟢	120.4 ms	127.3 ms	+5.7% (1.06× slower) ⚠️	19.12 MiB	20.96 MiB	+9.6% (1.10× larger) ⚠️	2.85B	2.55B	-10.5% (1.12× fewer) 🟢
`Int.gcd`	1.991 s	1.956 s	-1.8%	104.490	106.350	+1.8%	5.40 GiB	4.53 GiB	-16.2% (1.19× smaller) 🟢	342.9 ms	340.7 ms	-0.6%	113.8 ms	127.5 ms	+12.0% (1.12× slower) ⚠️	18.67 MiB	20.51 MiB	+9.9% (1.10× larger) ⚠️	1.83B	1.64B	-10.3% (1.12× fewer) 🟢
`Nat.sub_le_of_le_add`	1.788 s	1.741 s	-2.7%	95.060	97.660	+2.7%	4.42 GiB	4.61 GiB	+4.3% ⚠️	337.3 ms	333.3 ms	-1.2%	113.0 ms	128.9 ms	+14.1% (1.14× slower) ⚠️	19.03 MiB	20.87 MiB	+9.7% (1.10× larger) ⚠️	1.58B	1.42B	-10.0% (1.11× fewer) 🟢
`Nat.add_comm`	1.019 s	1.084 s	+6.3% (1.06× slower) ⚠️	41.200	38.740	-6.0% (1.06× slower) ⚠️	4.17 GiB	3.49 GiB	-16.4% (1.20× smaller) 🟢	245.4 ms	246.5 ms	+0.4%	107.8 ms	115.9 ms	+7.4% (1.07× slower) ⚠️	17.20 MiB	19.04 MiB	+10.7% (1.11× larger) ⚠️	268.28M	249.57M	-7.0% (1.07× fewer) 🟢

Workflow logs

gabriel-barrett · 2026-08-13T15:47:07Z

Inline blake3_compress_inner_j. This has the effect of increasing the total width, though not at the same proportion as what we gain on efficiency, for the lookup itself is costly, especially when the input arguments are big.

Eventually we will have to decrease the proof size of the recursive verifier, but there are ways around it see #551

gabriel-barrett · 2026-08-14T15:52:31Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T15:52:48Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T15:56:36Z

`!benchmark` — main vs `69ce736`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `4e89546` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	35.764 s	31.074 s	-13.1% (1.15× faster) 🟢	107.68 GiB	81.50 GiB	-24.3% (1.32× smaller) 🟢	5.30 MiB	6.27 MiB	+18.3% (1.18× larger) ⚠️	33.1 ms	38.1 ms	+15.3% (1.15× slower) ⚠️	5.961 s	4.727 s	-20.7% (1.26× faster) 🟢	180.38B	143.49B	-20.5% (1.26× fewer) 🟢	15.713 s	14.745 s	-6.2% (1.07× faster) 🟢	11.21 MiB	12.14 MiB	+8.3% (1.08× larger) ⚠️	69.0 ms	75.7 ms	+9.7% (1.10× slower) ⚠️	32.74 GiB	29.83 GiB	-8.9% (1.10× smaller) 🟢
`Vector.append`	32.628 s	27.362 s	-16.1% (1.19× faster) 🟢	95.00 GiB	73.59 GiB	-22.5% (1.29× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	32.4 ms	39.9 ms	+23.0% (1.23× slower) ⚠️	5.158 s	4.255 s	-17.5% (1.21× faster) 🟢	157.02B	128.00B	-18.5% (1.23× fewer) 🟢	3.885 s	3.795 s	-2.3%	10.42 MiB	11.34 MiB	+8.9% (1.09× larger) ⚠️	65.7 ms	69.1 ms	+5.1% (1.05× slower) ⚠️	6.80 GiB	6.89 GiB	+1.4%
`Nat.add_comm`	21.106 s	17.709 s	-16.1% (1.19× faster) 🟢	58.08 GiB	44.38 GiB	-23.6% (1.31× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	34.6 ms	38.9 ms	+12.3% (1.12× slower) ⚠️	3.752 s	3.105 s	-17.2% (1.21× faster) 🟢	113.85B	92.75B	-18.5% (1.23× fewer) 🟢	1.024 s	1.034 s	+1.0%	8.93 MiB	9.86 MiB	+10.4% (1.10× larger) ⚠️	52.6 ms	60.8 ms	+15.7% (1.16× slower) ⚠️	3.79 GiB	4.06 GiB	+7.0% (1.07× larger) ⚠️

Workflow logs

argument-ci-bot · 2026-08-14T15:56:40Z

`!benchmark` — main vs `69ce736`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `4e89546` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	39.589 s	38.272 s	-3.3% 🟢	38.950	40.290	+3.4% 🟢	73.66 GiB	70.70 GiB	-4.0% 🟢	9.861 s	9.625 s	-2.4%	131.2 ms	146.6 ms	+11.7% (1.12× slower) ⚠️	21.59 MiB	23.38 MiB	+8.3% (1.08× larger) ⚠️	136.02B	130.81B	-3.8% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	38.143 s	36.140 s	-5.2% (1.06× faster) 🟢	70.550	74.460	+5.5% (1.06× faster) 🟢	73.09 GiB	67.14 GiB	-8.1% (1.09× smaller) 🟢	9.759 s	9.374 s	-3.9% 🟢	136.4 ms	145.1 ms	+6.4% (1.06× slower) ⚠️	21.69 MiB	23.47 MiB	+8.2% (1.08× larger) ⚠️	138.80B	126.89B	-8.6% (1.09× fewer) 🟢
`Char.ofOrdinal_le_of_le`	29.836 s	29.502 s	-1.1%	88.690	89.690	+1.1%	58.53 GiB	58.49 GiB	-0.1%	6.903 s	6.630 s	-4.0% 🟢	137.5 ms	201.3 ms	+46.4% (1.46× slower) ⚠️	21.64 MiB	23.42 MiB	+8.3% (1.08× larger) ⚠️	99.42B	90.63B	-8.8% (1.10× fewer) 🟢
`Vector.extract_append._proof_2`	21.860 s	21.323 s	-2.5%	59.560	61.060	+2.5%	39.70 GiB	38.22 GiB	-3.7% 🟢	5.365 s	5.263 s	-1.9%	130.4 ms	148.2 ms	+13.7% (1.14× slower) ⚠️	21.30 MiB	23.08 MiB	+8.4% (1.08× larger) ⚠️	77.08B	73.31B	-4.9% (1.05× fewer) 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.638 s	16.652 s	-5.6% (1.06× faster) 🟢	102.740	108.810	+5.9% (1.06× faster) 🟢	34.77 GiB	31.77 GiB	-8.6% (1.09× smaller) 🟢	3.634 s	3.537 s	-2.7%	138.0 ms	149.8 ms	+8.5% (1.09× slower) ⚠️	21.47 MiB	23.26 MiB	+8.3% (1.08× larger) ⚠️	54.85B	49.57B	-9.6% (1.11× fewer) 🟢
`String.split`	16.598 s	15.578 s	-6.1% (1.07× faster) 🟢	106.340	113.300	+6.5% (1.07× faster) 🟢	32.71 GiB	29.74 GiB	-9.1% (1.10× smaller) 🟢	3.376 s	3.252 s	-3.7% 🟢	133.0 ms	141.9 ms	+6.7% (1.07× slower) ⚠️	21.68 MiB	23.46 MiB	+8.2% (1.08× larger) ⚠️	50.25B	45.59B	-9.3% (1.10× fewer) 🟢
`List.mergeSort`	12.202 s	11.714 s	-4.0% 🟢	118.670	123.610	+4.2% 🟢	23.77 GiB	22.24 GiB	-6.4% (1.07× smaller) 🟢	2.438 s	2.333 s	-4.3% 🟢	129.5 ms	147.4 ms	+13.7% (1.14× slower) ⚠️	21.53 MiB	23.32 MiB	+8.3% (1.08× larger) ⚠️	36.10B	32.44B	-10.1% (1.11× fewer) 🟢
`Vector.append`	3.968 s	3.886 s	-2.1%	122.230	124.810	+2.1%	6.69 GiB	6.60 GiB	-1.3%	684.6 ms	653.3 ms	-4.6% 🟢	132.6 ms	134.9 ms	+1.7%	20.13 MiB	21.92 MiB	+8.9% (1.09× larger) ⚠️	8.02B	7.24B	-9.7% (1.11× fewer) 🟢
`Nat.gcd_comm`	3.235 s	3.104 s	-4.1% 🟢	120.550	125.640	+4.2% 🟢	5.07 GiB	5.78 GiB	+14.0% (1.14× larger) ⚠️	524.7 ms	507.6 ms	-3.3% 🟢	123.7 ms	133.3 ms	+7.8% (1.08× slower) ⚠️	19.90 MiB	21.68 MiB	+9.0% (1.09× larger) ⚠️	5.25B	4.60B	-12.4% (1.14× fewer) 🟢
`String.append`	2.399 s	2.299 s	-4.2% 🟢	126.740	132.250	+4.3% 🟢	4.62 GiB	4.64 GiB	+0.3%	403.9 ms	393.2 ms	-2.7%	123.5 ms	127.6 ms	+3.3% ⚠️	19.17 MiB	20.96 MiB	+9.3% (1.09× larger) ⚠️	2.90B	2.55B	-12.1% (1.14× fewer) 🟢
`Int.gcd`	1.991 s	2.001 s	+0.5%	104.480	103.950	-0.5%	4.50 GiB	4.53 GiB	+0.7%	355.0 ms	343.3 ms	-3.3% 🟢	118.2 ms	129.5 ms	+9.5% (1.10× slower) ⚠️	18.73 MiB	20.51 MiB	+9.5% (1.10× larger) ⚠️	1.87B	1.64B	-11.9% (1.14× fewer) 🟢
`Nat.sub_le_of_le_add`	1.815 s	1.788 s	-1.5%	93.650	95.090	+1.5%	4.60 GiB	4.62 GiB	+0.3%	343.9 ms	333.7 ms	-2.9%	113.8 ms	128.4 ms	+12.8% (1.13× slower) ⚠️	19.08 MiB	20.87 MiB	+9.4% (1.09× larger) ⚠️	1.60B	1.42B	-11.5% (1.13× fewer) 🟢
`Nat.add_comm`	1.011 s	1.050 s	+3.9% ⚠️	41.550	40	-3.7% ⚠️	3.54 GiB	4.17 GiB	+17.7% (1.18× larger) ⚠️	255.5 ms	251.9 ms	-1.4%	106.7 ms	116.6 ms	+9.3% (1.09× slower) ⚠️	17.25 MiB	19.04 MiB	+10.4% (1.10× larger) ⚠️	271.37M	249.57M	-8.0% (1.09× fewer) 🟢

Workflow logs

Rebased onto the fused xor-rotation branch: the seven inline calls now compose with the two-operand u32_xor_rotr7/12 forms. Codegen regenerated and all 71 kernel FFT pins + the shard aggregate re-measured on top of the new base (median -5.4%, range -11.4%..-0.2%; shard 6.08B -> 5.21B, -14.4%). aiur (cargo + prove + cross), ixvm (pins + parity), multi-stark and recursive-verifier suites pass; fmt, clippy, deny clean.

gabriel-barrett · 2026-08-14T17:51:01Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T17:51:11Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T17:55:05Z

`!benchmark` — main vs `b6a4f22`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	34.284 s	31.716 s	-7.5% (1.08× faster) 🟢	102.25 GiB	81.04 GiB	-20.7% (1.26× smaller) 🟢	5.27 MiB	6.27 MiB	+18.9% (1.19× larger) ⚠️	34.7 ms	37.9 ms	+9.2% (1.09× slower) ⚠️	5.645 s	4.839 s	-14.3% (1.17× faster) 🟢	173.38B	143.49B	-17.2% (1.21× fewer) 🟢	15.805 s	15.068 s	-4.7% 🟢	11.19 MiB	12.14 MiB	+8.5% (1.09× larger) ⚠️	68.0 ms	75.8 ms	+11.6% (1.12× slower) ⚠️	32.25 GiB	29.75 GiB	-7.7% (1.08× smaller) 🟢
`Vector.append`	31.433 s	28.063 s	-10.7% (1.12× faster) 🟢	90.02 GiB	73.45 GiB	-18.4% (1.23× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	32.5 ms	37.4 ms	+15.3% (1.15× slower) ⚠️	4.991 s	4.283 s	-14.2% (1.17× faster) 🟢	152.01B	128.00B	-15.8% (1.19× fewer) 🟢	3.853 s	3.825 s	-0.7%	10.39 MiB	11.34 MiB	+9.2% (1.09× larger) ⚠️	65.2 ms	68.8 ms	+5.5% (1.06× slower) ⚠️	7.23 GiB	6.34 GiB	-12.3% (1.14× smaller) 🟢
`Nat.add_comm`	20.507 s	17.985 s	-12.3% (1.14× faster) 🟢	55.58 GiB	44.60 GiB	-19.8% (1.25× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	34.9 ms	38.4 ms	+10.2% (1.10× slower) ⚠️	3.591 s	3.070 s	-14.5% (1.17× faster) 🟢	107.43B	92.75B	-13.7% (1.16× fewer) 🟢	1.034 s	1.031 s	-0.4%	8.90 MiB	9.86 MiB	+10.7% (1.11× larger) ⚠️	53.3 ms	61.2 ms	+14.9% (1.15× slower) ⚠️	3.98 GiB	3.56 GiB	-10.7% (1.12× smaller) 🟢

Workflow logs

argument-ci-bot · 2026-08-14T17:55:08Z

`!benchmark` — main vs `b6a4f22`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	37.550 s	37.412 s	-0.4%	41.070	41.220	+0.4%	73.20 GiB	70.64 GiB	-3.5% 🟢	9.577 s	9.391 s	-1.9%	138.9 ms	151.1 ms	+8.8% (1.09× slower) ⚠️	21.54 MiB	23.38 MiB	+8.5% (1.09× larger) ⚠️	135.19B	130.81B	-3.2% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	37.157 s	34.546 s	-7.0% (1.08× faster) 🟢	72.420	77.900	+7.6% (1.08× faster) 🟢	72.13 GiB	67.22 GiB	-6.8% (1.07× smaller) 🟢	9.613 s	9.122 s	-5.1% (1.05× faster) 🟢	197.0 ms	148.7 ms	-24.5% (1.33× faster) 🟢	21.63 MiB	23.47 MiB	+8.5% (1.09× larger) ⚠️	136.87B	126.89B	-7.3% (1.08× fewer) 🟢
`Char.ofOrdinal_le_of_le`	28.515 s	28.454 s	-0.2%	92.790	92.990	+0.2%	58.03 GiB	58.51 GiB	+0.8%	6.602 s	6.421 s	-2.8%	131.8 ms	150.6 ms	+14.2% (1.14× slower) ⚠️	21.58 MiB	23.42 MiB	+8.5% (1.09× larger) ⚠️	98.00B	90.63B	-7.5% (1.08× fewer) 🟢
`Vector.extract_append._proof_2`	21.362 s	20.460 s	-4.2% 🟢	60.950	63.640	+4.4% 🟢	39.42 GiB	38.21 GiB	-3.1% 🟢	5.182 s	5.110 s	-1.4%	128.2 ms	152.2 ms	+18.7% (1.19× slower) ⚠️	21.24 MiB	23.08 MiB	+8.7% (1.09× larger) ⚠️	76.48B	73.31B	-4.1% 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.040 s	15.965 s	-6.3% (1.07× faster) 🟢	106.340	113.500	+6.7% (1.07× faster) 🟢	34.22 GiB	31.82 GiB	-7.0% (1.08× smaller) 🟢	3.560 s	3.361 s	-5.6% (1.06× faster) 🟢	134.9 ms	143.8 ms	+6.6% (1.07× slower) ⚠️	21.42 MiB	23.26 MiB	+8.6% (1.09× larger) ⚠️	54.00B	49.57B	-8.2% (1.09× fewer) 🟢
`String.split`	16.080 s	15.024 s	-6.6% (1.07× faster) 🟢	109.770	117.480	+7.0% (1.07× faster) 🟢	32.26 GiB	29.72 GiB	-7.9% (1.09× smaller) 🟢	3.243 s	3.158 s	-2.6%	132.5 ms	141.8 ms	+7.0% (1.07× slower) ⚠️	21.62 MiB	23.46 MiB	+8.5% (1.09× larger) ⚠️	49.51B	45.59B	-7.9% (1.09× fewer) 🟢
`List.mergeSort`	11.944 s	11.350 s	-5.0% (1.05× faster) 🟢	121.230	127.570	+5.2% (1.05× faster) 🟢	23.52 GiB	22.26 GiB	-5.4% (1.06× smaller) 🟢	2.435 s	2.308 s	-5.2% (1.05× faster) 🟢	133.7 ms	142.9 ms	+6.9% (1.07× slower) ⚠️	21.48 MiB	23.32 MiB	+8.6% (1.09× larger) ⚠️	35.52B	32.44B	-8.7% (1.09× fewer) 🟢
`Vector.append`	3.982 s	3.778 s	-5.1% (1.05× faster) 🟢	121.780	128.380	+5.4% (1.05× faster) 🟢	7.11 GiB	6.06 GiB	-14.8% (1.17× smaller) 🟢	666.1 ms	647.1 ms	-2.8%	135.2 ms	131.0 ms	-3.1% 🟢	20.08 MiB	21.92 MiB	+9.2% (1.09× larger) ⚠️	7.90B	7.24B	-8.3% (1.09× fewer) 🟢
`Nat.gcd_comm`	3.194 s	3.051 s	-4.5% 🟢	122.110	127.830	+4.7% 🟢	4.92 GiB	4.81 GiB	-2.2%	529.7 ms	504.2 ms	-4.8% (1.05× faster) 🟢	122.8 ms	131.9 ms	+7.4% (1.07× slower) ⚠️	19.84 MiB	21.68 MiB	+9.3% (1.09× larger) ⚠️	5.15B	4.60B	-10.7% (1.12× fewer) 🟢
`String.append`	2.339 s	2.245 s	-4.0% 🟢	129.970	135.390	+4.2% 🟢	4.63 GiB	4.79 GiB	+3.5% ⚠️	400.3 ms	382.6 ms	-4.4% 🟢	120.4 ms	127.3 ms	+5.7% (1.06× slower) ⚠️	19.12 MiB	20.96 MiB	+9.6% (1.10× larger) ⚠️	2.85B	2.55B	-10.5% (1.12× fewer) 🟢
`Int.gcd`	1.991 s	1.956 s	-1.8%	104.490	106.350	+1.8%	5.40 GiB	4.53 GiB	-16.2% (1.19× smaller) 🟢	342.9 ms	340.7 ms	-0.6%	113.8 ms	127.5 ms	+12.0% (1.12× slower) ⚠️	18.67 MiB	20.51 MiB	+9.9% (1.10× larger) ⚠️	1.83B	1.64B	-10.3% (1.12× fewer) 🟢
`Nat.sub_le_of_le_add`	1.788 s	1.741 s	-2.7%	95.060	97.660	+2.7%	4.42 GiB	4.61 GiB	+4.3% ⚠️	337.3 ms	333.3 ms	-1.2%	113.0 ms	128.9 ms	+14.1% (1.14× slower) ⚠️	19.03 MiB	20.87 MiB	+9.7% (1.10× larger) ⚠️	1.58B	1.42B	-10.0% (1.11× fewer) 🟢
`Nat.add_comm`	1.019 s	1.084 s	+6.3% (1.06× slower) ⚠️	41.200	38.740	-6.0% (1.06× slower) ⚠️	4.17 GiB	3.49 GiB	-16.4% (1.20× smaller) 🟢	245.4 ms	246.5 ms	+0.4%	107.8 ms	115.9 ms	+7.4% (1.07× slower) ⚠️	17.20 MiB	19.04 MiB	+10.7% (1.11× larger) ⚠️	268.28M	249.57M	-7.0% (1.07× fewer) 🟢

Workflow logs

gabriel-barrett · 2026-08-13T15:47:07Z

Inline blake3_compress_inner_j. This has the effect of increasing the total width, though not at the same proportion as what we gain on efficiency, for the lookup itself is costly, especially when the input arguments are big.

Eventually we will have to decrease the proof size of the recursive verifier, but there are ways around it see #551

gabriel-barrett · 2026-08-14T15:52:31Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T15:52:48Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T15:56:36Z

`!benchmark` — main vs `69ce736`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `4e89546` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	35.764 s	31.074 s	-13.1% (1.15× faster) 🟢	107.68 GiB	81.50 GiB	-24.3% (1.32× smaller) 🟢	5.30 MiB	6.27 MiB	+18.3% (1.18× larger) ⚠️	33.1 ms	38.1 ms	+15.3% (1.15× slower) ⚠️	5.961 s	4.727 s	-20.7% (1.26× faster) 🟢	180.38B	143.49B	-20.5% (1.26× fewer) 🟢	15.713 s	14.745 s	-6.2% (1.07× faster) 🟢	11.21 MiB	12.14 MiB	+8.3% (1.08× larger) ⚠️	69.0 ms	75.7 ms	+9.7% (1.10× slower) ⚠️	32.74 GiB	29.83 GiB	-8.9% (1.10× smaller) 🟢
`Vector.append`	32.628 s	27.362 s	-16.1% (1.19× faster) 🟢	95.00 GiB	73.59 GiB	-22.5% (1.29× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	32.4 ms	39.9 ms	+23.0% (1.23× slower) ⚠️	5.158 s	4.255 s	-17.5% (1.21× faster) 🟢	157.02B	128.00B	-18.5% (1.23× fewer) 🟢	3.885 s	3.795 s	-2.3%	10.42 MiB	11.34 MiB	+8.9% (1.09× larger) ⚠️	65.7 ms	69.1 ms	+5.1% (1.05× slower) ⚠️	6.80 GiB	6.89 GiB	+1.4%
`Nat.add_comm`	21.106 s	17.709 s	-16.1% (1.19× faster) 🟢	58.08 GiB	44.38 GiB	-23.6% (1.31× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	34.6 ms	38.9 ms	+12.3% (1.12× slower) ⚠️	3.752 s	3.105 s	-17.2% (1.21× faster) 🟢	113.85B	92.75B	-18.5% (1.23× fewer) 🟢	1.024 s	1.034 s	+1.0%	8.93 MiB	9.86 MiB	+10.4% (1.10× larger) ⚠️	52.6 ms	60.8 ms	+15.7% (1.16× slower) ⚠️	3.79 GiB	4.06 GiB	+7.0% (1.07× larger) ⚠️

Workflow logs

argument-ci-bot · 2026-08-14T15:56:40Z

`!benchmark` — main vs `69ce736`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `4e89546` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	39.589 s	38.272 s	-3.3% 🟢	38.950	40.290	+3.4% 🟢	73.66 GiB	70.70 GiB	-4.0% 🟢	9.861 s	9.625 s	-2.4%	131.2 ms	146.6 ms	+11.7% (1.12× slower) ⚠️	21.59 MiB	23.38 MiB	+8.3% (1.08× larger) ⚠️	136.02B	130.81B	-3.8% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	38.143 s	36.140 s	-5.2% (1.06× faster) 🟢	70.550	74.460	+5.5% (1.06× faster) 🟢	73.09 GiB	67.14 GiB	-8.1% (1.09× smaller) 🟢	9.759 s	9.374 s	-3.9% 🟢	136.4 ms	145.1 ms	+6.4% (1.06× slower) ⚠️	21.69 MiB	23.47 MiB	+8.2% (1.08× larger) ⚠️	138.80B	126.89B	-8.6% (1.09× fewer) 🟢
`Char.ofOrdinal_le_of_le`	29.836 s	29.502 s	-1.1%	88.690	89.690	+1.1%	58.53 GiB	58.49 GiB	-0.1%	6.903 s	6.630 s	-4.0% 🟢	137.5 ms	201.3 ms	+46.4% (1.46× slower) ⚠️	21.64 MiB	23.42 MiB	+8.3% (1.08× larger) ⚠️	99.42B	90.63B	-8.8% (1.10× fewer) 🟢
`Vector.extract_append._proof_2`	21.860 s	21.323 s	-2.5%	59.560	61.060	+2.5%	39.70 GiB	38.22 GiB	-3.7% 🟢	5.365 s	5.263 s	-1.9%	130.4 ms	148.2 ms	+13.7% (1.14× slower) ⚠️	21.30 MiB	23.08 MiB	+8.4% (1.08× larger) ⚠️	77.08B	73.31B	-4.9% (1.05× fewer) 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.638 s	16.652 s	-5.6% (1.06× faster) 🟢	102.740	108.810	+5.9% (1.06× faster) 🟢	34.77 GiB	31.77 GiB	-8.6% (1.09× smaller) 🟢	3.634 s	3.537 s	-2.7%	138.0 ms	149.8 ms	+8.5% (1.09× slower) ⚠️	21.47 MiB	23.26 MiB	+8.3% (1.08× larger) ⚠️	54.85B	49.57B	-9.6% (1.11× fewer) 🟢
`String.split`	16.598 s	15.578 s	-6.1% (1.07× faster) 🟢	106.340	113.300	+6.5% (1.07× faster) 🟢	32.71 GiB	29.74 GiB	-9.1% (1.10× smaller) 🟢	3.376 s	3.252 s	-3.7% 🟢	133.0 ms	141.9 ms	+6.7% (1.07× slower) ⚠️	21.68 MiB	23.46 MiB	+8.2% (1.08× larger) ⚠️	50.25B	45.59B	-9.3% (1.10× fewer) 🟢
`List.mergeSort`	12.202 s	11.714 s	-4.0% 🟢	118.670	123.610	+4.2% 🟢	23.77 GiB	22.24 GiB	-6.4% (1.07× smaller) 🟢	2.438 s	2.333 s	-4.3% 🟢	129.5 ms	147.4 ms	+13.7% (1.14× slower) ⚠️	21.53 MiB	23.32 MiB	+8.3% (1.08× larger) ⚠️	36.10B	32.44B	-10.1% (1.11× fewer) 🟢
`Vector.append`	3.968 s	3.886 s	-2.1%	122.230	124.810	+2.1%	6.69 GiB	6.60 GiB	-1.3%	684.6 ms	653.3 ms	-4.6% 🟢	132.6 ms	134.9 ms	+1.7%	20.13 MiB	21.92 MiB	+8.9% (1.09× larger) ⚠️	8.02B	7.24B	-9.7% (1.11× fewer) 🟢
`Nat.gcd_comm`	3.235 s	3.104 s	-4.1% 🟢	120.550	125.640	+4.2% 🟢	5.07 GiB	5.78 GiB	+14.0% (1.14× larger) ⚠️	524.7 ms	507.6 ms	-3.3% 🟢	123.7 ms	133.3 ms	+7.8% (1.08× slower) ⚠️	19.90 MiB	21.68 MiB	+9.0% (1.09× larger) ⚠️	5.25B	4.60B	-12.4% (1.14× fewer) 🟢
`String.append`	2.399 s	2.299 s	-4.2% 🟢	126.740	132.250	+4.3% 🟢	4.62 GiB	4.64 GiB	+0.3%	403.9 ms	393.2 ms	-2.7%	123.5 ms	127.6 ms	+3.3% ⚠️	19.17 MiB	20.96 MiB	+9.3% (1.09× larger) ⚠️	2.90B	2.55B	-12.1% (1.14× fewer) 🟢
`Int.gcd`	1.991 s	2.001 s	+0.5%	104.480	103.950	-0.5%	4.50 GiB	4.53 GiB	+0.7%	355.0 ms	343.3 ms	-3.3% 🟢	118.2 ms	129.5 ms	+9.5% (1.10× slower) ⚠️	18.73 MiB	20.51 MiB	+9.5% (1.10× larger) ⚠️	1.87B	1.64B	-11.9% (1.14× fewer) 🟢
`Nat.sub_le_of_le_add`	1.815 s	1.788 s	-1.5%	93.650	95.090	+1.5%	4.60 GiB	4.62 GiB	+0.3%	343.9 ms	333.7 ms	-2.9%	113.8 ms	128.4 ms	+12.8% (1.13× slower) ⚠️	19.08 MiB	20.87 MiB	+9.4% (1.09× larger) ⚠️	1.60B	1.42B	-11.5% (1.13× fewer) 🟢
`Nat.add_comm`	1.011 s	1.050 s	+3.9% ⚠️	41.550	40	-3.7% ⚠️	3.54 GiB	4.17 GiB	+17.7% (1.18× larger) ⚠️	255.5 ms	251.9 ms	-1.4%	106.7 ms	116.6 ms	+9.3% (1.09× slower) ⚠️	17.25 MiB	19.04 MiB	+10.4% (1.10× larger) ⚠️	271.37M	249.57M	-8.0% (1.09× fewer) 🟢

Workflow logs

Rebased onto the fused xor-rotation branch: the seven inline calls now compose with the two-operand u32_xor_rotr7/12 forms. Codegen regenerated and all 71 kernel FFT pins + the shard aggregate re-measured on top of the new base (median -5.4%, range -11.4%..-0.2%; shard 6.08B -> 5.21B, -14.4%). aiur (cargo + prove + cross), ixvm (pins + parity), multi-stark and recursive-verifier suites pass; fmt, clippy, deny clean.

gabriel-barrett · 2026-08-14T17:51:01Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T17:51:11Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T17:55:05Z

`!benchmark` — main vs `b6a4f22`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	34.284 s	31.716 s	-7.5% (1.08× faster) 🟢	102.25 GiB	81.04 GiB	-20.7% (1.26× smaller) 🟢	5.27 MiB	6.27 MiB	+18.9% (1.19× larger) ⚠️	34.7 ms	37.9 ms	+9.2% (1.09× slower) ⚠️	5.645 s	4.839 s	-14.3% (1.17× faster) 🟢	173.38B	143.49B	-17.2% (1.21× fewer) 🟢	15.805 s	15.068 s	-4.7% 🟢	11.19 MiB	12.14 MiB	+8.5% (1.09× larger) ⚠️	68.0 ms	75.8 ms	+11.6% (1.12× slower) ⚠️	32.25 GiB	29.75 GiB	-7.7% (1.08× smaller) 🟢
`Vector.append`	31.433 s	28.063 s	-10.7% (1.12× faster) 🟢	90.02 GiB	73.45 GiB	-18.4% (1.23× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	32.5 ms	37.4 ms	+15.3% (1.15× slower) ⚠️	4.991 s	4.283 s	-14.2% (1.17× faster) 🟢	152.01B	128.00B	-15.8% (1.19× fewer) 🟢	3.853 s	3.825 s	-0.7%	10.39 MiB	11.34 MiB	+9.2% (1.09× larger) ⚠️	65.2 ms	68.8 ms	+5.5% (1.06× slower) ⚠️	7.23 GiB	6.34 GiB	-12.3% (1.14× smaller) 🟢
`Nat.add_comm`	20.507 s	17.985 s	-12.3% (1.14× faster) 🟢	55.58 GiB	44.60 GiB	-19.8% (1.25× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	34.9 ms	38.4 ms	+10.2% (1.10× slower) ⚠️	3.591 s	3.070 s	-14.5% (1.17× faster) 🟢	107.43B	92.75B	-13.7% (1.16× fewer) 🟢	1.034 s	1.031 s	-0.4%	8.90 MiB	9.86 MiB	+10.7% (1.11× larger) ⚠️	53.3 ms	61.2 ms	+14.9% (1.15× slower) ⚠️	3.98 GiB	3.56 GiB	-10.7% (1.12× smaller) 🟢

Workflow logs

argument-ci-bot · 2026-08-14T17:55:08Z

`!benchmark` — main vs `b6a4f22`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	37.550 s	37.412 s	-0.4%	41.070	41.220	+0.4%	73.20 GiB	70.64 GiB	-3.5% 🟢	9.577 s	9.391 s	-1.9%	138.9 ms	151.1 ms	+8.8% (1.09× slower) ⚠️	21.54 MiB	23.38 MiB	+8.5% (1.09× larger) ⚠️	135.19B	130.81B	-3.2% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	37.157 s	34.546 s	-7.0% (1.08× faster) 🟢	72.420	77.900	+7.6% (1.08× faster) 🟢	72.13 GiB	67.22 GiB	-6.8% (1.07× smaller) 🟢	9.613 s	9.122 s	-5.1% (1.05× faster) 🟢	197.0 ms	148.7 ms	-24.5% (1.33× faster) 🟢	21.63 MiB	23.47 MiB	+8.5% (1.09× larger) ⚠️	136.87B	126.89B	-7.3% (1.08× fewer) 🟢
`Char.ofOrdinal_le_of_le`	28.515 s	28.454 s	-0.2%	92.790	92.990	+0.2%	58.03 GiB	58.51 GiB	+0.8%	6.602 s	6.421 s	-2.8%	131.8 ms	150.6 ms	+14.2% (1.14× slower) ⚠️	21.58 MiB	23.42 MiB	+8.5% (1.09× larger) ⚠️	98.00B	90.63B	-7.5% (1.08× fewer) 🟢
`Vector.extract_append._proof_2`	21.362 s	20.460 s	-4.2% 🟢	60.950	63.640	+4.4% 🟢	39.42 GiB	38.21 GiB	-3.1% 🟢	5.182 s	5.110 s	-1.4%	128.2 ms	152.2 ms	+18.7% (1.19× slower) ⚠️	21.24 MiB	23.08 MiB	+8.7% (1.09× larger) ⚠️	76.48B	73.31B	-4.1% 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.040 s	15.965 s	-6.3% (1.07× faster) 🟢	106.340	113.500	+6.7% (1.07× faster) 🟢	34.22 GiB	31.82 GiB	-7.0% (1.08× smaller) 🟢	3.560 s	3.361 s	-5.6% (1.06× faster) 🟢	134.9 ms	143.8 ms	+6.6% (1.07× slower) ⚠️	21.42 MiB	23.26 MiB	+8.6% (1.09× larger) ⚠️	54.00B	49.57B	-8.2% (1.09× fewer) 🟢
`String.split`	16.080 s	15.024 s	-6.6% (1.07× faster) 🟢	109.770	117.480	+7.0% (1.07× faster) 🟢	32.26 GiB	29.72 GiB	-7.9% (1.09× smaller) 🟢	3.243 s	3.158 s	-2.6%	132.5 ms	141.8 ms	+7.0% (1.07× slower) ⚠️	21.62 MiB	23.46 MiB	+8.5% (1.09× larger) ⚠️	49.51B	45.59B	-7.9% (1.09× fewer) 🟢
`List.mergeSort`	11.944 s	11.350 s	-5.0% (1.05× faster) 🟢	121.230	127.570	+5.2% (1.05× faster) 🟢	23.52 GiB	22.26 GiB	-5.4% (1.06× smaller) 🟢	2.435 s	2.308 s	-5.2% (1.05× faster) 🟢	133.7 ms	142.9 ms	+6.9% (1.07× slower) ⚠️	21.48 MiB	23.32 MiB	+8.6% (1.09× larger) ⚠️	35.52B	32.44B	-8.7% (1.09× fewer) 🟢
`Vector.append`	3.982 s	3.778 s	-5.1% (1.05× faster) 🟢	121.780	128.380	+5.4% (1.05× faster) 🟢	7.11 GiB	6.06 GiB	-14.8% (1.17× smaller) 🟢	666.1 ms	647.1 ms	-2.8%	135.2 ms	131.0 ms	-3.1% 🟢	20.08 MiB	21.92 MiB	+9.2% (1.09× larger) ⚠️	7.90B	7.24B	-8.3% (1.09× fewer) 🟢
`Nat.gcd_comm`	3.194 s	3.051 s	-4.5% 🟢	122.110	127.830	+4.7% 🟢	4.92 GiB	4.81 GiB	-2.2%	529.7 ms	504.2 ms	-4.8% (1.05× faster) 🟢	122.8 ms	131.9 ms	+7.4% (1.07× slower) ⚠️	19.84 MiB	21.68 MiB	+9.3% (1.09× larger) ⚠️	5.15B	4.60B	-10.7% (1.12× fewer) 🟢
`String.append`	2.339 s	2.245 s	-4.0% 🟢	129.970	135.390	+4.2% 🟢	4.63 GiB	4.79 GiB	+3.5% ⚠️	400.3 ms	382.6 ms	-4.4% 🟢	120.4 ms	127.3 ms	+5.7% (1.06× slower) ⚠️	19.12 MiB	20.96 MiB	+9.6% (1.10× larger) ⚠️	2.85B	2.55B	-10.5% (1.12× fewer) 🟢
`Int.gcd`	1.991 s	1.956 s	-1.8%	104.490	106.350	+1.8%	5.40 GiB	4.53 GiB	-16.2% (1.19× smaller) 🟢	342.9 ms	340.7 ms	-0.6%	113.8 ms	127.5 ms	+12.0% (1.12× slower) ⚠️	18.67 MiB	20.51 MiB	+9.9% (1.10× larger) ⚠️	1.83B	1.64B	-10.3% (1.12× fewer) 🟢
`Nat.sub_le_of_le_add`	1.788 s	1.741 s	-2.7%	95.060	97.660	+2.7%	4.42 GiB	4.61 GiB	+4.3% ⚠️	337.3 ms	333.3 ms	-1.2%	113.0 ms	128.9 ms	+14.1% (1.14× slower) ⚠️	19.03 MiB	20.87 MiB	+9.7% (1.10× larger) ⚠️	1.58B	1.42B	-10.0% (1.11× fewer) 🟢
`Nat.add_comm`	1.019 s	1.084 s	+6.3% (1.06× slower) ⚠️	41.200	38.740	-6.0% (1.06× slower) ⚠️	4.17 GiB	3.49 GiB	-16.4% (1.20× smaller) 🟢	245.4 ms	246.5 ms	+0.4%	107.8 ms	115.9 ms	+7.4% (1.07× slower) ⚠️	17.20 MiB	19.04 MiB	+10.7% (1.11× larger) ⚠️	268.28M	249.57M	-7.0% (1.07× fewer) 🟢

Workflow logs

gabriel-barrett · 2026-08-13T15:47:07Z

Inline blake3_compress_inner_j. This has the effect of increasing the total width, though not at the same proportion as what we gain on efficiency, for the lookup itself is costly, especially when the input arguments are big.

Eventually we will have to decrease the proof size of the recursive verifier, but there are ways around it see #551

gabriel-barrett · 2026-08-14T15:52:31Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T15:52:48Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T15:56:36Z

`!benchmark` — main vs `69ce736`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `4e89546` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	35.764 s	31.074 s	-13.1% (1.15× faster) 🟢	107.68 GiB	81.50 GiB	-24.3% (1.32× smaller) 🟢	5.30 MiB	6.27 MiB	+18.3% (1.18× larger) ⚠️	33.1 ms	38.1 ms	+15.3% (1.15× slower) ⚠️	5.961 s	4.727 s	-20.7% (1.26× faster) 🟢	180.38B	143.49B	-20.5% (1.26× fewer) 🟢	15.713 s	14.745 s	-6.2% (1.07× faster) 🟢	11.21 MiB	12.14 MiB	+8.3% (1.08× larger) ⚠️	69.0 ms	75.7 ms	+9.7% (1.10× slower) ⚠️	32.74 GiB	29.83 GiB	-8.9% (1.10× smaller) 🟢
`Vector.append`	32.628 s	27.362 s	-16.1% (1.19× faster) 🟢	95.00 GiB	73.59 GiB	-22.5% (1.29× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	32.4 ms	39.9 ms	+23.0% (1.23× slower) ⚠️	5.158 s	4.255 s	-17.5% (1.21× faster) 🟢	157.02B	128.00B	-18.5% (1.23× fewer) 🟢	3.885 s	3.795 s	-2.3%	10.42 MiB	11.34 MiB	+8.9% (1.09× larger) ⚠️	65.7 ms	69.1 ms	+5.1% (1.05× slower) ⚠️	6.80 GiB	6.89 GiB	+1.4%
`Nat.add_comm`	21.106 s	17.709 s	-16.1% (1.19× faster) 🟢	58.08 GiB	44.38 GiB	-23.6% (1.31× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	34.6 ms	38.9 ms	+12.3% (1.12× slower) ⚠️	3.752 s	3.105 s	-17.2% (1.21× faster) 🟢	113.85B	92.75B	-18.5% (1.23× fewer) 🟢	1.024 s	1.034 s	+1.0%	8.93 MiB	9.86 MiB	+10.4% (1.10× larger) ⚠️	52.6 ms	60.8 ms	+15.7% (1.16× slower) ⚠️	3.79 GiB	4.06 GiB	+7.0% (1.07× larger) ⚠️

Workflow logs

argument-ci-bot · 2026-08-14T15:56:40Z

`!benchmark` — main vs `69ce736`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `4e89546` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	39.589 s	38.272 s	-3.3% 🟢	38.950	40.290	+3.4% 🟢	73.66 GiB	70.70 GiB	-4.0% 🟢	9.861 s	9.625 s	-2.4%	131.2 ms	146.6 ms	+11.7% (1.12× slower) ⚠️	21.59 MiB	23.38 MiB	+8.3% (1.08× larger) ⚠️	136.02B	130.81B	-3.8% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	38.143 s	36.140 s	-5.2% (1.06× faster) 🟢	70.550	74.460	+5.5% (1.06× faster) 🟢	73.09 GiB	67.14 GiB	-8.1% (1.09× smaller) 🟢	9.759 s	9.374 s	-3.9% 🟢	136.4 ms	145.1 ms	+6.4% (1.06× slower) ⚠️	21.69 MiB	23.47 MiB	+8.2% (1.08× larger) ⚠️	138.80B	126.89B	-8.6% (1.09× fewer) 🟢
`Char.ofOrdinal_le_of_le`	29.836 s	29.502 s	-1.1%	88.690	89.690	+1.1%	58.53 GiB	58.49 GiB	-0.1%	6.903 s	6.630 s	-4.0% 🟢	137.5 ms	201.3 ms	+46.4% (1.46× slower) ⚠️	21.64 MiB	23.42 MiB	+8.3% (1.08× larger) ⚠️	99.42B	90.63B	-8.8% (1.10× fewer) 🟢
`Vector.extract_append._proof_2`	21.860 s	21.323 s	-2.5%	59.560	61.060	+2.5%	39.70 GiB	38.22 GiB	-3.7% 🟢	5.365 s	5.263 s	-1.9%	130.4 ms	148.2 ms	+13.7% (1.14× slower) ⚠️	21.30 MiB	23.08 MiB	+8.4% (1.08× larger) ⚠️	77.08B	73.31B	-4.9% (1.05× fewer) 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.638 s	16.652 s	-5.6% (1.06× faster) 🟢	102.740	108.810	+5.9% (1.06× faster) 🟢	34.77 GiB	31.77 GiB	-8.6% (1.09× smaller) 🟢	3.634 s	3.537 s	-2.7%	138.0 ms	149.8 ms	+8.5% (1.09× slower) ⚠️	21.47 MiB	23.26 MiB	+8.3% (1.08× larger) ⚠️	54.85B	49.57B	-9.6% (1.11× fewer) 🟢
`String.split`	16.598 s	15.578 s	-6.1% (1.07× faster) 🟢	106.340	113.300	+6.5% (1.07× faster) 🟢	32.71 GiB	29.74 GiB	-9.1% (1.10× smaller) 🟢	3.376 s	3.252 s	-3.7% 🟢	133.0 ms	141.9 ms	+6.7% (1.07× slower) ⚠️	21.68 MiB	23.46 MiB	+8.2% (1.08× larger) ⚠️	50.25B	45.59B	-9.3% (1.10× fewer) 🟢
`List.mergeSort`	12.202 s	11.714 s	-4.0% 🟢	118.670	123.610	+4.2% 🟢	23.77 GiB	22.24 GiB	-6.4% (1.07× smaller) 🟢	2.438 s	2.333 s	-4.3% 🟢	129.5 ms	147.4 ms	+13.7% (1.14× slower) ⚠️	21.53 MiB	23.32 MiB	+8.3% (1.08× larger) ⚠️	36.10B	32.44B	-10.1% (1.11× fewer) 🟢
`Vector.append`	3.968 s	3.886 s	-2.1%	122.230	124.810	+2.1%	6.69 GiB	6.60 GiB	-1.3%	684.6 ms	653.3 ms	-4.6% 🟢	132.6 ms	134.9 ms	+1.7%	20.13 MiB	21.92 MiB	+8.9% (1.09× larger) ⚠️	8.02B	7.24B	-9.7% (1.11× fewer) 🟢
`Nat.gcd_comm`	3.235 s	3.104 s	-4.1% 🟢	120.550	125.640	+4.2% 🟢	5.07 GiB	5.78 GiB	+14.0% (1.14× larger) ⚠️	524.7 ms	507.6 ms	-3.3% 🟢	123.7 ms	133.3 ms	+7.8% (1.08× slower) ⚠️	19.90 MiB	21.68 MiB	+9.0% (1.09× larger) ⚠️	5.25B	4.60B	-12.4% (1.14× fewer) 🟢
`String.append`	2.399 s	2.299 s	-4.2% 🟢	126.740	132.250	+4.3% 🟢	4.62 GiB	4.64 GiB	+0.3%	403.9 ms	393.2 ms	-2.7%	123.5 ms	127.6 ms	+3.3% ⚠️	19.17 MiB	20.96 MiB	+9.3% (1.09× larger) ⚠️	2.90B	2.55B	-12.1% (1.14× fewer) 🟢
`Int.gcd`	1.991 s	2.001 s	+0.5%	104.480	103.950	-0.5%	4.50 GiB	4.53 GiB	+0.7%	355.0 ms	343.3 ms	-3.3% 🟢	118.2 ms	129.5 ms	+9.5% (1.10× slower) ⚠️	18.73 MiB	20.51 MiB	+9.5% (1.10× larger) ⚠️	1.87B	1.64B	-11.9% (1.14× fewer) 🟢
`Nat.sub_le_of_le_add`	1.815 s	1.788 s	-1.5%	93.650	95.090	+1.5%	4.60 GiB	4.62 GiB	+0.3%	343.9 ms	333.7 ms	-2.9%	113.8 ms	128.4 ms	+12.8% (1.13× slower) ⚠️	19.08 MiB	20.87 MiB	+9.4% (1.09× larger) ⚠️	1.60B	1.42B	-11.5% (1.13× fewer) 🟢
`Nat.add_comm`	1.011 s	1.050 s	+3.9% ⚠️	41.550	40	-3.7% ⚠️	3.54 GiB	4.17 GiB	+17.7% (1.18× larger) ⚠️	255.5 ms	251.9 ms	-1.4%	106.7 ms	116.6 ms	+9.3% (1.09× slower) ⚠️	17.25 MiB	19.04 MiB	+10.4% (1.10× larger) ⚠️	271.37M	249.57M	-8.0% (1.09× fewer) 🟢

Workflow logs

Rebased onto the fused xor-rotation branch: the seven inline calls now compose with the two-operand u32_xor_rotr7/12 forms. Codegen regenerated and all 71 kernel FFT pins + the shard aggregate re-measured on top of the new base (median -5.4%, range -11.4%..-0.2%; shard 6.08B -> 5.21B, -14.4%). aiur (cargo + prove + cross), ixvm (pins + parity), multi-stark and recursive-verifier suites pass; fmt, clippy, deny clean.

gabriel-barrett · 2026-08-14T17:51:01Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T17:51:11Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T17:55:05Z

`!benchmark` — main vs `b6a4f22`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	34.284 s	31.716 s	-7.5% (1.08× faster) 🟢	102.25 GiB	81.04 GiB	-20.7% (1.26× smaller) 🟢	5.27 MiB	6.27 MiB	+18.9% (1.19× larger) ⚠️	34.7 ms	37.9 ms	+9.2% (1.09× slower) ⚠️	5.645 s	4.839 s	-14.3% (1.17× faster) 🟢	173.38B	143.49B	-17.2% (1.21× fewer) 🟢	15.805 s	15.068 s	-4.7% 🟢	11.19 MiB	12.14 MiB	+8.5% (1.09× larger) ⚠️	68.0 ms	75.8 ms	+11.6% (1.12× slower) ⚠️	32.25 GiB	29.75 GiB	-7.7% (1.08× smaller) 🟢
`Vector.append`	31.433 s	28.063 s	-10.7% (1.12× faster) 🟢	90.02 GiB	73.45 GiB	-18.4% (1.23× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	32.5 ms	37.4 ms	+15.3% (1.15× slower) ⚠️	4.991 s	4.283 s	-14.2% (1.17× faster) 🟢	152.01B	128.00B	-15.8% (1.19× fewer) 🟢	3.853 s	3.825 s	-0.7%	10.39 MiB	11.34 MiB	+9.2% (1.09× larger) ⚠️	65.2 ms	68.8 ms	+5.5% (1.06× slower) ⚠️	7.23 GiB	6.34 GiB	-12.3% (1.14× smaller) 🟢
`Nat.add_comm`	20.507 s	17.985 s	-12.3% (1.14× faster) 🟢	55.58 GiB	44.60 GiB	-19.8% (1.25× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	34.9 ms	38.4 ms	+10.2% (1.10× slower) ⚠️	3.591 s	3.070 s	-14.5% (1.17× faster) 🟢	107.43B	92.75B	-13.7% (1.16× fewer) 🟢	1.034 s	1.031 s	-0.4%	8.90 MiB	9.86 MiB	+10.7% (1.11× larger) ⚠️	53.3 ms	61.2 ms	+14.9% (1.15× slower) ⚠️	3.98 GiB	3.56 GiB	-10.7% (1.12× smaller) 🟢

Workflow logs

argument-ci-bot · 2026-08-14T17:55:08Z

`!benchmark` — main vs `b6a4f22`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	37.550 s	37.412 s	-0.4%	41.070	41.220	+0.4%	73.20 GiB	70.64 GiB	-3.5% 🟢	9.577 s	9.391 s	-1.9%	138.9 ms	151.1 ms	+8.8% (1.09× slower) ⚠️	21.54 MiB	23.38 MiB	+8.5% (1.09× larger) ⚠️	135.19B	130.81B	-3.2% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	37.157 s	34.546 s	-7.0% (1.08× faster) 🟢	72.420	77.900	+7.6% (1.08× faster) 🟢	72.13 GiB	67.22 GiB	-6.8% (1.07× smaller) 🟢	9.613 s	9.122 s	-5.1% (1.05× faster) 🟢	197.0 ms	148.7 ms	-24.5% (1.33× faster) 🟢	21.63 MiB	23.47 MiB	+8.5% (1.09× larger) ⚠️	136.87B	126.89B	-7.3% (1.08× fewer) 🟢
`Char.ofOrdinal_le_of_le`	28.515 s	28.454 s	-0.2%	92.790	92.990	+0.2%	58.03 GiB	58.51 GiB	+0.8%	6.602 s	6.421 s	-2.8%	131.8 ms	150.6 ms	+14.2% (1.14× slower) ⚠️	21.58 MiB	23.42 MiB	+8.5% (1.09× larger) ⚠️	98.00B	90.63B	-7.5% (1.08× fewer) 🟢
`Vector.extract_append._proof_2`	21.362 s	20.460 s	-4.2% 🟢	60.950	63.640	+4.4% 🟢	39.42 GiB	38.21 GiB	-3.1% 🟢	5.182 s	5.110 s	-1.4%	128.2 ms	152.2 ms	+18.7% (1.19× slower) ⚠️	21.24 MiB	23.08 MiB	+8.7% (1.09× larger) ⚠️	76.48B	73.31B	-4.1% 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.040 s	15.965 s	-6.3% (1.07× faster) 🟢	106.340	113.500	+6.7% (1.07× faster) 🟢	34.22 GiB	31.82 GiB	-7.0% (1.08× smaller) 🟢	3.560 s	3.361 s	-5.6% (1.06× faster) 🟢	134.9 ms	143.8 ms	+6.6% (1.07× slower) ⚠️	21.42 MiB	23.26 MiB	+8.6% (1.09× larger) ⚠️	54.00B	49.57B	-8.2% (1.09× fewer) 🟢
`String.split`	16.080 s	15.024 s	-6.6% (1.07× faster) 🟢	109.770	117.480	+7.0% (1.07× faster) 🟢	32.26 GiB	29.72 GiB	-7.9% (1.09× smaller) 🟢	3.243 s	3.158 s	-2.6%	132.5 ms	141.8 ms	+7.0% (1.07× slower) ⚠️	21.62 MiB	23.46 MiB	+8.5% (1.09× larger) ⚠️	49.51B	45.59B	-7.9% (1.09× fewer) 🟢
`List.mergeSort`	11.944 s	11.350 s	-5.0% (1.05× faster) 🟢	121.230	127.570	+5.2% (1.05× faster) 🟢	23.52 GiB	22.26 GiB	-5.4% (1.06× smaller) 🟢	2.435 s	2.308 s	-5.2% (1.05× faster) 🟢	133.7 ms	142.9 ms	+6.9% (1.07× slower) ⚠️	21.48 MiB	23.32 MiB	+8.6% (1.09× larger) ⚠️	35.52B	32.44B	-8.7% (1.09× fewer) 🟢
`Vector.append`	3.982 s	3.778 s	-5.1% (1.05× faster) 🟢	121.780	128.380	+5.4% (1.05× faster) 🟢	7.11 GiB	6.06 GiB	-14.8% (1.17× smaller) 🟢	666.1 ms	647.1 ms	-2.8%	135.2 ms	131.0 ms	-3.1% 🟢	20.08 MiB	21.92 MiB	+9.2% (1.09× larger) ⚠️	7.90B	7.24B	-8.3% (1.09× fewer) 🟢
`Nat.gcd_comm`	3.194 s	3.051 s	-4.5% 🟢	122.110	127.830	+4.7% 🟢	4.92 GiB	4.81 GiB	-2.2%	529.7 ms	504.2 ms	-4.8% (1.05× faster) 🟢	122.8 ms	131.9 ms	+7.4% (1.07× slower) ⚠️	19.84 MiB	21.68 MiB	+9.3% (1.09× larger) ⚠️	5.15B	4.60B	-10.7% (1.12× fewer) 🟢
`String.append`	2.339 s	2.245 s	-4.0% 🟢	129.970	135.390	+4.2% 🟢	4.63 GiB	4.79 GiB	+3.5% ⚠️	400.3 ms	382.6 ms	-4.4% 🟢	120.4 ms	127.3 ms	+5.7% (1.06× slower) ⚠️	19.12 MiB	20.96 MiB	+9.6% (1.10× larger) ⚠️	2.85B	2.55B	-10.5% (1.12× fewer) 🟢
`Int.gcd`	1.991 s	1.956 s	-1.8%	104.490	106.350	+1.8%	5.40 GiB	4.53 GiB	-16.2% (1.19× smaller) 🟢	342.9 ms	340.7 ms	-0.6%	113.8 ms	127.5 ms	+12.0% (1.12× slower) ⚠️	18.67 MiB	20.51 MiB	+9.9% (1.10× larger) ⚠️	1.83B	1.64B	-10.3% (1.12× fewer) 🟢
`Nat.sub_le_of_le_add`	1.788 s	1.741 s	-2.7%	95.060	97.660	+2.7%	4.42 GiB	4.61 GiB	+4.3% ⚠️	337.3 ms	333.3 ms	-1.2%	113.0 ms	128.9 ms	+14.1% (1.14× slower) ⚠️	19.03 MiB	20.87 MiB	+9.7% (1.10× larger) ⚠️	1.58B	1.42B	-10.0% (1.11× fewer) 🟢
`Nat.add_comm`	1.019 s	1.084 s	+6.3% (1.06× slower) ⚠️	41.200	38.740	-6.0% (1.06× slower) ⚠️	4.17 GiB	3.49 GiB	-16.4% (1.20× smaller) 🟢	245.4 ms	246.5 ms	+0.4%	107.8 ms	115.9 ms	+7.4% (1.07× slower) ⚠️	17.20 MiB	19.04 MiB	+10.7% (1.11× larger) ⚠️	268.28M	249.57M	-7.0% (1.07× fewer) 🟢

Workflow logs

gabriel-barrett · 2026-08-13T15:47:07Z

Inline blake3_compress_inner_j. This has the effect of increasing the total width, though not at the same proportion as what we gain on efficiency, for the lookup itself is costly, especially when the input arguments are big.

Eventually we will have to decrease the proof size of the recursive verifier, but there are ways around it see #551

gabriel-barrett · 2026-08-14T15:52:31Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T15:52:48Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T15:56:36Z

`!benchmark` — main vs `69ce736`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `4e89546` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	35.764 s	31.074 s	-13.1% (1.15× faster) 🟢	107.68 GiB	81.50 GiB	-24.3% (1.32× smaller) 🟢	5.30 MiB	6.27 MiB	+18.3% (1.18× larger) ⚠️	33.1 ms	38.1 ms	+15.3% (1.15× slower) ⚠️	5.961 s	4.727 s	-20.7% (1.26× faster) 🟢	180.38B	143.49B	-20.5% (1.26× fewer) 🟢	15.713 s	14.745 s	-6.2% (1.07× faster) 🟢	11.21 MiB	12.14 MiB	+8.3% (1.08× larger) ⚠️	69.0 ms	75.7 ms	+9.7% (1.10× slower) ⚠️	32.74 GiB	29.83 GiB	-8.9% (1.10× smaller) 🟢
`Vector.append`	32.628 s	27.362 s	-16.1% (1.19× faster) 🟢	95.00 GiB	73.59 GiB	-22.5% (1.29× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	32.4 ms	39.9 ms	+23.0% (1.23× slower) ⚠️	5.158 s	4.255 s	-17.5% (1.21× faster) 🟢	157.02B	128.00B	-18.5% (1.23× fewer) 🟢	3.885 s	3.795 s	-2.3%	10.42 MiB	11.34 MiB	+8.9% (1.09× larger) ⚠️	65.7 ms	69.1 ms	+5.1% (1.05× slower) ⚠️	6.80 GiB	6.89 GiB	+1.4%
`Nat.add_comm`	21.106 s	17.709 s	-16.1% (1.19× faster) 🟢	58.08 GiB	44.38 GiB	-23.6% (1.31× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	34.6 ms	38.9 ms	+12.3% (1.12× slower) ⚠️	3.752 s	3.105 s	-17.2% (1.21× faster) 🟢	113.85B	92.75B	-18.5% (1.23× fewer) 🟢	1.024 s	1.034 s	+1.0%	8.93 MiB	9.86 MiB	+10.4% (1.10× larger) ⚠️	52.6 ms	60.8 ms	+15.7% (1.16× slower) ⚠️	3.79 GiB	4.06 GiB	+7.0% (1.07× larger) ⚠️

Workflow logs

argument-ci-bot · 2026-08-14T15:56:40Z

`!benchmark` — main vs `69ce736`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `4e89546` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	39.589 s	38.272 s	-3.3% 🟢	38.950	40.290	+3.4% 🟢	73.66 GiB	70.70 GiB	-4.0% 🟢	9.861 s	9.625 s	-2.4%	131.2 ms	146.6 ms	+11.7% (1.12× slower) ⚠️	21.59 MiB	23.38 MiB	+8.3% (1.08× larger) ⚠️	136.02B	130.81B	-3.8% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	38.143 s	36.140 s	-5.2% (1.06× faster) 🟢	70.550	74.460	+5.5% (1.06× faster) 🟢	73.09 GiB	67.14 GiB	-8.1% (1.09× smaller) 🟢	9.759 s	9.374 s	-3.9% 🟢	136.4 ms	145.1 ms	+6.4% (1.06× slower) ⚠️	21.69 MiB	23.47 MiB	+8.2% (1.08× larger) ⚠️	138.80B	126.89B	-8.6% (1.09× fewer) 🟢
`Char.ofOrdinal_le_of_le`	29.836 s	29.502 s	-1.1%	88.690	89.690	+1.1%	58.53 GiB	58.49 GiB	-0.1%	6.903 s	6.630 s	-4.0% 🟢	137.5 ms	201.3 ms	+46.4% (1.46× slower) ⚠️	21.64 MiB	23.42 MiB	+8.3% (1.08× larger) ⚠️	99.42B	90.63B	-8.8% (1.10× fewer) 🟢
`Vector.extract_append._proof_2`	21.860 s	21.323 s	-2.5%	59.560	61.060	+2.5%	39.70 GiB	38.22 GiB	-3.7% 🟢	5.365 s	5.263 s	-1.9%	130.4 ms	148.2 ms	+13.7% (1.14× slower) ⚠️	21.30 MiB	23.08 MiB	+8.4% (1.08× larger) ⚠️	77.08B	73.31B	-4.9% (1.05× fewer) 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.638 s	16.652 s	-5.6% (1.06× faster) 🟢	102.740	108.810	+5.9% (1.06× faster) 🟢	34.77 GiB	31.77 GiB	-8.6% (1.09× smaller) 🟢	3.634 s	3.537 s	-2.7%	138.0 ms	149.8 ms	+8.5% (1.09× slower) ⚠️	21.47 MiB	23.26 MiB	+8.3% (1.08× larger) ⚠️	54.85B	49.57B	-9.6% (1.11× fewer) 🟢
`String.split`	16.598 s	15.578 s	-6.1% (1.07× faster) 🟢	106.340	113.300	+6.5% (1.07× faster) 🟢	32.71 GiB	29.74 GiB	-9.1% (1.10× smaller) 🟢	3.376 s	3.252 s	-3.7% 🟢	133.0 ms	141.9 ms	+6.7% (1.07× slower) ⚠️	21.68 MiB	23.46 MiB	+8.2% (1.08× larger) ⚠️	50.25B	45.59B	-9.3% (1.10× fewer) 🟢
`List.mergeSort`	12.202 s	11.714 s	-4.0% 🟢	118.670	123.610	+4.2% 🟢	23.77 GiB	22.24 GiB	-6.4% (1.07× smaller) 🟢	2.438 s	2.333 s	-4.3% 🟢	129.5 ms	147.4 ms	+13.7% (1.14× slower) ⚠️	21.53 MiB	23.32 MiB	+8.3% (1.08× larger) ⚠️	36.10B	32.44B	-10.1% (1.11× fewer) 🟢
`Vector.append`	3.968 s	3.886 s	-2.1%	122.230	124.810	+2.1%	6.69 GiB	6.60 GiB	-1.3%	684.6 ms	653.3 ms	-4.6% 🟢	132.6 ms	134.9 ms	+1.7%	20.13 MiB	21.92 MiB	+8.9% (1.09× larger) ⚠️	8.02B	7.24B	-9.7% (1.11× fewer) 🟢
`Nat.gcd_comm`	3.235 s	3.104 s	-4.1% 🟢	120.550	125.640	+4.2% 🟢	5.07 GiB	5.78 GiB	+14.0% (1.14× larger) ⚠️	524.7 ms	507.6 ms	-3.3% 🟢	123.7 ms	133.3 ms	+7.8% (1.08× slower) ⚠️	19.90 MiB	21.68 MiB	+9.0% (1.09× larger) ⚠️	5.25B	4.60B	-12.4% (1.14× fewer) 🟢
`String.append`	2.399 s	2.299 s	-4.2% 🟢	126.740	132.250	+4.3% 🟢	4.62 GiB	4.64 GiB	+0.3%	403.9 ms	393.2 ms	-2.7%	123.5 ms	127.6 ms	+3.3% ⚠️	19.17 MiB	20.96 MiB	+9.3% (1.09× larger) ⚠️	2.90B	2.55B	-12.1% (1.14× fewer) 🟢
`Int.gcd`	1.991 s	2.001 s	+0.5%	104.480	103.950	-0.5%	4.50 GiB	4.53 GiB	+0.7%	355.0 ms	343.3 ms	-3.3% 🟢	118.2 ms	129.5 ms	+9.5% (1.10× slower) ⚠️	18.73 MiB	20.51 MiB	+9.5% (1.10× larger) ⚠️	1.87B	1.64B	-11.9% (1.14× fewer) 🟢
`Nat.sub_le_of_le_add`	1.815 s	1.788 s	-1.5%	93.650	95.090	+1.5%	4.60 GiB	4.62 GiB	+0.3%	343.9 ms	333.7 ms	-2.9%	113.8 ms	128.4 ms	+12.8% (1.13× slower) ⚠️	19.08 MiB	20.87 MiB	+9.4% (1.09× larger) ⚠️	1.60B	1.42B	-11.5% (1.13× fewer) 🟢
`Nat.add_comm`	1.011 s	1.050 s	+3.9% ⚠️	41.550	40	-3.7% ⚠️	3.54 GiB	4.17 GiB	+17.7% (1.18× larger) ⚠️	255.5 ms	251.9 ms	-1.4%	106.7 ms	116.6 ms	+9.3% (1.09× slower) ⚠️	17.25 MiB	19.04 MiB	+10.4% (1.10× larger) ⚠️	271.37M	249.57M	-8.0% (1.09× fewer) 🟢

Workflow logs

Rebased onto the fused xor-rotation branch: the seven inline calls now compose with the two-operand u32_xor_rotr7/12 forms. Codegen regenerated and all 71 kernel FFT pins + the shard aggregate re-measured on top of the new base (median -5.4%, range -11.4%..-0.2%; shard 6.08B -> 5.21B, -14.4%). aiur (cargo + prove + cross), ixvm (pins + parity), multi-stark and recursive-verifier suites pass; fmt, clippy, deny clean.

gabriel-barrett · 2026-08-14T17:51:01Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T17:51:11Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T17:55:05Z

`!benchmark` — main vs `b6a4f22`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	34.284 s	31.716 s	-7.5% (1.08× faster) 🟢	102.25 GiB	81.04 GiB	-20.7% (1.26× smaller) 🟢	5.27 MiB	6.27 MiB	+18.9% (1.19× larger) ⚠️	34.7 ms	37.9 ms	+9.2% (1.09× slower) ⚠️	5.645 s	4.839 s	-14.3% (1.17× faster) 🟢	173.38B	143.49B	-17.2% (1.21× fewer) 🟢	15.805 s	15.068 s	-4.7% 🟢	11.19 MiB	12.14 MiB	+8.5% (1.09× larger) ⚠️	68.0 ms	75.8 ms	+11.6% (1.12× slower) ⚠️	32.25 GiB	29.75 GiB	-7.7% (1.08× smaller) 🟢
`Vector.append`	31.433 s	28.063 s	-10.7% (1.12× faster) 🟢	90.02 GiB	73.45 GiB	-18.4% (1.23× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	32.5 ms	37.4 ms	+15.3% (1.15× slower) ⚠️	4.991 s	4.283 s	-14.2% (1.17× faster) 🟢	152.01B	128.00B	-15.8% (1.19× fewer) 🟢	3.853 s	3.825 s	-0.7%	10.39 MiB	11.34 MiB	+9.2% (1.09× larger) ⚠️	65.2 ms	68.8 ms	+5.5% (1.06× slower) ⚠️	7.23 GiB	6.34 GiB	-12.3% (1.14× smaller) 🟢
`Nat.add_comm`	20.507 s	17.985 s	-12.3% (1.14× faster) 🟢	55.58 GiB	44.60 GiB	-19.8% (1.25× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	34.9 ms	38.4 ms	+10.2% (1.10× slower) ⚠️	3.591 s	3.070 s	-14.5% (1.17× faster) 🟢	107.43B	92.75B	-13.7% (1.16× fewer) 🟢	1.034 s	1.031 s	-0.4%	8.90 MiB	9.86 MiB	+10.7% (1.11× larger) ⚠️	53.3 ms	61.2 ms	+14.9% (1.15× slower) ⚠️	3.98 GiB	3.56 GiB	-10.7% (1.12× smaller) 🟢

Workflow logs

argument-ci-bot · 2026-08-14T17:55:08Z

`!benchmark` — main vs `b6a4f22`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	37.550 s	37.412 s	-0.4%	41.070	41.220	+0.4%	73.20 GiB	70.64 GiB	-3.5% 🟢	9.577 s	9.391 s	-1.9%	138.9 ms	151.1 ms	+8.8% (1.09× slower) ⚠️	21.54 MiB	23.38 MiB	+8.5% (1.09× larger) ⚠️	135.19B	130.81B	-3.2% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	37.157 s	34.546 s	-7.0% (1.08× faster) 🟢	72.420	77.900	+7.6% (1.08× faster) 🟢	72.13 GiB	67.22 GiB	-6.8% (1.07× smaller) 🟢	9.613 s	9.122 s	-5.1% (1.05× faster) 🟢	197.0 ms	148.7 ms	-24.5% (1.33× faster) 🟢	21.63 MiB	23.47 MiB	+8.5% (1.09× larger) ⚠️	136.87B	126.89B	-7.3% (1.08× fewer) 🟢
`Char.ofOrdinal_le_of_le`	28.515 s	28.454 s	-0.2%	92.790	92.990	+0.2%	58.03 GiB	58.51 GiB	+0.8%	6.602 s	6.421 s	-2.8%	131.8 ms	150.6 ms	+14.2% (1.14× slower) ⚠️	21.58 MiB	23.42 MiB	+8.5% (1.09× larger) ⚠️	98.00B	90.63B	-7.5% (1.08× fewer) 🟢
`Vector.extract_append._proof_2`	21.362 s	20.460 s	-4.2% 🟢	60.950	63.640	+4.4% 🟢	39.42 GiB	38.21 GiB	-3.1% 🟢	5.182 s	5.110 s	-1.4%	128.2 ms	152.2 ms	+18.7% (1.19× slower) ⚠️	21.24 MiB	23.08 MiB	+8.7% (1.09× larger) ⚠️	76.48B	73.31B	-4.1% 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.040 s	15.965 s	-6.3% (1.07× faster) 🟢	106.340	113.500	+6.7% (1.07× faster) 🟢	34.22 GiB	31.82 GiB	-7.0% (1.08× smaller) 🟢	3.560 s	3.361 s	-5.6% (1.06× faster) 🟢	134.9 ms	143.8 ms	+6.6% (1.07× slower) ⚠️	21.42 MiB	23.26 MiB	+8.6% (1.09× larger) ⚠️	54.00B	49.57B	-8.2% (1.09× fewer) 🟢
`String.split`	16.080 s	15.024 s	-6.6% (1.07× faster) 🟢	109.770	117.480	+7.0% (1.07× faster) 🟢	32.26 GiB	29.72 GiB	-7.9% (1.09× smaller) 🟢	3.243 s	3.158 s	-2.6%	132.5 ms	141.8 ms	+7.0% (1.07× slower) ⚠️	21.62 MiB	23.46 MiB	+8.5% (1.09× larger) ⚠️	49.51B	45.59B	-7.9% (1.09× fewer) 🟢
`List.mergeSort`	11.944 s	11.350 s	-5.0% (1.05× faster) 🟢	121.230	127.570	+5.2% (1.05× faster) 🟢	23.52 GiB	22.26 GiB	-5.4% (1.06× smaller) 🟢	2.435 s	2.308 s	-5.2% (1.05× faster) 🟢	133.7 ms	142.9 ms	+6.9% (1.07× slower) ⚠️	21.48 MiB	23.32 MiB	+8.6% (1.09× larger) ⚠️	35.52B	32.44B	-8.7% (1.09× fewer) 🟢
`Vector.append`	3.982 s	3.778 s	-5.1% (1.05× faster) 🟢	121.780	128.380	+5.4% (1.05× faster) 🟢	7.11 GiB	6.06 GiB	-14.8% (1.17× smaller) 🟢	666.1 ms	647.1 ms	-2.8%	135.2 ms	131.0 ms	-3.1% 🟢	20.08 MiB	21.92 MiB	+9.2% (1.09× larger) ⚠️	7.90B	7.24B	-8.3% (1.09× fewer) 🟢
`Nat.gcd_comm`	3.194 s	3.051 s	-4.5% 🟢	122.110	127.830	+4.7% 🟢	4.92 GiB	4.81 GiB	-2.2%	529.7 ms	504.2 ms	-4.8% (1.05× faster) 🟢	122.8 ms	131.9 ms	+7.4% (1.07× slower) ⚠️	19.84 MiB	21.68 MiB	+9.3% (1.09× larger) ⚠️	5.15B	4.60B	-10.7% (1.12× fewer) 🟢
`String.append`	2.339 s	2.245 s	-4.0% 🟢	129.970	135.390	+4.2% 🟢	4.63 GiB	4.79 GiB	+3.5% ⚠️	400.3 ms	382.6 ms	-4.4% 🟢	120.4 ms	127.3 ms	+5.7% (1.06× slower) ⚠️	19.12 MiB	20.96 MiB	+9.6% (1.10× larger) ⚠️	2.85B	2.55B	-10.5% (1.12× fewer) 🟢
`Int.gcd`	1.991 s	1.956 s	-1.8%	104.490	106.350	+1.8%	5.40 GiB	4.53 GiB	-16.2% (1.19× smaller) 🟢	342.9 ms	340.7 ms	-0.6%	113.8 ms	127.5 ms	+12.0% (1.12× slower) ⚠️	18.67 MiB	20.51 MiB	+9.9% (1.10× larger) ⚠️	1.83B	1.64B	-10.3% (1.12× fewer) 🟢
`Nat.sub_le_of_le_add`	1.788 s	1.741 s	-2.7%	95.060	97.660	+2.7%	4.42 GiB	4.61 GiB	+4.3% ⚠️	337.3 ms	333.3 ms	-1.2%	113.0 ms	128.9 ms	+14.1% (1.14× slower) ⚠️	19.03 MiB	20.87 MiB	+9.7% (1.10× larger) ⚠️	1.58B	1.42B	-10.0% (1.11× fewer) 🟢
`Nat.add_comm`	1.019 s	1.084 s	+6.3% (1.06× slower) ⚠️	41.200	38.740	-6.0% (1.06× slower) ⚠️	4.17 GiB	3.49 GiB	-16.4% (1.20× smaller) 🟢	245.4 ms	246.5 ms	+0.4%	107.8 ms	115.9 ms	+7.4% (1.07× slower) ⚠️	17.20 MiB	19.04 MiB	+10.7% (1.11× larger) ⚠️	268.28M	249.57M	-7.0% (1.07× fewer) 🟢

Workflow logs

gabriel-barrett · 2026-08-13T15:47:07Z

Inline blake3_compress_inner_j. This has the effect of increasing the total width, though not at the same proportion as what we gain on efficiency, for the lookup itself is costly, especially when the input arguments are big.

Eventually we will have to decrease the proof size of the recursive verifier, but there are ways around it see #551

gabriel-barrett · 2026-08-14T15:52:31Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T15:52:48Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T15:56:36Z

`!benchmark` — main vs `69ce736`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `4e89546` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	35.764 s	31.074 s	-13.1% (1.15× faster) 🟢	107.68 GiB	81.50 GiB	-24.3% (1.32× smaller) 🟢	5.30 MiB	6.27 MiB	+18.3% (1.18× larger) ⚠️	33.1 ms	38.1 ms	+15.3% (1.15× slower) ⚠️	5.961 s	4.727 s	-20.7% (1.26× faster) 🟢	180.38B	143.49B	-20.5% (1.26× fewer) 🟢	15.713 s	14.745 s	-6.2% (1.07× faster) 🟢	11.21 MiB	12.14 MiB	+8.3% (1.08× larger) ⚠️	69.0 ms	75.7 ms	+9.7% (1.10× slower) ⚠️	32.74 GiB	29.83 GiB	-8.9% (1.10× smaller) 🟢
`Vector.append`	32.628 s	27.362 s	-16.1% (1.19× faster) 🟢	95.00 GiB	73.59 GiB	-22.5% (1.29× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	32.4 ms	39.9 ms	+23.0% (1.23× slower) ⚠️	5.158 s	4.255 s	-17.5% (1.21× faster) 🟢	157.02B	128.00B	-18.5% (1.23× fewer) 🟢	3.885 s	3.795 s	-2.3%	10.42 MiB	11.34 MiB	+8.9% (1.09× larger) ⚠️	65.7 ms	69.1 ms	+5.1% (1.05× slower) ⚠️	6.80 GiB	6.89 GiB	+1.4%
`Nat.add_comm`	21.106 s	17.709 s	-16.1% (1.19× faster) 🟢	58.08 GiB	44.38 GiB	-23.6% (1.31× smaller) 🟢	5.30 MiB	6.23 MiB	+17.5% (1.18× larger) ⚠️	34.6 ms	38.9 ms	+12.3% (1.12× slower) ⚠️	3.752 s	3.105 s	-17.2% (1.21× faster) 🟢	113.85B	92.75B	-18.5% (1.23× fewer) 🟢	1.024 s	1.034 s	+1.0%	8.93 MiB	9.86 MiB	+10.4% (1.10× larger) ⚠️	52.6 ms	60.8 ms	+15.7% (1.16× slower) ⚠️	3.79 GiB	4.06 GiB	+7.0% (1.07× larger) ⚠️

Workflow logs

argument-ci-bot · 2026-08-14T15:56:40Z

`!benchmark` — main vs `69ce736`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `4e89546` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	39.589 s	38.272 s	-3.3% 🟢	38.950	40.290	+3.4% 🟢	73.66 GiB	70.70 GiB	-4.0% 🟢	9.861 s	9.625 s	-2.4%	131.2 ms	146.6 ms	+11.7% (1.12× slower) ⚠️	21.59 MiB	23.38 MiB	+8.3% (1.08× larger) ⚠️	136.02B	130.81B	-3.8% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	38.143 s	36.140 s	-5.2% (1.06× faster) 🟢	70.550	74.460	+5.5% (1.06× faster) 🟢	73.09 GiB	67.14 GiB	-8.1% (1.09× smaller) 🟢	9.759 s	9.374 s	-3.9% 🟢	136.4 ms	145.1 ms	+6.4% (1.06× slower) ⚠️	21.69 MiB	23.47 MiB	+8.2% (1.08× larger) ⚠️	138.80B	126.89B	-8.6% (1.09× fewer) 🟢
`Char.ofOrdinal_le_of_le`	29.836 s	29.502 s	-1.1%	88.690	89.690	+1.1%	58.53 GiB	58.49 GiB	-0.1%	6.903 s	6.630 s	-4.0% 🟢	137.5 ms	201.3 ms	+46.4% (1.46× slower) ⚠️	21.64 MiB	23.42 MiB	+8.3% (1.08× larger) ⚠️	99.42B	90.63B	-8.8% (1.10× fewer) 🟢
`Vector.extract_append._proof_2`	21.860 s	21.323 s	-2.5%	59.560	61.060	+2.5%	39.70 GiB	38.22 GiB	-3.7% 🟢	5.365 s	5.263 s	-1.9%	130.4 ms	148.2 ms	+13.7% (1.14× slower) ⚠️	21.30 MiB	23.08 MiB	+8.4% (1.08× larger) ⚠️	77.08B	73.31B	-4.9% (1.05× fewer) 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.638 s	16.652 s	-5.6% (1.06× faster) 🟢	102.740	108.810	+5.9% (1.06× faster) 🟢	34.77 GiB	31.77 GiB	-8.6% (1.09× smaller) 🟢	3.634 s	3.537 s	-2.7%	138.0 ms	149.8 ms	+8.5% (1.09× slower) ⚠️	21.47 MiB	23.26 MiB	+8.3% (1.08× larger) ⚠️	54.85B	49.57B	-9.6% (1.11× fewer) 🟢
`String.split`	16.598 s	15.578 s	-6.1% (1.07× faster) 🟢	106.340	113.300	+6.5% (1.07× faster) 🟢	32.71 GiB	29.74 GiB	-9.1% (1.10× smaller) 🟢	3.376 s	3.252 s	-3.7% 🟢	133.0 ms	141.9 ms	+6.7% (1.07× slower) ⚠️	21.68 MiB	23.46 MiB	+8.2% (1.08× larger) ⚠️	50.25B	45.59B	-9.3% (1.10× fewer) 🟢
`List.mergeSort`	12.202 s	11.714 s	-4.0% 🟢	118.670	123.610	+4.2% 🟢	23.77 GiB	22.24 GiB	-6.4% (1.07× smaller) 🟢	2.438 s	2.333 s	-4.3% 🟢	129.5 ms	147.4 ms	+13.7% (1.14× slower) ⚠️	21.53 MiB	23.32 MiB	+8.3% (1.08× larger) ⚠️	36.10B	32.44B	-10.1% (1.11× fewer) 🟢
`Vector.append`	3.968 s	3.886 s	-2.1%	122.230	124.810	+2.1%	6.69 GiB	6.60 GiB	-1.3%	684.6 ms	653.3 ms	-4.6% 🟢	132.6 ms	134.9 ms	+1.7%	20.13 MiB	21.92 MiB	+8.9% (1.09× larger) ⚠️	8.02B	7.24B	-9.7% (1.11× fewer) 🟢
`Nat.gcd_comm`	3.235 s	3.104 s	-4.1% 🟢	120.550	125.640	+4.2% 🟢	5.07 GiB	5.78 GiB	+14.0% (1.14× larger) ⚠️	524.7 ms	507.6 ms	-3.3% 🟢	123.7 ms	133.3 ms	+7.8% (1.08× slower) ⚠️	19.90 MiB	21.68 MiB	+9.0% (1.09× larger) ⚠️	5.25B	4.60B	-12.4% (1.14× fewer) 🟢
`String.append`	2.399 s	2.299 s	-4.2% 🟢	126.740	132.250	+4.3% 🟢	4.62 GiB	4.64 GiB	+0.3%	403.9 ms	393.2 ms	-2.7%	123.5 ms	127.6 ms	+3.3% ⚠️	19.17 MiB	20.96 MiB	+9.3% (1.09× larger) ⚠️	2.90B	2.55B	-12.1% (1.14× fewer) 🟢
`Int.gcd`	1.991 s	2.001 s	+0.5%	104.480	103.950	-0.5%	4.50 GiB	4.53 GiB	+0.7%	355.0 ms	343.3 ms	-3.3% 🟢	118.2 ms	129.5 ms	+9.5% (1.10× slower) ⚠️	18.73 MiB	20.51 MiB	+9.5% (1.10× larger) ⚠️	1.87B	1.64B	-11.9% (1.14× fewer) 🟢
`Nat.sub_le_of_le_add`	1.815 s	1.788 s	-1.5%	93.650	95.090	+1.5%	4.60 GiB	4.62 GiB	+0.3%	343.9 ms	333.7 ms	-2.9%	113.8 ms	128.4 ms	+12.8% (1.13× slower) ⚠️	19.08 MiB	20.87 MiB	+9.4% (1.09× larger) ⚠️	1.60B	1.42B	-11.5% (1.13× fewer) 🟢
`Nat.add_comm`	1.011 s	1.050 s	+3.9% ⚠️	41.550	40	-3.7% ⚠️	3.54 GiB	4.17 GiB	+17.7% (1.18× larger) ⚠️	255.5 ms	251.9 ms	-1.4%	106.7 ms	116.6 ms	+9.3% (1.09× slower) ⚠️	17.25 MiB	19.04 MiB	+10.4% (1.10× larger) ⚠️	271.37M	249.57M	-8.0% (1.09× fewer) 🟢

Workflow logs

Rebased onto the fused xor-rotation branch: the seven inline calls now compose with the two-operand u32_xor_rotr7/12 forms. Codegen regenerated and all 71 kernel FFT pins + the shard aggregate re-measured on top of the new base (median -5.4%, range -11.4%..-0.2%; shard 6.08B -> 5.21B, -14.4%). aiur (cargo + prove + cross), ixvm (pins + parity), multi-stark and recursive-verifier suites pass; fmt, clippy, deny clean.

gabriel-barrett · 2026-08-14T17:51:01Z

!benchmark aiur-recursive fresh

gabriel-barrett · 2026-08-14T17:51:11Z

!benchmark aiur fresh

argument-ci-bot · 2026-08-14T17:55:05Z

`!benchmark` — main vs `b6a4f22`

backends: aiur-recursive · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur-recursive` · `InitStd` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

3 constants · 3 with regressions · 3 with improvements (|Δ| > 3.0% on any metric).

constant	recursive-prove-time (main)	recursive-prove-time (PR)	Δ%	recursive-peak-ram (main)	recursive-peak-ram (PR)	Δ%	recursive-proof-size (main)	recursive-proof-size (PR)	Δ%	recursive-verify-time (main)	recursive-verify-time (PR)	Δ%	recursive-execute-time (main)	recursive-execute-time (PR)	Δ%	recursive-fft-cost (main)	recursive-fft-cost (PR)	Δ%	prove-time (main)	prove-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%
`String.split`	34.284 s	31.716 s	-7.5% (1.08× faster) 🟢	102.25 GiB	81.04 GiB	-20.7% (1.26× smaller) 🟢	5.27 MiB	6.27 MiB	+18.9% (1.19× larger) ⚠️	34.7 ms	37.9 ms	+9.2% (1.09× slower) ⚠️	5.645 s	4.839 s	-14.3% (1.17× faster) 🟢	173.38B	143.49B	-17.2% (1.21× fewer) 🟢	15.805 s	15.068 s	-4.7% 🟢	11.19 MiB	12.14 MiB	+8.5% (1.09× larger) ⚠️	68.0 ms	75.8 ms	+11.6% (1.12× slower) ⚠️	32.25 GiB	29.75 GiB	-7.7% (1.08× smaller) 🟢
`Vector.append`	31.433 s	28.063 s	-10.7% (1.12× faster) 🟢	90.02 GiB	73.45 GiB	-18.4% (1.23× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	32.5 ms	37.4 ms	+15.3% (1.15× slower) ⚠️	4.991 s	4.283 s	-14.2% (1.17× faster) 🟢	152.01B	128.00B	-15.8% (1.19× fewer) 🟢	3.853 s	3.825 s	-0.7%	10.39 MiB	11.34 MiB	+9.2% (1.09× larger) ⚠️	65.2 ms	68.8 ms	+5.5% (1.06× slower) ⚠️	7.23 GiB	6.34 GiB	-12.3% (1.14× smaller) 🟢
`Nat.add_comm`	20.507 s	17.985 s	-12.3% (1.14× faster) 🟢	55.58 GiB	44.60 GiB	-19.8% (1.25× smaller) 🟢	5.27 MiB	6.23 MiB	+18.1% (1.18× larger) ⚠️	34.9 ms	38.4 ms	+10.2% (1.10× slower) ⚠️	3.591 s	3.070 s	-14.5% (1.17× faster) 🟢	107.43B	92.75B	-13.7% (1.16× fewer) 🟢	1.034 s	1.031 s	-0.4%	8.90 MiB	9.86 MiB	+10.7% (1.11× larger) ⚠️	53.3 ms	61.2 ms	+14.9% (1.15× slower) ⚠️	3.98 GiB	3.56 GiB	-10.7% (1.12× smaller) 🟢

Workflow logs

argument-ci-bot · 2026-08-14T17:55:08Z

`!benchmark` — main vs `b6a4f22`

backends: aiur=prove · envs: InitStd · set: primary · shard: 0 · baseline: fresh (base-SHA run, bencher bypassed)

`aiur` · `InitStd` · `prove` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

13 constants · 13 with regressions · 13 with improvements (|Δ| > 3.0% on any metric).

comparison table (13 constants)

constant	prove-time (main)	prove-time (PR)	Δ%	throughput (const/s) (main)	throughput (const/s) (PR)	Δ%	peak-ram (main)	peak-ram (PR)	Δ%	execute-time (main)	execute-time (PR)	Δ%	verify-time (main)	verify-time (PR)	Δ%	proof-size (main)	proof-size (PR)	Δ%	fft-cost (main)	fft-cost (PR)	Δ%
`Array.extract_append`	37.550 s	37.412 s	-0.4%	41.070	41.220	+0.4%	73.20 GiB	70.64 GiB	-3.5% 🟢	9.577 s	9.391 s	-1.9%	138.9 ms	151.1 ms	+8.8% (1.09× slower) ⚠️	21.54 MiB	23.38 MiB	+8.5% (1.09× larger) ⚠️	135.19B	130.81B	-3.2% 🟢
`ByteArray.utf8DecodeChar?_utf8EncodeChar_append`	37.157 s	34.546 s	-7.0% (1.08× faster) 🟢	72.420	77.900	+7.6% (1.08× faster) 🟢	72.13 GiB	67.22 GiB	-6.8% (1.07× smaller) 🟢	9.613 s	9.122 s	-5.1% (1.05× faster) 🟢	197.0 ms	148.7 ms	-24.5% (1.33× faster) 🟢	21.63 MiB	23.47 MiB	+8.5% (1.09× larger) ⚠️	136.87B	126.89B	-7.3% (1.08× fewer) 🟢
`Char.ofOrdinal_le_of_le`	28.515 s	28.454 s	-0.2%	92.790	92.990	+0.2%	58.03 GiB	58.51 GiB	+0.8%	6.602 s	6.421 s	-2.8%	131.8 ms	150.6 ms	+14.2% (1.14× slower) ⚠️	21.58 MiB	23.42 MiB	+8.5% (1.09× larger) ⚠️	98.00B	90.63B	-7.5% (1.08× fewer) 🟢
`Vector.extract_append._proof_2`	21.362 s	20.460 s	-4.2% 🟢	60.950	63.640	+4.4% 🟢	39.42 GiB	38.21 GiB	-3.1% 🟢	5.182 s	5.110 s	-1.4%	128.2 ms	152.2 ms	+18.7% (1.19× slower) ⚠️	21.24 MiB	23.08 MiB	+8.7% (1.09× larger) ⚠️	76.48B	73.31B	-4.1% 🟢
`_private.Init.Data.Range.Polymorphic.SInt.0.Int64.instRxcHasSize_eq`	17.040 s	15.965 s	-6.3% (1.07× faster) 🟢	106.340	113.500	+6.7% (1.07× faster) 🟢	34.22 GiB	31.82 GiB	-7.0% (1.08× smaller) 🟢	3.560 s	3.361 s	-5.6% (1.06× faster) 🟢	134.9 ms	143.8 ms	+6.6% (1.07× slower) ⚠️	21.42 MiB	23.26 MiB	+8.6% (1.09× larger) ⚠️	54.00B	49.57B	-8.2% (1.09× fewer) 🟢
`String.split`	16.080 s	15.024 s	-6.6% (1.07× faster) 🟢	109.770	117.480	+7.0% (1.07× faster) 🟢	32.26 GiB	29.72 GiB	-7.9% (1.09× smaller) 🟢	3.243 s	3.158 s	-2.6%	132.5 ms	141.8 ms	+7.0% (1.07× slower) ⚠️	21.62 MiB	23.46 MiB	+8.5% (1.09× larger) ⚠️	49.51B	45.59B	-7.9% (1.09× fewer) 🟢
`List.mergeSort`	11.944 s	11.350 s	-5.0% (1.05× faster) 🟢	121.230	127.570	+5.2% (1.05× faster) 🟢	23.52 GiB	22.26 GiB	-5.4% (1.06× smaller) 🟢	2.435 s	2.308 s	-5.2% (1.05× faster) 🟢	133.7 ms	142.9 ms	+6.9% (1.07× slower) ⚠️	21.48 MiB	23.32 MiB	+8.6% (1.09× larger) ⚠️	35.52B	32.44B	-8.7% (1.09× fewer) 🟢
`Vector.append`	3.982 s	3.778 s	-5.1% (1.05× faster) 🟢	121.780	128.380	+5.4% (1.05× faster) 🟢	7.11 GiB	6.06 GiB	-14.8% (1.17× smaller) 🟢	666.1 ms	647.1 ms	-2.8%	135.2 ms	131.0 ms	-3.1% 🟢	20.08 MiB	21.92 MiB	+9.2% (1.09× larger) ⚠️	7.90B	7.24B	-8.3% (1.09× fewer) 🟢
`Nat.gcd_comm`	3.194 s	3.051 s	-4.5% 🟢	122.110	127.830	+4.7% 🟢	4.92 GiB	4.81 GiB	-2.2%	529.7 ms	504.2 ms	-4.8% (1.05× faster) 🟢	122.8 ms	131.9 ms	+7.4% (1.07× slower) ⚠️	19.84 MiB	21.68 MiB	+9.3% (1.09× larger) ⚠️	5.15B	4.60B	-10.7% (1.12× fewer) 🟢
`String.append`	2.339 s	2.245 s	-4.0% 🟢	129.970	135.390	+4.2% 🟢	4.63 GiB	4.79 GiB	+3.5% ⚠️	400.3 ms	382.6 ms	-4.4% 🟢	120.4 ms	127.3 ms	+5.7% (1.06× slower) ⚠️	19.12 MiB	20.96 MiB	+9.6% (1.10× larger) ⚠️	2.85B	2.55B	-10.5% (1.12× fewer) 🟢
`Int.gcd`	1.991 s	1.956 s	-1.8%	104.490	106.350	+1.8%	5.40 GiB	4.53 GiB	-16.2% (1.19× smaller) 🟢	342.9 ms	340.7 ms	-0.6%	113.8 ms	127.5 ms	+12.0% (1.12× slower) ⚠️	18.67 MiB	20.51 MiB	+9.9% (1.10× larger) ⚠️	1.83B	1.64B	-10.3% (1.12× fewer) 🟢
`Nat.sub_le_of_le_add`	1.788 s	1.741 s	-2.7%	95.060	97.660	+2.7%	4.42 GiB	4.61 GiB	+4.3% ⚠️	337.3 ms	333.3 ms	-1.2%	113.0 ms	128.9 ms	+14.1% (1.14× slower) ⚠️	19.03 MiB	20.87 MiB	+9.7% (1.10× larger) ⚠️	1.58B	1.42B	-10.0% (1.11× fewer) 🟢
`Nat.add_comm`	1.019 s	1.084 s	+6.3% (1.06× slower) ⚠️	41.200	38.740	-6.0% (1.06× slower) ⚠️	4.17 GiB	3.49 GiB	-16.4% (1.20× smaller) 🟢	245.4 ms	246.5 ms	+0.4%	107.8 ms	115.9 ms	+7.4% (1.07× slower) ⚠️	17.20 MiB	19.04 MiB	+10.7% (1.11× larger) ⚠️	268.28M	249.57M	-7.0% (1.07× fewer) 🟢

Workflow logs

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 2b2a4fd to 1617d60Compare August 13, 2026 16:25

argumentcomputer deleted a comment from argument-ci-botBotAug 13, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 1f5a367 to 2b83e35Compare August 13, 2026 17:24

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 2b83e35 to 3d03094Compare August 13, 2026 21:20

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from d6f0857 to 69ce736Compare August 14, 2026 14:45

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 69ce736 to b6a4f22Compare August 14, 2026 17:50

gabriel-barrett closed this Aug 15, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 2b2a4fd to 1617d60Compare August 13, 2026 16:25

argumentcomputer deleted a comment from argument-ci-botBotAug 13, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 1f5a367 to 2b83e35Compare August 13, 2026 17:24

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 2b83e35 to 3d03094Compare August 13, 2026 21:20

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from d6f0857 to 69ce736Compare August 14, 2026 14:45

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 69ce736 to b6a4f22Compare August 14, 2026 17:50

gabriel-barrett closed this Aug 15, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 2b2a4fd to 1617d60Compare August 13, 2026 16:25

argumentcomputer deleted a comment from argument-ci-botBotAug 13, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 1f5a367 to 2b83e35Compare August 13, 2026 17:24

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 2b83e35 to 3d03094Compare August 13, 2026 21:20

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from d6f0857 to 69ce736Compare August 14, 2026 14:45

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 69ce736 to b6a4f22Compare August 14, 2026 17:50

gabriel-barrett closed this Aug 15, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 2b2a4fd to 1617d60Compare August 13, 2026 16:25

argumentcomputer deleted a comment from argument-ci-botBotAug 13, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 1f5a367 to 2b83e35Compare August 13, 2026 17:24

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 2b83e35 to 3d03094Compare August 13, 2026 21:20

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from d6f0857 to 69ce736Compare August 14, 2026 14:45

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 69ce736 to b6a4f22Compare August 14, 2026 17:50

gabriel-barrett closed this Aug 15, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 2b2a4fd to 1617d60Compare August 13, 2026 16:25

argumentcomputer deleted a comment from argument-ci-botBotAug 13, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 1f5a367 to 2b83e35Compare August 13, 2026 17:24

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 2b83e35 to 3d03094Compare August 13, 2026 21:20

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from d6f0857 to 69ce736Compare August 14, 2026 14:45

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 69ce736 to b6a4f22Compare August 14, 2026 17:50

gabriel-barrett closed this Aug 15, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 2b2a4fd to 1617d60Compare August 13, 2026 16:25

argumentcomputer deleted a comment from argument-ci-botBotAug 13, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 1f5a367 to 2b83e35Compare August 13, 2026 17:24

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 2b83e35 to 3d03094Compare August 13, 2026 21:20

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from d6f0857 to 69ce736Compare August 14, 2026 14:45

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 69ce736 to b6a4f22Compare August 14, 2026 17:50

gabriel-barrett closed this Aug 15, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 2b2a4fd to 1617d60Compare August 13, 2026 16:25

argumentcomputer deleted a comment from argument-ci-botBotAug 13, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 1f5a367 to 2b83e35Compare August 13, 2026 17:24

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 2b83e35 to 3d03094Compare August 13, 2026 21:20

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from d6f0857 to 69ce736Compare August 14, 2026 14:45

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 69ce736 to b6a4f22Compare August 14, 2026 17:50

gabriel-barrett closed this Aug 15, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 2b2a4fd to 1617d60Compare August 13, 2026 16:25

argumentcomputer deleted a comment from argument-ci-botBotAug 13, 2026

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from 1f5a367 to 2b83e35Compare August 13, 2026 17:24

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 2b83e35 to 3d03094Compare August 13, 2026 21:20

This comment was marked as outdated.
Sign in to view

gabriel-barrett force-pushed the perf/inline-blake3-round branch 3 times, most recently from d6f0857 to 69ce736Compare August 14, 2026 14:45

gabriel-barrett force-pushed the perf/inline-blake3-round branch from 69ce736 to b6a4f22Compare August 14, 2026 17:50

gabriel-barrett closed this Aug 15, 2026

Uh oh!

Conversation

gabriel-barrett commented Aug 13, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment was marked as outdated.

This comment was marked as outdated.

This comment was marked as outdated.

This comment was marked as outdated.

This comment was marked as outdated.

This comment was marked as outdated.

gabriel-barrett commented Aug 14, 2026

Uh oh!

gabriel-barrett commented Aug 14, 2026

Uh oh!

argument-ci-botBot commented Aug 14, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

!benchmark — main vs 69ce736

aiur-recursive · InitStd — main from: base run @ 4e89546 (fresh — bencher bypassed)

Uh oh!

argument-ci-botBot commented Aug 14, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

!benchmark — main vs 69ce736

aiur · InitStd · prove — main from: base run @ 4e89546 (fresh — bencher bypassed)

Uh oh!

gabriel-barrett commented Aug 14, 2026

Uh oh!

gabriel-barrett commented Aug 14, 2026

Uh oh!

argument-ci-botBot commented Aug 14, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

!benchmark — main vs b6a4f22

aiur-recursive · InitStd — main from: base run @ 97aa19e (fresh — bencher bypassed)

Uh oh!

argument-ci-botBot commented Aug 14, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

!benchmark — main vs b6a4f22

aiur · InitStd · prove — main from: base run @ 97aa19e (fresh — bencher bypassed)

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

gabriel-barrett commented Aug 13, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment was marked as outdated.

This comment was marked as outdated.

This comment was marked as outdated.

This comment was marked as outdated.

This comment was marked as outdated.

This comment was marked as outdated.

gabriel-barrett commented Aug 14, 2026

Uh oh!

gabriel-barrett commented Aug 14, 2026

Uh oh!

argument-ci-botBot commented Aug 14, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

!benchmark — main vs 69ce736

aiur-recursive · InitStd — main from: base run @ 4e89546 (fresh — bencher bypassed)

Uh oh!

argument-ci-botBot commented Aug 14, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

!benchmark — main vs 69ce736

aiur · InitStd · prove — main from: base run @ 4e89546 (fresh — bencher bypassed)

Uh oh!

gabriel-barrett commented Aug 14, 2026

Uh oh!

gabriel-barrett commented Aug 14, 2026

Uh oh!

argument-ci-botBot commented Aug 14, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

!benchmark — main vs b6a4f22

aiur-recursive · InitStd — main from: base run @ 97aa19e (fresh — bencher bypassed)

gabriel-barrett commented Aug 13, 2026•
edited
Loading

argument-ci-botBot commented Aug 14, 2026•
edited
Loading

`!benchmark` — main vs `69ce736`

`aiur-recursive` · `InitStd` — main from: base run @ `4e89546` (fresh — bencher bypassed)

argument-ci-botBot commented Aug 14, 2026•
edited
Loading

`!benchmark` — main vs `69ce736`

`aiur` · `InitStd` · `prove` — main from: base run @ `4e89546` (fresh — bencher bypassed)

argument-ci-botBot commented Aug 14, 2026•
edited
Loading

`!benchmark` — main vs `b6a4f22`

`aiur-recursive` · `InitStd` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

argument-ci-botBot commented Aug 14, 2026•
edited
Loading

`!benchmark` — main vs `b6a4f22`

`aiur` · `InitStd` · `prove` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

gabriel-barrett commented Aug 13, 2026•
edited
Loading

argument-ci-botBot commented Aug 14, 2026•
edited
Loading

`!benchmark` — main vs `69ce736`

`aiur-recursive` · `InitStd` — main from: base run @ `4e89546` (fresh — bencher bypassed)

argument-ci-botBot commented Aug 14, 2026•
edited
Loading

`!benchmark` — main vs `69ce736`

`aiur` · `InitStd` · `prove` — main from: base run @ `4e89546` (fresh — bencher bypassed)

argument-ci-botBot commented Aug 14, 2026•
edited
Loading

`!benchmark` — main vs `b6a4f22`

`aiur-recursive` · `InitStd` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

argument-ci-botBot commented Aug 14, 2026•
edited
Loading

`!benchmark` — main vs `b6a4f22`

`aiur` · `InitStd` · `prove` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

gabriel-barrett commented Aug 13, 2026•
edited
Loading

argument-ci-botBot commented Aug 14, 2026•
edited
Loading

`!benchmark` — main vs `69ce736`

`aiur-recursive` · `InitStd` — main from: base run @ `4e89546` (fresh — bencher bypassed)

argument-ci-botBot commented Aug 14, 2026•
edited
Loading

`!benchmark` — main vs `69ce736`

`aiur` · `InitStd` · `prove` — main from: base run @ `4e89546` (fresh — bencher bypassed)

argument-ci-botBot commented Aug 14, 2026•
edited
Loading

`!benchmark` — main vs `b6a4f22`

`aiur-recursive` · `InitStd` — main from: base run @ `97aa19e` (fresh — bencher bypassed)

argument-ci-botBot commented Aug 14, 2026•
edited
Loading

`!benchmark` — main vs `b6a4f22`

`aiur` · `InitStd` · `prove` — main from: base run @ `97aa19e` (fresh — bencher bypassed)