Reduce the default page size from 256 KB to 32 KB

[nicole-graus]

Description

Lowers the DEFAULT_PAGE_SIZE to 32 KB (1<<15). Regenerates the static zero-init page commitments (blowups 2/4/8) for the new size and updates the page-count and page-boundary tests accordingly.

20-tx ethrex block

Recursion cycles

Page size	Recursion cycles	vs 256 KB	Keccak calls	Inner blob
4 KB	5.594 B	-15.3%	7.57 M	294 MB
8 KB	4.989 B	-24.5%	6.69 M	263 MB
16 KB	4.680 B	-29.2%	6.19 M	244 MB
32 KB	4.608 B	-30.3%	5.95 M	232 MB
64 KB	4.731 B	-28.4%	5.81 M	221 MB
256 KB	6.607 B	—	6.64 M	209 MB
512 KB	9.442 B	+42.9%	8.08 M	205 MB
1 MB	15.305 B	+131.6%	11.16 M	203 MB

Prove Time

Page size	log2	Prove mean	vs 256KB	Verify	Peak RSS
4 KB	12	88.39 s	+2.7%	5.51 s	41.6 GB
8 KB	13	86.21 s	+0.2%	5.02 s	41.6 GB
16 KB	14	85.65 s	-0.5%	4.74 s	41.6 GB
32 KB	15	84.2 s	-2.2%	4.7 s	41.8 GB
64 KB	16	84.59 s	-1.7%	4.41 s	41.6 GB
256 KB	18	86.05 s	—	4.56 s	41.8 GB
512 KB	19	87.20 s	+1.3%	4.81 s	41.7 GB
1 MB	20	88.74 s	+3.1%	5.29 s	41.6 GB
4 MB	22	100.65 s	+17.0%	7.83 s	41.7 GB

[nicole-graus]

/bench-verify

[github-actions]

⏳ Benchmark started on the bench server. The recursion-guest cycle comparison adds guest builds on top of the verifier bench, longer on a cold runner. The bench server is occupied until it finishes.

[github-actions]

Verifier benchmark — ca58d945fc vs main (20 pairs)

=== Verify ABBA result ===

Metric	main	PR	Δ
Verify time (per-side)	2.852s	4.223s	+48.06% 🔴
Proof size	152.60 MiB	239.03 MiB	+56.64% 🔴

Per-side (⚠️ PR REJECTS the baseline's valid proof — likely a VERIFY REGRESSION, not a format change): A/B/B/A cancels machine drift but not proof-specific variance — read the Verify-time Δ as approximate.

  pairs: 20   mean A (PR): 4.223s   mean B (main): 2.852s
  [parametric] paired-t   mean +48.06%   sd 2.42%   se 0.54%
               95% CI: [+46.93%, +49.19%]   (t df=19 = 2.093)
  [robust]     median +48.42%   Wilcoxon W+=210 W-=0  p(exact)=1.9e-06  (z=+3.90)

  run-to-run jitter:    A CV 1.40%   B CV 0.58%        (lower = steadier)
  within-session drift: +0.42% over the run, 1st->2nd half +0.20%

🔴 REAL REGRESSION — PR verifies ~48.06% slower (paired-t and Wilcoxon agree).

_{Drift-free interleaved A/B/B/A measurement. - = PR faster. Trust the verdict when paired-t and Wilcoxon agree.}

---

Recursion guest cycles (main vs PR)

=== Recursion-guest cycle comparison — single query (blowup=2, 1 query) — deterministic to ~±100k cycles ===
REF_B (baseline) origin/main 0893b6b guest=recursion-min.elf
REF_A (PR) ca58d94 ca58d94 guest=recursion-min.elf

Metric	REF_B (baseline)	REF_A (PR)	Δ (A-B)
Guest cycles	41.4M	41.2M	-0.2M (-0.53%)
Keccak calls	3025	2959	-66

note: cycles reproduce to ~±100k (build codegen + proof nondeterminism); treat sub-100k deltas as noise, not signal.

raw (exact integer counts)

ref_b_sha=0893b6b5598800c02fae2d0f5eb5104b74937068 ref_b_elf=recursion-min.elf ref_b_cycles=41381557 ref_b_keccak=3025 ref_b_execute_wall_s=1
ref_a_sha=ca58d945fc2a02a26e41fa98d6b40ef64042d2d1 ref_a_elf=recursion-min.elf ref_a_cycles=41163399 ref_a_keccak=2959 ref_a_execute_wall_s=1
delta_cycles=-218158 delta_keccak=-66

(blowup2 regime unavailable for these refs — see the workflow log.)

(blowup4 regime unavailable for these refs — see the workflow log.)

=== Recursion-guest cycle comparison — 128-bit (blowup=4, 110 queries) — real ethrex block, 4 transfers — deterministic to ~±100k cycles ===
REF_B (baseline) origin/main 0893b6b guest=recursion-cont-blowup4.elf
REF_A (PR) ca58d94 ca58d94 guest=recursion-cont-blowup4.elf

Metric	REF_B (baseline)	REF_A (PR)	Δ (A-B)
Guest cycles	1025.2M	1099.9M	+74.7M (+7.29%)
Keccak calls	1506942	1711085	+204143

note: cycles reproduce to ~±100k (build codegen + proof nondeterminism); treat sub-100k deltas as noise, not signal.

raw (exact integer counts)

ref_b_sha=0893b6b5598800c02fae2d0f5eb5104b74937068 ref_b_elf=recursion-cont-blowup4.elf ref_b_cycles=1025165414 ref_b_keccak=1506942 ref_b_execute_wall_s=17
ref_a_sha=ca58d945fc2a02a26e41fa98d6b40ef64042d2d1 ref_a_elf=recursion-cont-blowup4.elf ref_a_cycles=1099881618 ref_a_keccak=1711085 ref_a_execute_wall_s=18
delta_cycles=74716204 delta_keccak=204143