fix: keep track of memory accesses since last checkpoint

[shuklaayush]

This pr primarily adds a page_indices_since_checkpoint buffer to track page accesses between checkpoints. This buffer is used to initialize the memory trace heights of the next segment

Resolves INT-5778

[codspeed-hq]

CodSpeed Performance Report

Merging #2332 will degrade performance by 79.78%

_{Comparing fix/e2-checkpoint-unique-memory-access (f1c9fba) with main (d7eab70)}

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

Summary

❌ 8 regressions
✅ 16 untouched
⏩ 36 skipped¹

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	WallTime	benchmark_execute_metered[quicksort]	9.1 ms	44.8 ms	-79.78%
❌	WallTime	benchmark_execute_metered[keccak256]	11.1 ms	46.8 ms	-76.25%
❌	WallTime	benchmark_execute_metered[fibonacci_iterative]	14.3 ms	49.1 ms	-70.92%
❌	WallTime	benchmark_execute_metered[bubblesort]	11.2 ms	46.7 ms	-75.9%
❌	WallTime	benchmark_execute_metered[revm_transfer]	23.9 ms	59.1 ms	-59.47%
❌	WallTime	benchmark_execute_metered[sha256]	9.3 ms	45 ms	-79.4%
❌	WallTime	benchmark_execute_metered[revm_snailtracer]	10 ms	42.2 ms	-76.37%
❌	WallTime	benchmark_execute_metered[fibonacci_recursive]	16.8 ms	52.7 ms	-68.12%

Footnotes

1. 36 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[shuklaayush]

we don't reset access adapter heights here because access adapters are shared between memory and normal traces and it won't be correct to reset them to 0. what we want ideally is to remove only the contribution of memory chips to access adapter heights

[github-actions]

group	app.proof_time_ms	app.cycles	app.cells_used	leaf.proof_time_ms	leaf.cycles	leaf.cells_used
verify_fibair	(+18 [+8.1%]) 241	322,610	2,058,654	-	-	-
fibonacci	(-9 [-0.9%]) 1,019	1,500,209	2,100,402	-	-	-
regex	(+43 [+1.9%]) 2,332	4,137,502	17,695,216	-	-	-
ecrecover	(-4 [-0.5%]) 737	122,859	(-2328 [-0.1%]) 2,262,772	-	-	-
pairing	(-28 [-1.9%]) 1,454	1,745,742	25,408,302	-	-	-

Commit: f1c9fba

Benchmark Workflow

[jonathanpwang]

LGTM

I added a non-debug assert to ensure the buffer len is not exceeded, since it would lead to UB behavior if it was exceeded at runtime. I think branch prediction will mean it doesn't incur a large perf diff.

[shuklaayush]

I added a non-debug assert to ensure the buffer len is not exceeded, since it would lead to UB behavior if it was exceeded at runtime. I think branch prediction will mean it doesn't incur a large perf diff.

We should also add it to the aot version