Chunk 32.0

This chunk transitioned from the intense engineering of Phase 12's memory backpressure fix into a phase of consolidation, documentation, and performance characterization. The first major task was formally recording the complex Phase 12 architecture—including the split GPU proving API, the use-after-free fix in the C++ `prep_msm_thread`, the early a/b/c free optimization, channel capacity auto-scaling, and the partition semaphore permit-through-send fix—into the project's main documentation (`cuzk-project.md`). The summary timeline, stopping points table, E2E test results, and file references were all updated to reflect the new state, and `cuzk.example.toml` was revised with the optimal defaults (`gpu_threads=32`, `partition_workers=12`). The second major effort was a systematic low-memory benchmark sweep across nine configurations (`pw=1/2/5/7/10/12` × `gw=1/2`) to characterize the engine's memory footprint and throughput for smaller systems. The results revealed a clean linear memory scaling formula (`~69 GiB baseline + pw × ~20 GiB`) and demonstrated that `gw=2` provides no throughput benefit below `pw=10` due to synthesis starvation. Concrete deployment guidance was established for 128 GiB (`pw=2 gw=1`, 152s/proof) through 768 GiB systems (`pw=12 gw=2`, 42.5s/proof). The overarching theme of this chunk is the shift from raw optimization to production readiness and configuration characterization. The complex engineering of Phase 12 was captured for future maintainers, and system integrators were provided with a clear, data-driven guide for configuring the engine on hardware with varying memory capacities. All changes were committed as `9bb657e5` on the `feat/cuzk` branch.

From Engineering to Production Readiness: The Consolidation and Characterization of Phase 12 2131 words

Message Articles

Subagent Sessions