Files
zeekay b0a5ca07b7 v0.52 — PVM GPU layout
Establish PVM (Platform-VM) GPU substrate scaffold for LP-137
GPU-native compliance:

  - pvm_gpu_layout.hpp     ValidatorSlot, StakeRecord/Op, SlashEvidence,
                            EpochState, PVMRoundDescriptor, ValidatorOp,
                            PVMTransitionResult — all 16-byte aligned with
                            explicit pad to fixed arena strides for
                            byte-stable host/device ABI
  - pvm_cpu_reference.cpp  Deterministic CPU oracle implementing four
                            canonical kernels: ValidatorSetApply,
                            StakeTransition, SlashingTransition,
                            EpochTransition (root composition)
  - pvm_layout_test        12 layout + reference-determinism checks; all
                            pass with byte-identical roots across runs

This closes the layout half of the P-Chain GPU-native gap. v0.53 ships
the matching CUDA + Metal transition kernels and the cross-backend
determinism harness.
2026-04-26 14:37:43 -07:00

2 lines
193 B
Plaintext

/usr/bin/c++ -O3 -DNDEBUG -arch arm64 -Wl,-search_paths_first -Wl,-headerpad_max_install_names "CMakeFiles/pvm-layout-test.dir/test/pvm_layout_test.cpp.o" -o pvm-layout-test liblux_pvm_core.a