Files
GR-raytracing/benchmarks/hip_psf_2026-09-07/clean_baseline_multi.log
T
wyj e1ec480669 HIP: accelerate PSF accumulation and restore parallel producers
Cooperate across 32 lanes per PSF and use two completion-protected staging slots with complete batch timing. Restore coarse OpenMP event production while serializing shared GPU submissions and direct fallback boundaries.

Add bounded benchmarks, streaming and renderer regressions, and preserve validation evidence and ownership documentation.
2026-09-06 21:38:53 -04:00

11 lines
879 B
Plaintext

PSF cache ready: 64x64 phases, radius 47 px, relative tail 1e-08, tail abs 1e-06, boundary 1e-07, build 0.689 s
Frame 0: finding and prefetching catalog tiles...
Catalog prefetch: 64440 requested, 3 newly loaded (21087 stars), 64437 unavailable in 0.033 s; 4 loader workers
Frame 0: catalog prefetch finished (64440 candidate tiles).
Frame 0: splatting 130882 lens triangles...
Frame 0: catalog splatting finished in 1.9 s; writing image...
Rendered 156837 images from 21087 catalog stars to /tmp/gr-hip-clean-baseline-multi.png (ok; imported lens map)
PSF splats: cached 156837, cached wing-clipped 0, direct fallbacks 0, discarded below min-Y 0
HIP PSF: 156837 events in 10 batches (10 timed); upload 0.000304 s, kernel 1.496493 s, download 0.026518 s
OMP_NUM_THREADS=16 timeout --kill-after=5s 60s --all-sky-catalog 4 1e-8 12.48s user 0.25s system 421% cpu 3.016 total