wyj
e1ec480669
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.
10 lines
873 B
Plaintext
10 lines
873 B
Plaintext
gr-hip-baseline-8192 gr-hip-final-8192 shape (3840, 2160, 3) samples 24883200 HDR_payload_exact True candidate_payload_sha256 423da7b025bf4815a5044e5d81c2f2330138cb8536b8058c19c9cba2d50f2f65
|
|
PNG_exact True
|
|
gr-hip-baseline-indexed gr-hip-final-indexed shape (3840, 2160, 3) samples 24883200 HDR_payload_exact True candidate_payload_sha256 423da7b025bf4815a5044e5d81c2f2330138cb8536b8058c19c9cba2d50f2f65
|
|
PNG_exact True
|
|
gr-hip-baseline-multi gr-hip-final-multi shape (3840, 2160, 3) samples 24883200 HDR_payload_exact True candidate_payload_sha256 a935becded15185af9e680aee9923ba51ecfee451a6bba34634a7e8be9a22ab8
|
|
PNG_exact True
|
|
gr-cpu-multi gr-hip-final-multi shape (3840, 2160, 3) samples 24883200 HDR_payload_exact True candidate_payload_sha256 a935becded15185af9e680aee9923ba51ecfee451a6bba34634a7e8be9a22ab8
|
|
PNG_exact True
|
|
clean baseline vs final: HDR_payload_exact True
|