| { |
| "environment": { |
| "platform": "macOS-27.0.1-arm64-arm-64bit", |
| "machine": "arm64", |
| "metal": { |
| "device_name": "Apple M5 Pro", |
| "max_recommended_working_set_size": 55662788608, |
| "memory_size": 68719476736, |
| "architecture": "applegpu_g17s", |
| "max_buffer_length": 41747087360, |
| "resource_limit": 499000 |
| }, |
| "versions": { |
| "mlx": "0.32.4", |
| "mlx-metal": "0.32.4", |
| "numpy": "2.5.4", |
| "safetensors": "0.8.0", |
| "transformers": "4.57.6", |
| "torch": "2.15.0", |
| "Pillow": "13.0.0", |
| "scikit-image": "0.26.0" |
| }, |
| "precision_environment": { |
| "MLX_ENABLE_TF32": "0", |
| "MLX_CONV_WINOGRAD": "0" |
| } |
| }, |
| "scope": "Real FC1 activations; fused dual GEMM + native precise sigmoid/BF16 boundaries. Compilation excluded, materialization included.", |
| "trials": [ |
| { |
| "tile": [ |
| 64, |
| 64, |
| 128 |
| ], |
| "simdgroups": 4, |
| "max_abs": 0.0, |
| "relative_rmse": 0.0, |
| "equal_fraction": 1.0, |
| "order": [ |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control" |
| ], |
| "control": { |
| "median_ms": 54.30633283685893, |
| "min_ms": 53.88062517158687, |
| "max_ms": 55.719291092827916, |
| "samples_ms": [ |
| 55.719291092827916, |
| 53.958292119205, |
| 55.07374997250736, |
| 54.05362509191036, |
| 54.42695785313845, |
| 55.30574987642467, |
| 54.18570782057941, |
| 53.88062517158687 |
| ] |
| }, |
| "candidate": { |
| "median_ms": 56.32895801682025, |
| "min_ms": 55.50845805555582, |
| "max_ms": 56.85487505979836, |
| "samples_ms": [ |
| 56.68666702695191, |
| 56.48208293132484, |
| 56.59312498755753, |
| 56.03404087014496, |
| 56.85487505979836, |
| 56.175833102315664, |
| 56.1213749460876, |
| 55.50845805555582 |
| ] |
| }, |
| "reduction_pct": -3.7244738768082097 |
| }, |
| { |
| "tile": [ |
| 64, |
| 64, |
| 256 |
| ], |
| "simdgroups": 4, |
| "max_abs": 0.0, |
| "relative_rmse": 0.0, |
| "equal_fraction": 1.0, |
| "order": [ |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control" |
| ], |
| "control": { |
| "median_ms": 54.211645503528416, |
| "min_ms": 53.07895806618035, |
| "max_ms": 54.641125025227666, |
| "samples_ms": [ |
| 53.07895806618035, |
| 53.51012502796948, |
| 53.82495792582631, |
| 54.23129093833268, |
| 54.351292084902525, |
| 54.44195889867842, |
| 54.192000068724155, |
| 54.641125025227666 |
| ] |
| }, |
| "candidate": { |
| "median_ms": 54.6483329962939, |
| "min_ms": 54.23354101367295, |
| "max_ms": 55.29358284547925, |
| "samples_ms": [ |
| 54.65449998155236, |
| 54.343624971807, |
| 54.23354101367295, |
| 55.09308306500316, |
| 54.53879199922085, |
| 55.29358284547925, |
| 54.64216601103544, |
| 54.97041717171669 |
| ] |
| }, |
| "reduction_pct": -0.8055234049980431 |
| }, |
| { |
| "tile": [ |
| 64, |
| 128, |
| 128 |
| ], |
| "simdgroups": 4, |
| "max_abs": 0.0, |
| "relative_rmse": 0.0, |
| "equal_fraction": 1.0, |
| "order": [ |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control" |
| ], |
| "control": { |
| "median_ms": 54.14856213610619, |
| "min_ms": 53.45358303748071, |
| "max_ms": 54.465292021632195, |
| "samples_ms": [ |
| 53.95558406598866, |
| 53.55166690424085, |
| 54.24279114231467, |
| 54.447249975055456, |
| 54.395125014707446, |
| 54.465292021632195, |
| 54.054333129897714, |
| 53.45358303748071 |
| ] |
| }, |
| "candidate": { |
| "median_ms": 56.264353916049004, |
| "min_ms": 55.47845899127424, |
| "max_ms": 57.0768341422081, |
| "samples_ms": [ |
| 56.573959067463875, |
| 55.54545810446143, |
| 57.04212514683604, |
| 55.47845899127424, |
| 56.27166596241295, |
| 57.0768341422081, |
| 56.188875110819936, |
| 56.257041869685054 |
| ] |
| }, |
| "reduction_pct": -3.9073831261199787 |
| }, |
| { |
| "tile": [ |
| 64, |
| 128, |
| 256 |
| ], |
| "simdgroups": 4, |
| "max_abs": 0.0, |
| "relative_rmse": 0.0, |
| "equal_fraction": 1.0, |
| "order": [ |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control" |
| ], |
| "control": { |
| "median_ms": 54.106041439808905, |
| "min_ms": 53.25645813718438, |
| "max_ms": 54.419915890321136, |
| "samples_ms": [ |
| 53.83545905351639, |
| 53.25645813718438, |
| 54.03195787221193, |
| 54.419915890321136, |
| 54.00166613981128, |
| 54.26054215058684, |
| 54.2540829628706, |
| 54.18012500740588 |
| ] |
| }, |
| "candidate": { |
| "median_ms": 54.481583065353334, |
| "min_ms": 53.830333054065704, |
| "max_ms": 55.55591708980501, |
| "samples_ms": [ |
| 54.57108304835856, |
| 54.2239579372108, |
| 53.830333054065704, |
| 54.371041944250464, |
| 54.39208308234811, |
| 55.21841603331268, |
| 55.55591708980501, |
| 54.768916917964816 |
| ] |
| }, |
| "reduction_pct": -0.6940844599806928 |
| }, |
| { |
| "tile": [ |
| 64, |
| 128, |
| 128 |
| ], |
| "simdgroups": 8, |
| "max_abs": 0.0, |
| "relative_rmse": 0.0, |
| "equal_fraction": 1.0, |
| "order": [ |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control" |
| ], |
| "control": { |
| "median_ms": 54.371103993617, |
| "min_ms": 53.39183402247727, |
| "max_ms": 54.588541854172945, |
| "samples_ms": [ |
| 54.588541854172945, |
| 54.46012504398823, |
| 54.064624942839146, |
| 54.12720795720816, |
| 54.30083302780986, |
| 53.39183402247727, |
| 54.49941707774997, |
| 54.44137495942414 |
| ] |
| }, |
| "candidate": { |
| "median_ms": 58.25266649480909, |
| "min_ms": 57.800541864708066, |
| "max_ms": 60.04566606134176, |
| "samples_ms": [ |
| 57.905124966055155, |
| 58.879791060462594, |
| 57.800541864708066, |
| 58.30283300019801, |
| 58.202499989420176, |
| 60.04566606134176, |
| 58.16129199229181, |
| 58.51491680368781 |
| ] |
| }, |
| "reduction_pct": -7.1390172648467365 |
| }, |
| { |
| "tile": [ |
| 64, |
| 128, |
| 256 |
| ], |
| "simdgroups": 8, |
| "max_abs": 0.0, |
| "relative_rmse": 0.0, |
| "equal_fraction": 1.0, |
| "order": [ |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control" |
| ], |
| "control": { |
| "median_ms": 54.36177097726613, |
| "min_ms": 53.352291928604245, |
| "max_ms": 54.752874886617064, |
| "samples_ms": [ |
| 53.91862499527633, |
| 54.35912497341633, |
| 54.45850011892617, |
| 53.352291928604245, |
| 54.752874886617064, |
| 53.97741589695215, |
| 54.36441698111594, |
| 54.52037509530783 |
| ] |
| }, |
| "candidate": { |
| "median_ms": 55.4332920582965, |
| "min_ms": 53.90954203903675, |
| "max_ms": 55.774624925106764, |
| "samples_ms": [ |
| 55.774624925106764, |
| 53.90954203903675, |
| 55.104624945670366, |
| 55.51516707055271, |
| 55.3514170460403, |
| 55.63316703774035, |
| 54.87812520004809, |
| 55.63899991102517 |
| ] |
| }, |
| "reduction_pct": -1.9710930342546673 |
| }, |
| { |
| "tile": [ |
| 128, |
| 128, |
| 128 |
| ], |
| "simdgroups": 8, |
| "max_abs": 0.0, |
| "relative_rmse": 0.0, |
| "equal_fraction": 1.0, |
| "order": [ |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control" |
| ], |
| "control": { |
| "median_ms": 53.93291695509106, |
| "min_ms": 53.539999993517995, |
| "max_ms": 54.6253330539912, |
| "samples_ms": [ |
| 54.0205838624388, |
| 53.539999993517995, |
| 53.84525004774332, |
| 54.28050016053021, |
| 53.790791891515255, |
| 54.6253330539912, |
| 54.07733400352299, |
| 53.810957819223404 |
| ] |
| }, |
| "candidate": { |
| "median_ms": 59.239438036456704, |
| "min_ms": 59.03545906767249, |
| "max_ms": 61.203999910503626, |
| "samples_ms": [ |
| 59.03545906767249, |
| 59.1114170383662, |
| 59.29441703483462, |
| 59.184459038078785, |
| 61.203999910503626, |
| 59.11066709086299, |
| 60.443833004683256, |
| 59.55045809969306 |
| ] |
| }, |
| "reduction_pct": -9.839113812042255 |
| }, |
| { |
| "tile": [ |
| 128, |
| 128, |
| 256 |
| ], |
| "simdgroups": 8, |
| "max_abs": 0.0, |
| "relative_rmse": 0.0, |
| "equal_fraction": 1.0, |
| "order": [ |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control", |
| "control", |
| "candidate", |
| "candidate", |
| "control" |
| ], |
| "control": { |
| "median_ms": 54.6687290770933, |
| "min_ms": 53.769208025187254, |
| "max_ms": 55.67850009538233, |
| "samples_ms": [ |
| 54.58470806479454, |
| 53.769208025187254, |
| 54.445333080366254, |
| 54.449375020340085, |
| 54.752750089392066, |
| 55.09354081004858, |
| 55.03437505103648, |
| 55.67850009538233 |
| ] |
| }, |
| "candidate": { |
| "median_ms": 56.32437497843057, |
| "min_ms": 54.99091697856784, |
| "max_ms": 57.51441698521376, |
| "samples_ms": [ |
| 54.99091697856784, |
| 56.50112498551607, |
| 56.34741601534188, |
| 55.63233280554414, |
| 57.322500040754676, |
| 55.95620907843113, |
| 57.51441698521376, |
| 56.30133394151926 |
| ] |
| }, |
| "reduction_pct": -3.0285062947091657 |
| } |
| ] |
| } |
|
|
