{ "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" } }, "repeats": 2, "warmup": 2, "scope": "Synthetic BF16 Q/K/V; dense online softmax. Different reduction/rounding may affect generated images.", "trials": [ { "tokens": 65, "bq": 0, "bk": 0, "simdgroups": 0, "median_ms": 0.595208490267396, "min_ms": 0.5699580069631338, "max_ms": 0.6204589735716581, "samples_ms": [ 0.5699580069631338, 0.6204589735716581 ], "max_abs": 0.0, "relative_rmse": 0.0, "equal_fraction": 1.0 }, { "tokens": 65, "bq": 16, "bk": 32, "simdgroups": 2, "median_ms": 0.8040000684559345, "min_ms": 0.7807090878486633, "max_ms": 0.8272910490632057, "samples_ms": [ 0.8272910490632057, 0.7807090878486633 ], "max_abs": 2.03515625, "relative_rmse": 1.1683274507522583, "equal_fraction": 0.0004206731100566685 }, { "tokens": 65, "bq": 16, "bk": 32, "simdgroups": 4, "median_ms": 0.789958517998457, "min_ms": 0.7098750211298466, "max_ms": 0.8700420148670673, "samples_ms": [ 0.8700420148670673, 0.7098750211298466 ], "max_abs": 2.03515625, "relative_rmse": 1.1683274507522583, "equal_fraction": 0.0004206731100566685 }, { "tokens": 65, "bq": 16, "bk": 64, "simdgroups": 2, "median_ms": 1.0451466077938676, "min_ms": 0.9804591536521912, "max_ms": 1.109834061935544, "samples_ms": [ 1.109834061935544, 0.9804591536521912 ], "max_abs": 2.08203125, "relative_rmse": 1.208585262298584, "equal_fraction": 0.00045501376735046506 }, { "tokens": 65, "bq": 16, "bk": 64, "simdgroups": 4, "median_ms": 1.023833523504436, "min_ms": 1.0045000817626715, "max_ms": 1.0431669652462006, "samples_ms": [ 1.0045000817626715, 1.0431669652462006 ], "max_abs": 2.08203125, "relative_rmse": 1.208585262298584, "equal_fraction": 0.00045501376735046506 }, { "tokens": 65, "bq": 16, "bk": 128, "simdgroups": 2, "median_ms": 1.1088334722444415, "min_ms": 1.107207965105772, "max_ms": 1.110458979383111, "samples_ms": [ 1.107207965105772, 1.110458979383111 ], "max_abs": 2.07421875, "relative_rmse": 1.2112047672271729, "equal_fraction": 0.00041638052789494395 }, { "tokens": 65, "bq": 16, "bk": 128, "simdgroups": 4, "median_ms": 1.1818960774689913, "min_ms": 1.180834136903286, "max_ms": 1.1829580180346966, "samples_ms": [ 1.1829580180346966, 1.180834136903286 ], "max_abs": 2.07421875, "relative_rmse": 1.2112047672271729, "equal_fraction": 0.00041638052789494395 }, { "tokens": 65, "bq": 32, "bk": 32, "simdgroups": 2, "median_ms": 1.3260626001283526, "min_ms": 1.2702920939773321, "max_ms": 1.3818331062793732, "samples_ms": [ 1.3818331062793732, 1.2702920939773321 ], "max_abs": 2.03515625, "relative_rmse": 1.1683274507522583, "equal_fraction": 0.0004206731100566685 }, { "tokens": 65, "bq": 32, "bk": 32, "simdgroups": 4, "median_ms": 1.1533955112099648, "min_ms": 1.1415830813348293, "max_ms": 1.1652079410851002, "samples_ms": [ 1.1415830813348293, 1.1652079410851002 ], "max_abs": 2.03515625, "relative_rmse": 1.1683274507522583, "equal_fraction": 0.0004206731100566685 }, { "tokens": 65, "bq": 32, "bk": 64, "simdgroups": 2, "median_ms": 1.3674584915861487, "min_ms": 1.3492919970303774, "max_ms": 1.38562498614192, "samples_ms": [ 1.3492919970303774, 1.38562498614192 ], "max_abs": 2.08203125, "relative_rmse": 1.208585262298584, "equal_fraction": 0.00045501376735046506 }, { "tokens": 65, "bq": 32, "bk": 64, "simdgroups": 4, "median_ms": 1.1829581344500184, "min_ms": 1.1795831378549337, "max_ms": 1.186333131045103, "samples_ms": [ 1.186333131045103, 1.1795831378549337 ], "max_abs": 2.08203125, "relative_rmse": 1.208585262298584, "equal_fraction": 0.00045501376735046506 }, { "tokens": 65, "bq": 32, "bk": 128, "simdgroups": 2, "median_ms": 1.5032080700621009, "min_ms": 1.4863749966025352, "max_ms": 1.5200411435216665, "samples_ms": [ 1.5200411435216665, 1.4863749966025352 ], "max_abs": 2.07421875, "relative_rmse": 1.2112047672271729, "equal_fraction": 0.00041638052789494395 }, { "tokens": 65, "bq": 32, "bk": 128, "simdgroups": 4, "median_ms": 1.6456249868497252, "min_ms": 1.5924999024719, "max_ms": 1.6987500712275505, "samples_ms": [ 1.5924999024719, 1.6987500712275505 ], "max_abs": 2.07421875, "relative_rmse": 1.2112047672271729, "equal_fraction": 0.00041638052789494395 } ] }