| {"ts": "2025-10-23T16:39:58Z", "run": "f65c2ebe9ed24f579d9b0ea89bc64c0e", "impl": "xformers_meff", "tags": {"family": "xformers", "backend": "memory_efficient", "compile": "none"}, "wl": {"name": "flux_L128", "batch": 1, "seq_len": 1152, "heads": 24, "head_dim": 128, "dtype": "bfloat16", "device": "cuda", "seed": 0}, "env": {"torch": "2.8.0+cu128", "cuda": "12.8", "gpu": "NVIDIA L4", "sm": "8.9", "py": "3.12.7", "plat": "Linux-5.15.0-1084-aws-x86_64-with-glibc2.31"}, "lat_ms": {"p10": 0.4138489894103259, "p50": 0.4307200142648071, "p90": 0.44611000339500606, "mean": 0.4555642022751272, "iqr": 0.017660000594332814, "raw_times": [0.4138489894103259, 0.4307200142648071, 0.44611000339500606, 0.42845000280067325, 0.5586920015048236], "has_warnings": false, "reps": 5, "warmup": 2}, "compile_ms": 0.3418480046093464, "peak_bytes": 87425024, "ok": true, "absmax": 0.0625, "corr": {"ok": true, "rtol": 0.02, "atol": 0.02, "absmax": 0.0625, "mae": 0.000362396240234375, "mse": 2.9206275939941406e-06, "ref": "sdpa_math_fp32"}, "err": null} | |
| {"ts": "2025-10-23T16:39:58Z", "run": "f65c2ebe9ed24f579d9b0ea89bc64c0e", "impl": "xformers_meff", "tags": {"family": "xformers", "backend": "memory_efficient", "compile": "none"}, "wl": {"name": "flux_L256", "batch": 1, "seq_len": 1280, "heads": 24, "head_dim": 128, "dtype": "bfloat16", "device": "cuda", "seed": 0}, "env": {"torch": "2.8.0+cu128", "cuda": "12.8", "gpu": "NVIDIA L4", "sm": "8.9", "py": "3.12.7", "plat": "Linux-5.15.0-1084-aws-x86_64-with-glibc2.31"}, "lat_ms": {"p10": 0.4239999980200082, "p50": 2.659967984072864, "p90": 2.6715080020949244, "mean": 1.9286082009784877, "iqr": 1.478391990531236, "raw_times": [2.6715080020949244, 2.6944490091409534, 2.659967984072864, 1.1931160115636885, 0.4239999980200082], "has_warnings": true, "reps": 5, "warmup": 2}, "compile_ms": 2.5627249851822853, "peak_bytes": 95027200, "ok": true, "absmax": 0.0625, "corr": {"ok": true, "rtol": 0.02, "atol": 0.02, "absmax": 0.0625, "mae": 0.00035858154296875, "mse": 2.8908252716064453e-06, "ref": "sdpa_math_fp32"}, "err": null} | |
| {"ts": "2025-10-23T16:39:58Z", "run": "f65c2ebe9ed24f579d9b0ea89bc64c0e", "impl": "xformers_meff", "tags": {"family": "xformers", "backend": "memory_efficient", "compile": "none"}, "wl": {"name": "flux_L320", "batch": 1, "seq_len": 1344, "heads": 24, "head_dim": 128, "dtype": "bfloat16", "device": "cuda", "seed": 0}, "env": {"torch": "2.8.0+cu128", "cuda": "12.8", "gpu": "NVIDIA L4", "sm": "8.9", "py": "3.12.7", "plat": "Linux-5.15.0-1084-aws-x86_64-with-glibc2.31"}, "lat_ms": {"p10": 0.5517620011232793, "p50": 0.55704201804474, "p90": 0.5575830000452697, "mean": 0.5856530042365193, "iqr": 0.0008110073395073414, "raw_times": [0.5575830000452697, 0.7051060092635453, 0.55704201804474, 0.5517620011232793, 0.5567719927057624], "has_warnings": false, "reps": 5, "warmup": 2}, "compile_ms": 2.122876001521945, "peak_bytes": 100007936, "ok": true, "absmax": 0.0625, "corr": {"ok": true, "rtol": 0.02, "atol": 0.02, "absmax": 0.0625, "mae": 0.0003566741943359375, "mse": 2.8759241104125977e-06, "ref": "sdpa_math_fp32"}, "err": null} | |
| {"ts": "2025-10-23T16:39:58Z", "run": "f65c2ebe9ed24f579d9b0ea89bc64c0e", "impl": "xformers_meff", "tags": {"family": "xformers", "backend": "memory_efficient", "compile": "none"}, "wl": {"name": "flux_L384", "batch": 1, "seq_len": 1408, "heads": 24, "head_dim": 128, "dtype": "bfloat16", "device": "cuda", "seed": 0}, "env": {"torch": "2.8.0+cu128", "cuda": "12.8", "gpu": "NVIDIA L4", "sm": "8.9", "py": "3.12.7", "plat": "Linux-5.15.0-1084-aws-x86_64-with-glibc2.31"}, "lat_ms": {"p10": 0.4601500113494694, "p50": 0.547812000149861, "p90": 2.7409789909143, "mean": 1.3905902043916285, "iqr": 2.2796589764766395, "raw_times": [2.7409789909143, 2.7426900051068515, 0.547812000149861, 0.4601500113494694, 0.4613200144376606], "has_warnings": true, "reps": 5, "warmup": 2}, "compile_ms": 2.7442800055723637, "peak_bytes": 104726528, "ok": true, "absmax": 0.0625, "corr": {"ok": true, "rtol": 0.02, "atol": 0.02, "absmax": 0.0625, "mae": 0.0003604888916015625, "mse": 2.8759241104125977e-06, "ref": "sdpa_math_fp32"}, "err": null} | |
| {"ts": "2025-10-23T16:39:58Z", "run": "f65c2ebe9ed24f579d9b0ea89bc64c0e", "impl": "xformers_meff", "tags": {"family": "xformers", "backend": "memory_efficient", "compile": "none"}, "wl": {"name": "flux_L448", "batch": 1, "seq_len": 1472, "heads": 24, "head_dim": 128, "dtype": "bfloat16", "device": "cuda", "seed": 0}, "env": {"torch": "2.8.0+cu128", "cuda": "12.8", "gpu": "NVIDIA L4", "sm": "8.9", "py": "3.12.7", "plat": "Linux-5.15.0-1084-aws-x86_64-with-glibc2.31"}, "lat_ms": {"p10": 0.6246439879760146, "p50": 0.7141460082493722, "p90": 0.8494379871990532, "mean": 1.1143721989355981, "iqr": 0.22109498968347907, "raw_times": [0.7141460082493722, 0.6283429975155741, 0.6246439879760146, 0.8494379871990532, 2.7552900137379766], "has_warnings": true, "reps": 5, "warmup": 2}, "compile_ms": 0.6235140026547015, "peak_bytes": 108855296, "ok": true, "absmax": 0.0625, "corr": {"ok": true, "rtol": 0.02, "atol": 0.02, "absmax": 0.0625, "mae": 0.00035858154296875, "mse": 2.86102294921875e-06, "ref": "sdpa_math_fp32"}, "err": null} | |
| {"ts": "2025-10-23T16:39:58Z", "run": "f65c2ebe9ed24f579d9b0ea89bc64c0e", "impl": "xformers_meff", "tags": {"family": "xformers", "backend": "memory_efficient", "compile": "none"}, "wl": {"name": "flux_L512", "batch": 1, "seq_len": 1536, "heads": 24, "head_dim": 128, "dtype": "bfloat16", "device": "cuda", "seed": 0}, "env": {"torch": "2.8.0+cu128", "cuda": "12.8", "gpu": "NVIDIA L4", "sm": "8.9", "py": "3.12.7", "plat": "Linux-5.15.0-1084-aws-x86_64-with-glibc2.31"}, "lat_ms": {"p10": 0.4940209910273552, "p50": 0.4979610093869269, "p90": 0.5128410120960325, "mean": 0.5156752071343362, "iqr": 0.01566999708302319, "raw_times": [0.5763820081483573, 0.4979610093869269, 0.4971710150130093, 0.5128410120960325, 0.4940209910273552], "has_warnings": false, "reps": 5, "warmup": 2}, "compile_ms": 2.7941610023844987, "peak_bytes": 114425856, "ok": true, "absmax": 0.0625, "corr": {"ok": true, "rtol": 0.02, "atol": 0.02, "absmax": 0.0625, "mae": 0.0003566741943359375, "mse": 2.8908252716064453e-06, "ref": "sdpa_math_fp32"}, "err": null} | |