{ "op": "com.microsoft.FastGelu", "tunableSpace": { "WORKGROUP_SIZE": [64, 128, 256] }, "cases": [ { "name": "fastgelu-f32-bias-4096x3072", "preset": "smoke", "vars": { "dtype": "float32" }, "inputs": { "X": { "shape": [4096, 3072], "dtype": "float32", "dist": "normal", "seed": 410, "scale": 2 }, "bias": { "shape": [3072], "dtype": "float32", "dist": "normal", "seed": 411, "scale": 1 } }, "outputs": { "Y": { "shape": [4096, 3072], "dtype": "float32" } }, "bench": { "metrics": [{ "type": "bandwidth", "value": "4096 * 3072 * 2 * dtypeBytes(args.dtype)" }] } }, { "name": "fastgelu-f16-bias-4096x3072", "preset": "model", "vars": { "dtype": "float16" }, "inputs": { "X": { "shape": [4096, 3072], "dtype": "float16", "dist": "normal", "seed": 412, "scale": 2 }, "bias": { "shape": [3072], "dtype": "float16", "dist": "normal", "seed": 413, "scale": 1 } }, "outputs": { "Y": { "shape": [4096, 3072], "dtype": "float16" } }, "bench": { "metrics": [{ "type": "bandwidth", "value": "4096 * 3072 * 2 * dtypeBytes(args.dtype)" }] } }, { "name": "fastgelu-f32-scalar-bias-odd-hidden", "preset": "stress", "vars": { "dtype": "float32" }, "inputs": { "X": { "shape": [2097152, 3], "dtype": "float32", "dist": "normal", "seed": 440, "scale": 2 }, "bias": { "shape": [3], "dtype": "float32", "dist": "normal", "seed": 441, "scale": 1 } }, "outputs": { "Y": { "shape": [2097152, 3], "dtype": "float32" } }, "bench": { "metrics": [{ "type": "bandwidth", "value": "2097152 * 3 * 2 * dtypeBytes(args.dtype)" }] } } ] }