{ "op": "ai.onnx.Sub", "tunableSpace": { "WORKGROUP_SIZE": [64, 128, 256] }, "cases": [ { "name": "sub-f32-4m", "preset": "smoke", "vars": { "dtype": "float32", "count": 4194304 }, "inputs": { "a": { "shape": [4194304], "dtype": "float32", "dist": "normal", "seed": 302, "scale": 2 }, "b": { "shape": [4194304], "dtype": "float32", "dist": "normal", "seed": 303, "scale": 2 } }, "outputs": { "c": { "shape": [4194304], "dtype": "float32" } }, "bench": { "primary": true, "metrics": [{ "type": "bandwidth", "value": "args.count * dtypeBytes(args.dtype) * 3" }] } }, { "name": "sub-f16-4m-same-shape", "preset": "smoke", "vars": { "dtype": "float16", "count": 4194304 }, "inputs": { "a": { "shape": [4194304], "dtype": "float16", "dist": "normal", "seed": 312, "scale": 2 }, "b": { "shape": [4194304], "dtype": "float16", "dist": "normal", "seed": 313, "scale": 2 } }, "outputs": { "c": { "shape": [4194304], "dtype": "float16" } }, "bench": { "metrics": [{ "type": "bandwidth", "value": "args.count * dtypeBytes(args.dtype) * 3" }] } }, { "name": "sub-f32-rank4-broadcast", "preset": "smoke", "vars": { "dtype": "float32", "count": 2097152 }, "inputs": { "a": { "shape": [8, 128, 32, 64], "dtype": "float32", "dist": "normal", "seed": 322, "scale": 2 }, "b": { "shape": [1, 128, 1, 64], "dtype": "float32", "dist": "normal", "seed": 323, "scale": 2 } }, "outputs": { "c": { "shape": [8, 128, 32, 64], "dtype": "float32" } }, "bench": { "metrics": [{ "type": "bandwidth", "value": "args.count * dtypeBytes(args.dtype) * 2" }] } }, { "name": "sub-f32-scalar-rhs-odd-lastdim", "preset": "smoke", "vars": { "dtype": "float32", "count": 3000009 }, "inputs": { "a": { "shape": [3, 1000003], "dtype": "float32", "dist": "normal", "seed": 332, "scale": 2 }, "b": { "shape": [], "dtype": "float32", "data": { "kind": "values", "values": [0.5] } } }, "outputs": { "c": { "shape": [3, 1000003], "dtype": "float32" } }, "bench": { "metrics": [{ "type": "bandwidth", "value": "args.count * dtypeBytes(args.dtype) * 2" }] } }, { "name": "sub-f32-broadcast-scalar-fallback-rank3-odd", "preset": "stress", "vars": { "dtype": "float32", "count": 3407872 }, "inputs": { "a": { "shape": [512, 512, 13], "dtype": "float32", "dist": "normal", "seed": 371, "scale": 2 }, "b": { "shape": [1, 1, 13], "dtype": "float32", "dist": "normal", "seed": 372, "scale": 2 } }, "outputs": { "c": { "shape": [512, 512, 13], "dtype": "float32", "dist": "empty" } }, "bench": { "metrics": [{ "type": "bandwidth", "value": "args.count * dtypeBytes(args.dtype) * 2" }] } } ] }