{ "tunableSpace": { "WORKGROUP_SIZE": [64, 128, 256] }, "cases": [ { "name": "1m_f32", "preset": "smoke", "inputs": { "a": { "dtype": "float32", "shape": [1048576] }, "b": { "dtype": "float32", "shape": [1048576] } }, "outputs": { "c": { "dtype": "bool", "shape": [1048576] } }, "bench": { "metrics": [{ "type": "bandwidth", "value": "(numel(shapes.a) + numel(shapes.b) + numel(shapes.c)) * 4" }] } }, { "name": "8m_f32_dram_bound", "preset": "stress", "inputs": { "a": { "dtype": "float32", "shape": [8388608] }, "b": { "dtype": "float32", "shape": [8388608] } }, "outputs": { "c": { "dtype": "bool", "shape": [8388608] } }, "bench": { "metrics": [{ "type": "bandwidth", "value": "(numel(shapes.a) + numel(shapes.b) + numel(shapes.c)) * 4" }] }, "provenance": { "notes": "Measures float32 equality comparison over a 32 MB boolean output stored as uint32." } }, { "name": "4m-1_f32_same_shape_odd", "inputs": { "a": { "dtype": "float32", "shape": [4194303] }, "b": { "dtype": "float32", "shape": [4194303] } }, "outputs": { "c": { "dtype": "bool", "shape": [4194303] } }, "bench": { "metrics": [ { "type": "bandwidth", "value": "(numel(shapes.a) + numel(shapes.b) + numel(shapes.c)) * 4", "name": "GB/s" } ] } }, { "name": "1m_f32_scalar_rhs_broadcast", "inputs": { "a": { "dtype": "float32", "shape": [1048576] }, "b": { "dtype": "float32", "shape": [1] } }, "outputs": { "c": { "dtype": "bool", "shape": [1048576] } }, "bench": { "metrics": [ { "type": "bandwidth", "value": "(numel(shapes.a) + numel(shapes.b) + numel(shapes.c)) * 4", "name": "GB/s" } ] } }, { "name": "1m_f32_broadcast_scalar_odd_last_dim", "preset": "edge", "inputs": { "a": { "dtype": "float32", "shape": [1024, 1025], "dist": "normal", "seed": 1, "scale": 2 }, "b": { "dtype": "float32", "shape": [1025], "dist": "normal", "seed": 2, "scale": 2 } }, "outputs": { "c": { "dtype": "bool", "shape": [1024, 1025], "dist": "empty" } }, "bench": { "metrics": [ { "type": "bandwidth", "value": "(numel(shapes.a) + numel(shapes.b) + numel(shapes.c)) * 4", "name": "GB/s" } ] } } ] }