m4_multiply_alphatensor
m4_multiply_alphatensor(out, a, b)
Performs 4x4 matrix multiplication using algorithm discovered by AlphaTensor NOTE: currently ~2.5 times slower than m4_multiply
Kind: global function
See: AlphaTensor2022 by Fawzi etAl
| Param | Type |
|---|---|
| out | Array.<number> | Float32Array |
| a | Array.<number> | Float32Array |
| b | Array.<number> | Float32Array |