onnxruntime
a85977dd - [webgpu] Limit that K must be divisible by 128 to apply dp4a matmul (#24078)

Commit
284 days ago
[webgpu] Limit that K must be divisible by 128 to apply dp4a matmul (#24078) The DP4AMatMulQuantize shader needs to make sure that K is divisible by 128. Otherwise, we need align the scale to have shape [M, ceil(K / 128)]. To simplify the shader, we limit that K must be divisible by 128 to apply dp4a matmul.
Author
Parents
Loading