Describe the enhancement requested
ByteStreamSplitEncode / ByteStreamSplitDecode does not use SIMD instruction when NumStreams == 2.
A performance improvement is to add support for it in ByteStreamSplitEncodeSimd128 / ByteStreamSplitDecodeSimd128.
|
return ByteStreamSplitDecodeScalar<2>(data, width, num_values, stride, out); |
Component(s)
C++