I'm writing fast math library for x86 and ARM
cpufp. A CPU tool for benchmarking the peak of floating points
585sgemm_hsw. This is an implementation of sgemm_kernel on L1d cache.
233conv3x3_m1. This is a demo how to write a high performance convolution run on apple silicon
56spmv. This is a tuned sparse matrix dense vector multiplication(SpMV) library
23fp_diff. This is a tool for comparing two float-point arrays.
3