matmul-benchmark is a set of matrix multiplying programs that I use for benchmarking different machines that I encounter.
Not the most accurate scientifically (or any ADJECTIVE-cally, for the matter), itz for the funz :)
These are the tests:
- matmul-static: matmul with
static
ally allocated int32 matrices - matmul-O3: same with O3 optimizations turned on
- matmul-O2: same with O2 optimizations turned on
- matmul-malloc: matmul with mallocated int32 matrices
- matmul-double: matmul with double precision floats
- matmul-float: matmul with single precision floats
- matmul-int64: matmul with 64 bit ints
- matmul.py: numpy based matmuls on float{32,64}, int{32,64}
Just
$ git clone https://github.com/bojle/matmul-benchmark
$ make run
and let it warm your processor