bluss / matrixmultiply

General matrix multiplication of f32 and f64 matrices in Rust. Supports matrices with general strides.

https://docs.rs/matrixmultiply/

bluss/matrixmultiply Issues

Give the user the ability to allocate all memory himself
Updated 3 months ago4
Allow usage of the crate without alloc
Updated 6 months ago
Transposed operations segfault
Updated 7 months ago1
Bazel: link or target filename contains space
Closed a year ago2
Links and other knowledge about applied matrix multiplication
Updated a year ago
0.3.5 fails to build with `--no-default-features --features cgemm`
Closed a year ago3
Apple M1 SIMD support
Closed a year ago6
Test involving many 6x6 matrices fails randomly on Mac OS
Closed a year ago16
Tiny matmul discrepancy between single- and multi- thread(?)
Closed a year ago19
Use optimal kernel parameters (architectures, matrix layouts)
Updated 3 years ago7
error: array lengths can't depend on generic parameters
Closed 4 years ago5
Integer matrices
Updated 5 years ago9
Building tests fails on non-x86 architectures
Closed 5 years ago8
Use ifunc strategy or other ways to only check target feature existance once
Updated 5 years ago1
Panic when benchmarking with target-feature=sse
Closed 6 years ago4
Revert the workaround for array zeroing
Closed 6 years ago2
Explore performance of _mm256_blend_ps vs _mm256_shuffle_ps
Closed 5 years ago3
Use fma, fused multiply add, for architectures supporting fma
Closed 5 years ago4
Test fma instructions on travis
Closed 5 years ago1
Don't shadow c in sgemm_kernel::kernel_x86_avx
Closed 6 years ago1
Investigate if _mm256_broadcast_ss outperforms _mm256_set1_ps
Closed 6 years ago3
Allow operations on transposed matrices, i.e. Op(A) and Op(B), and DSYRK
Updated 6 years ago8
ICE's on nightly rust: resolving bounds after type-checking
Closed 7 years ago1
SNB Performance
Closed 8 years ago2
Performance regression on nightly
Closed 8 years ago
ref_mat_mul not always the slower version
Closed 8 years ago16