search

ashvardanian / SimSIMD

Up to 200x Faster Inner Products and Vector Similarity — for Python, JavaScript, Rust, C, and Swift, supporting f64, f32, f16 real & complex, i8, and binary vectors using SIMD for both x86 AVX2 & AVX-512 and Arm NEON & SVE 📐
https://ashvardanian.com/posts/simsimd-faster-scipy/
Apache License 2.0
806 stars 42 forks source link

Adjust compilation flags to support SVE and all AVX-512 extensions #4

Closed ashvardanian closed 9 months ago

ashvardanian commented 12 months ago

The setup.py needs to pass custom compilation flags depending on which target architecture (x86 vs Arm) and which compiler (GCC vs LLVM) is used.

ashvardanian commented 12 months ago

@gurgenyegoryan

ashvardanian commented 12 months ago

We must use the most recent compiler possible. When dealing with Ubuntu, the Debian package manager won't be enough. Let's install Homebrew into VMs, and build with GCC 13, which is available even for older Ubuntu 20.04, common these days.