SIMD

Modern processors have capabilities for SIMD (single instruction multiple data) like SSE or AVX instruction sets on x64 or NEON on ARM architecture. Now, in 2022, C/C++ compilers still have problems at automatic vectorization to generate these vector instructions. Assembler or compiler intrinsic functions can be used to enforce the usage ..

• https://en.wikichip.org/wiki/x86/extensions
• https://en.wikichip.org/wiki/arm/versions
• https://en.wikichip.org/wiki/arm_holdings#Microarchitectures
• https://en.wikipedia.org/wiki/Raspberry_Pi#Specifications
  • https://en.wikipedia.org/wiki/Comparison_of_ARMv8-A_processors
  • does contain the L1/L2/L3 cache sizes
  • Raspberry Pi 4 has Broadcom Chip BCM2711

• https://www.intel.com/content/www/us/en/docs/intrinsics-guide/index.html
• https://github.com/VcDevel/Vc
• https://github.com/hayguen/MIPP
  • it's origin https://github.com/aff3ct/MIPP
• SSE/AVX to Neon
  • https://github.com/DLTcollab/sse2neon
  • https://github.com/kunpengcompute/AvxToNeon
• Neon for x86/SSE
  • https://github.com/intel/ARM_NEON_2_x86_SSE
• SIMD Everywhere
  • https://simd-everywhere.github.io/blog/
  • https://github.com/simd-everywhere/simde
  • https://github.com/jpcima/simde