Base64 coding and decoding with SIMD instructions (SSE/AVX2/AVX512F/AVX512BW/AVX512VBMI/ARM Neon)
C++ Python Makefile Other
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.


base64 using SIMD instructions


Repository contains code for encoding and decoding base64 using SIMD instructions. Depending on CPU's architecture, vectorized encoding is faster than scalar versions by factor from 2 to 4; decoding is faster 2 .. 2.7 times.

There are several versions of procedures utilizing following instructions sets:

  • SSE,
  • AVX2,
  • AVX512F,
  • AVX512BW,
  • AVX512VBMI,
  • AVX512VL,
  • BMI2, and
  • ARM Neon.

Vectorization approaches were described in a series of articles:

Daniel Lemire and I wrote also paper Faster Base64 Encoding and Decoding Using AVX2 Instructions which was published by ACM Transactiona on the Web.

Performance results from various machines are located in subdirectories results.

Project organization

There are separate subdirectories for both algorithms, however both have the same structure. Each project contains three programs:

  • verify --- does simple validation of particular parts of algorithms,
  • check --- validates whole procedures,
  • speed --- compares speed of different variants of procedures.


command tools instruction sets
make verify, check, speed scalar, SSE, BMI2
make avx2 verify_avx2, check_avx2, speed_avx2 scalar, SSE, BMI2, AVX2
make avx512 verify_avx512, check_avx512, speed_avx512 scalar, SSE, BMI2, AVX2, AVX512F
make avx512bw verify_avx512bw, check_avx512bw, speed_avx512bw scalar, SSE, BMI2, AVX2, AVX512F, AVX512BW
make avx512vbmi verify_avx512vbmi, check_avx512vbmi scalar, SSE, BMI2, AVX2, AVX512F, AVX512BW, AVX512VBMI
make xop verify_xop, check_xop, speed_xop scalar, SSE and AMD XOP
make arm verify_arm, check_arm, speed_arm scalar, ARM Neon

Type make run, make run_avx2, make run_avx512 or make run_avx512bw to run all programs.

BMI2 presence is determined based on /proc/cpuinfo or counterpart. When an AVX2 or AVX512 targets are used then BMI2 is enabled by default.


To compile AVX512 versions of the programs at least GCC 5.3 is required. GCC 4.9.2 hasn't got AVX512 support.

Please download Intel Software Development Emulator in order to run AVX512 variants via make run_avx512, run_avx512bw or run_avx512vbmi. The emulator path should be added to the PATH.

Known problems

Both encoding and decoding don't match the base64 specification, there is no processing of data tail, i.e. encoder never produces '=' chars at the end, and decoder doesn't handle them at all.

All these shortcoming are not present in a brilliant library by Alfred Klomp:

See also