simdjson / simdjson
Parsing gigabytes of JSON per second : used by Facebook/Meta Velox, the Node.js runtime, ClickHouse, WatermelonDB, Apache Doris, Milvus, StarRocks
simdjson / simdjson
FareedKhan-dev / kimi-k3-in-c
A 2.78-trillion-parameter Kimi K3 running inference on a single CPU in 8.24 GB of RAM. Portable C99: no BLAS, no framework, no GPU.
google / highway
Performance-portable, length-agnostic SIMD with runtime dispatch
OpenNMT / CTranslate2
Fast inference engine for Transformer models
ashvardanian / StringZilla
Up to 100x faster strings for C, C++, CUDA, Python, Rust, Swift, JS, & Go, leveraging NEON, AVX2, AVX-512, SVE, GPGPU, & SWAR to accelerate search, hashing, sorting, edit distances, sketches, and memory ops 🦖
simd-everywhere / simde
Implementations of SIMD instruction sets for systems which don't natively support them.
simdutf / simdutf
Unicode routines (UTF8, UTF16, UTF32) and Base64: billions of characters per second using SSE2, AVX2, NEON, AVX-512, RISC-V Vector Extension, LoongArch64, POWER. Part of Node.js, WebKit/Safari, Ladybird, Chromium, Cloudflare Workers, Ghostty and Bun.
RoaringBitmap / CRoaring
Roaring bitmaps in C (and C++), with SIMD (AVX2, AVX-512 and NEON) optimizations: used by Apache Doris, ClickHouse, Alibaba Tair, Redpanda, YDB and StarRocks
microsoft / DirectXMath
DirectXMath is an all inline SIMD C++ linear algebra library for use in games and graphics apps
jfalcou / eve
Expressive Vector Engine - SIMD in C++ Goes Brrrr
kimwalisch / primesieve
🚀 Fast prime number generator
numpy / x86-simd-sort
C++ template library for high performance SIMD based sorting algorithms
powturbo / TurboPFor-Integer-Compression
Fastest Integer Compression
prabanta-dev / janas
37 tok/s from Qwen3-Next-80B (48 GB) on a 32 GB laptop, 24 without drafts vs 2.8 for llama.cpp; reading 62 vs 16 tok/s. Faster than llama.cpp on all 13 models compared in Q4_K_M. LLMs bigger than your RAM, pure C, CPU or GPU.
sybil-solutions / glm-flash-lite
GLM-5.3-Flash EXL3 on one 24 GB RTX 3090 + DDR4: elastic GPU expert cache, zero-copy experts, AVX2 CPU tier, OpenAI API
sergqwer / strata-nvfp4
Strata fork: Qwen3.8-Flash-Next 125B MoE in NVFP4 on one RTX 20-50 card (12 GB+) and 64 GB of RAM or more. Setup installs our GPTQ quants (huihui abliterated, OrcaRouter) from Hugging Face. CPU experts on AVX-512 or AVX2 (+AVX-VNNI), all experts in RAM from 92 GB, a low-RAM mode below, vision, W4A8 prompts. Windows / CUDA 13.
sybil-solutions / dsv41-flash-offload
DeepSeek-V4.1-Flash EXL3 on one 24 GB RTX 3090 + DDR4 + NVMe: staged-DMA prefill, AVX2 CPU expert tier, elastic VRAM expert cache, Engram on disk, OpenAI API
actual-computer / toks
exact tokenizers ids, as fast as the hardware allows
mattn / tensai
A tiny neural-network framework in pure Go with AVX2 SIMD kernels (GOEXPERIMENT=simd)
wasif-exe / wire
Wire v9: from-scratch ultra-low-latency HFT stack in Rust and C. 24 crates: TCP/IP, AF_XDP, VFIO PCIe DMA, LMAX Disruptor, AVX2/AVX-512 SIMD, flat-array matching engine (11 cyc), HugePage slab (1.5 ns), lock-free MPMC, CME MDP 3.0 SBE, eBPF XDP risk. Zero heap allocs on the hot path. 46/46 tests pass.