sve2

BLAKE3 cryptographic hash for .NET - pure managed C#, SIMD accelerated (AVX-512, AVX2, SSE, ARM SVE2 and NEON), multithreaded, no native dependencies. Faster than SHA-256 and MD5.

💎 REAL-TIME MUSICAL STEM SEPARATION 💎 Legacy S24 Ultra Edition. Optimized for Snapdragon 8 Gen 3 (Cortex-X4). Discontinued due to device theft.

Educational guide for developers to understand Arm Neoverse processors and Armv9 architecture for server and cloud use cases

Use Github Action to build mpv for Windows with latest commit.

Three fast non-cryptographic hashes for Go, one package each: XXH3 (64/128-bit), XXH64 and rapidhash. Bit-identical to their reference C implementations. Generated SSE2/AVX2/AVX-512/NEON/SVE2 kernels, verified by simulation where the host cannot run them. No dependencies outside the standard library.

Dynamically dispatches your functions to the best arch-specific implementation at program startup in a very portable way.

Unified pure-CPU compute engine covering MSC2020 math, graphics & quantum physics. Powered by TableGen, uops silicon simulation, and autonomous SFVM/sfGC runtime, it delivers multi-limb emulation (F8-F256, I/U8-U256) and SIMD steering across AVX10.2/512/APX/AMX, SVE2, and RVV with hardware port contention modeling and LLVM ORCJIT.

Go code generator for fast, allocation-light json.Unmarshaler implementations, plus a zero-copy toolkit for reading, editing and validating raw JSON

Fork used for optimization of crypto algorithms on AArch64

SVE2 SIMD Instruction set examples

Argon2 (RFC 9106) password hashing for .NET in managed C#, with AVX-512, AVX2, NEON and SVE2 kernels

Hardware-accelerated whole-slice SIMD vector primitives for Go (AVX-512, AVX2, NEON, SVE2, RVV) — zero cgo, zero dependencies, 100% portable Go fallback.

SDC excitation engine for arm64 servers — a stress-ng fork that spends every cycle maximizing silent-data-corruption excitation (di/dt, SMT contention, cache/TLB/interconnect pressure, SDC-directed data shaping). Detection is delegated to SDCShield.