7 posts
OxiCUDA 0.5.2 fixes a class of PTX portability bugs that CUDA 11.x silently tolerated and CUDA 12.9+ toolchains reject outright (non-ASCII bytes in generated comments), closes several Windows-specific test and lock-contention bugs, and fixes a numerical-correctness bug in the tensor-network DMRG excited-states solver. 38,675 tests, ~1.30M SLoC, 74 crates.
OxiCUDA 0.5.3 fixes an async-copy race condition where cuMemcpyDtoHAsync/cuMemcpyHtoDAsync and DeviceBuffer::copy_from_host could return before the transfer had actually landed, letting the very next read observe stale or zeroed device/host memory. 38,675 tests, ~1.30M SLoC, 74 crates.
OxiCUDA 0.5.0, the sovereign GPU-compute layer for the COOLJAPAN ecosystem, fixes F64-precision PTX codegen bugs that made ptxas reject elementwise and reduction kernels, closes a mixed-precision GEMM accumulator bug, and hits zero unwrap()/expect() in library code. 38,622 tests, ~1.30M SLoC, 73 crates.
SplitRS 0.3.5 fixes eight correctness issues found splitting a real sqlite3-parser crate: file-backed mod declarations, --split-impl-blocks cross-chunk calls, bitflags type exports, and grouped-import collisions, resolving E0583, E0624, E0425, and E0252. 608 tests passing, up from 570 in 0.3.4.
SplitRS 0.3.4 fixes five correctness issues in --split-nested-mods surfaced by a focused review: E0603 visibility widening for relocated private mods, E0425/E0599 parent-scope glob retention, and E0432/E0659 macro-import ambiguity. 570 tests passing. The AST-based Rust refactoring tool for the COOLJAPAN ecosystem.
OxiArc 0.3.3 is a correctness release: Brotli now round-trips high-entropy and incompressible data byte-for-byte across all quality levels 1–11. A package-merge Huffman code generator and an extended insert-length table fix two long-standing encoder bugs. Pure Rust, on-the-wire compatible, 1,679 tests passing.
OxiFFT 0.1.3 is a correctness patch for the Pure Rust FFTW replacement: it eliminates an infinite-recursion bug in the size-512/1024/4096 SIMD fallback dispatch for non-f32/f64 element types, drops a redundant buffer allocation, and consolidates licensing to a single Apache-2.0 LICENSE.