All issues

Search results

cooperative_groups: typed match masks use absolute warp-lane positions

bugSomething isn't workingSomething isn't working

TBDNot yet triaged; scope or disposition undecidedNot yet triaged; scope or disposition undecided

Status: Open.

#1312 In NVlabs/cuda-oxide;

· YunusMutlu opened on Sep 20, 2026

cooperative_groups: block_reduce/block_scan mishandle partial final warps

bugSomething isn't workingSomething isn't working

TBDNot yet triaged; scope or disposition undecidedNot yet triaged; scope or disposition undecided

Status: Open.

#1310 In NVlabs/cuda-oxide;

· YunusMutlu opened on Sep 20, 2026

Migrate LLVM inline asm semantics to upstream Pliron attributes

Status: Open.

#1308 In NVlabs/cuda-oxide;

· uurl opened on Sep 19, 2026

Target feature scan splits LLVM IR on ';', so conjunctive predicates match across the whole module (integer redux + any f32 reads as ReduxF32)

Status: Open.

#1303 In NVlabs/cuda-oxide;

· vyncint opened on Sep 19, 2026

cargo oxide doctor's clang check fails after following its own 'apt install clang-21' remedy

Status: Open.

#1302 In NVlabs/cuda-oxide;

· vyncint opened on Sep 19, 2026

llc is the pipeline anchor, but nothing checks it against the LLVM that wrote the IR

Status: Open.

#1300 In NVlabs/cuda-oxide;

· sachinsharma3191 opened on Sep 18, 2026

intrinsics-gen: runtime_validation = "executed" is unreachable for compact admissions (30 guards, no Runtime-stage producer)

Status: Open.

#1298 In NVlabs/cuda-oxide;

· 0z5a opened on Sep 18, 2026

sparse_mma resolver conflates type-kind with metadata mode, blocking admission of imported intrinsics

Status: Open.

#1296 In NVlabs/cuda-oxide;

· 0z5a opened on Sep 18, 2026

Merged PTX bundles collide on .file indices, so lineinfo/debug builds cannot load through the merged loader

Status: Open.

#1292 In NVlabs/cuda-oxide;

· navneet83 opened on Sep 17, 2026

ltoir_cache: the no-symlink open uses x86-64 flag values, inert on aarch64

Status: Open.

#1290 In NVlabs/cuda-oxide;

· navneet83 opened on Sep 17, 2026

load() reads the process executable, so a #[cuda_module] compiled into a shared object cannot find its own artifact

Status: Open.

#1284 In NVlabs/cuda-oxide;

· xavierforge opened on Sep 17, 2026

feat(wgmma): add m64n64k32 FP8 E4M3 full-drain path

enhancementNew feature or requestNew feature or request

TBDNot yet triaged; scope or disposition undecidedNot yet triaged; scope or disposition undecided

Status: Open.

#1281 In NVlabs/cuda-oxide;

· Vishalkulkarni45 opened on Sep 17, 2026