Repository navigation

All pull requests

Labels (9)Milestones (0)New pull request

Open

(36)

Closed

(7)

Backend: follow-up copies for kernel outputs that cannot share its layout
#44AdvaitTahilyani opened Nov 5, 2026
Solver: tie a fused kernel's kept results to one chare layout
#43AdvaitTahilyani opened Nov 5, 2026
[Draft] Size limits: dimension guard, 64-bit counts and offsets, device-side creation, vecmat
#42AdvaitTahilyani opened Nov 5, 2026
[Draft] Convolve and bincount: shared filter per PE, overlap-save, device histograms
#41AdvaitTahilyani opened Nov 5, 2026
Get: one gather message per tile piece instead of one per row
#40AdvaitTahilyani opened Nov 5, 2026
Indexing: tile-grid scans, device mask packing, device folded offsets, repeat
#39AdvaitTahilyani opened Nov 5, 2026
GPU JIT: cap fused kernels at the CUDA parameter limit
#38AdvaitTahilyani opened Nov 5, 2026
linalg LU: lookahead row swaps; stream-order chunk copies
#37AdvaitTahilyani opened Nov 5, 2026
Linalg: device svd/lstsq, distributed solve_triangular, complex/upper tiled cholesky
#36AdvaitTahilyani opened Nov 5, 2026
Linalg: distributed LU for solve, inv, det, slogdet
#35AdvaitTahilyani opened Nov 5, 2026
Matmul: cuBLAS for complex, stream-ordered int GEMM
#34AdvaitTahilyani opened Nov 5, 2026
RESHAPE/BROADCAST: device messages to other processes
#33AdvaitTahilyani opened Nov 5, 2026
Index ops: event ordering instead of per-message fences; GATHER/SCATTER iterate only meeting chares
#32AdvaitTahilyani opened Nov 5, 2026
SCAN: coordinator fold of line totals
#31AdvaitTahilyani opened Nov 5, 2026
Stream-ordered frees and send ordering without host fences
#30AdvaitTahilyani opened Nov 5, 2026
Sort: N-D sort/argsort along a split axis without gathering
#29AdvaitTahilyani opened Nov 5, 2026
Sort: complex dtypes in the distributed 1-D sample sort
#28AdvaitTahilyani opened Nov 5, 2026
Stream-ordered matvec, dot and axis reductions
#27AdvaitTahilyani opened Nov 4, 2026
Sort with many chares per GPU
#26AdvaitTahilyani opened Nov 4, 2026
Reuse large arrays across repeats
#25AdvaitTahilyani opened Nov 4, 2026
Balanced, deterministic tile decomposition
#24AdvaitTahilyani opened Nov 4, 2026
Support messages and fetches over 2 GiB
#23AdvaitTahilyani opened Nov 4, 2026
Device-resident linear algebra, tiled Cholesky and TSQR
#22AdvaitTahilyani opened Nov 4, 2026
On-device gather, scatter and expand
#21AdvaitTahilyani opened Nov 4, 2026
Batched matmul across GPUs
#20AdvaitTahilyani opened Nov 4, 2026

Pull requests