-
Notifications
You must be signed in to change notification settings - Fork 15
Pull requests: huawei-csl/pto-kernels
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
examples: add fused_hadamard_quant_a5 and fused_hadamard_quant_b32_a5 (Hadamard + MXFP4 in one launch)
#237
opened Sep 3, 2026 by
Mocchibird
Collaborator
•
Draft
examples/ptodsl: add fast_hadamard_vmi
#230
opened Aug 18, 2026 by
MirkoDeVita98
Collaborator
Loading…
fix(causal_conv1d): make K a dynamic parameter
#212
opened Jul 3, 2026 by
zouzias
Collaborator
Loading…
feature(causal_conv1d): import kernel from jit_cpp
GDN
#195
opened Jun 27, 2026 by
zouzias
Collaborator
Loading…
Add paged attention highperf JIT example
#193
opened Jun 26, 2026 by
MirkoDeVita98
Collaborator
Loading…
Update the skills.md for accuracy guidelines
#175
opened Jun 4, 2026 by
asobczyk
Collaborator
Loading…
Idealized C-V data exchange kernels for A5
A5
#174
opened Jun 1, 2026 by
learning-chip
Collaborator
Loading…
Add Ascend950 pure-vector simulator examples for SiLU and SwiGLU.
A5
#172
opened May 27, 2026 by
learning-chip
Collaborator
Loading…
Minimum demo to highlight cross-core sync API differences
A5
#158
opened May 11, 2026 by
learning-chip
Collaborator
Loading…
1 task
[Feat] Implement doubly-stochastic Sinkhorn normalization kernel
#134
opened Apr 21, 2026 by
Mocchibird
Collaborator
•
Draft
Chunkwise gated linear attention reaching 60~80 TFLOP/s, with step-by-step optimization records
#88
opened Apr 5, 2026 by
learning-chip
Collaborator
Loading…
9 of 17 tasks
compare host vs device-side chunk metadata computation
#84
opened Apr 1, 2026 by
learning-chip
Collaborator
•
Draft
ProTip!
Filter pull requests by the default branch with base:main.