forked from ggml-org/llama.cpp
-
Notifications
You must be signed in to change notification settings - Fork 8
Pull requests: AMD-Ecosystem/llama.cpp
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
llama: let the backend pick n_ubatch, and default to 2048 on RDNA3.5
#77
opened Jul 29, 2026 by
roberteg16
•
Draft
3 tasks done
tools: add mmq-tune, an MMQ tile-width autotuning harness
#76
opened Jul 29, 2026 by
roberteg16
•
Draft
3 tasks done
ggml-cuda: f32 tall-skinny GEMM kernel for gfx1151 (RDNA3.5)
#74
opened Jul 28, 2026 by
roberteg16
•
Draft
Merge latest upstream changes in master to gfx11.
#72
opened Jul 27, 2026 by
liangliangchang
•
Draft
feat(cuda): fuse activations and residual add into mmv f/q epilogues
#67
opened Jul 24, 2026 by
roberteg16
•
Draft
tests: add MoE MMQ benchmark with routing-distribution generator
#62
opened Jul 20, 2026 by
roberteg16
•
Draft
3 tasks
gfx1151 (Strix Halo): fuse attn_k+v into single MMVQ dispatch
#59
opened Jul 17, 2026 by
jeffli-xilinx
Loading…
7 tasks done
ggml-cuda: GEMM weight row padding + one-time K-padded f16 dequant for prefill
#57
opened Jul 17, 2026 by
roberteg16
•
Draft
4 of 5 tasks
ProTip!
Adding no:label will show everything without a label.