Lumpiasty

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 22:33:06 +00:00

cf5eeccdae remove ffmpeg subpath mount

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 22:31:03 +00:00

d8de1ec47b update supervisord image

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 22:26:53 +00:00

714d83eb86 add useful dependencies to supervisord

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 22:08:32 +00:00

bb7e358b25 move llama-swap workspace to /root

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 22:06:15 +00:00

806a6f38d4 add WORKSPACE env var to supervisord image

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 21:44:15 +00:00

89d610886f Use correct arch image of supervisord

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 21:41:52 +00:00

5e566bd6b7 trigger images rebuild on ci changes

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 21:35:26 +00:00

dfbd89d4cf build correct arch supervisord

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 21:34:59 +00:00

802af7a6b3 build correct arch supervisord

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 21:32:58 +00:00

ea888c57fb use supervisord instead of gitops llama.cpp

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 21:31:57 +00:00

a49454e757 use supervisord instead of gitops llama.cpp

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 21:14:33 +00:00

2d980550eb move coredns image under docker/

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 21:10:00 +00:00

ef6a2d4b46 make build-images ci job parallel

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-12 21:03:47 +00:00

23c5b8e51a add supervisord image build

Lumpiasty pushed to fable5/prefetch-experts at Lumpiasty/llama.cpp

2026-07-12 13:59:12 +00:00

8a545c7820 ggml-vulkan: implement host buffer pinning for faster H2D uploads

5e7f6271c0 docs(readme): add usage + benchmark instructions for the MoE-offload optimizations

5f83fbbe7c ggml : size prefetch slots per layer and fix fallback use-after-free

1163cb3493 ggml : overlap offloaded expert weight uploads with compute

20f5994bfe llama : pin mmap-backed CPU weights for faster H2D uploads

Compare 10 commits »

Lumpiasty created branch fable5/prefetch-experts in Lumpiasty/llama.cpp

2026-07-12 13:59:12 +00:00

Lumpiasty created repository Lumpiasty/llama.cpp

2026-07-12 13:50:12 +00:00

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-07 19:56:22 +00:00

82a5f98c0f add Qwythos LLM

Lumpiasty pushed to fresh-start at Lumpiasty/klaster

2026-07-06 22:01:50 +00:00

c7b9f37fe0 update llama-swap image

Lumpiasty pushed to main at Lumpiasty/nixos-config

2026-07-05 21:19:20 +00:00

1c59901eb1 update NetworkManager to 1.58-rc1

b49bbadd4f use native networkmanager option for enabling clat

Compare 2 commits »