@pupposandro: PFlash now run @poolsideai's Laguna-XS.2 (33B-A3B MoE) on a single RTX 3090. - 111 tok/s decode @ short ctx - 128K TTFT…
Summary
PFlash now supports running @poolsideai's Laguna-XS.2 (33B-A3B MoE) on a single RTX 3090, achieving 111 tok/s decode and 5.4x faster prefill than llama.cpp, with NIAH passes up to 131K context.
View Cached Full Text
Cached at: 05/15/26, 11:10 PM
PFlash now run @poolsideai’s Laguna-XS.2 (33B-A3B MoE) on a single RTX 3090.
- 111 tok/s decode @ short ctx
- 128K TTFT in 15.91s, 5.4x faster prefill vs llama.cpp
- NIAH passes every (ctx, keep) point up to 131K
- first MoE target supported by PFlash
- hand-rolled CUDA, ggml only, no libllama
great collab w/ @eisokant, @eric_alcaide, and the rest of the @poolsideai team. looking forward to working more on their great coding models.
repo + GGUF in first comment.
repo: http://github.com/Luce-Org/lucebox-hub… GGUF: http://huggingface.co/Lucebox/Laguna-XS.2-GGUF…
Thanks!
Similar Articles
@pupposandro: https://x.com/pupposandro/status/2054241934164492328
The article announces support for DFlash and PFlash speculative decoding in llama.cpp for AMD Strix Halo iGPUs, demonstrating significant speedups in inference performance using ROCm.
Qwen3.8-Flash-Next on 2x3090 + DDR4: 17 → 25-29 t/s decode with the expert cache PR
User benchmarks and details a GPU-resident expert cache PR in llama.cpp that boosts decode speed for Qwen3.8-Flash-Next on a dual RTX 3090 system from 17 to 25-29 tokens per second.
@pupposandro: 2.5x faster than llama.cpp on Strix Halo. We just shipped DFlash + PFlash for the AMD Ryzen AI MAX+ 395 iGPU (gfx1151, …
A new toolset (DFlash + PFlash) achieves 2.5x faster inference than llama.cpp on AMD Ryzen AI MAX+ 395 iGPU, demonstrating significant speedups for Qwen3.6-27B with 128 GiB unified memory.
I tested freshly merged DFlash in llama.cpp on Qwen 3.6 27B Local AI win. 4.44x faster at 36K context. Here are my findings RTX 6000 PRO.
A user benchmarks the newly merged DFlash speculative decoding method in llama.cpp on Qwen 3.6 27B, achieving up to 4.44x speedup at 36K context compared to baseline, with detailed leaderboard and quality tests.
Deepseek V4 Flash running on RTX 5090 MoE
User shares optimization benchmarks for DeepSeek-V4-Flash (Q2_K) running on an RTX 5090 using a fork of llama.cpp, achieving 21.3 tokens/s generation and 1 million context size.