What are you looking for?

llama.cpp b9966 for sm-tensor

AI·llama.cpp build b9966 fixes regex recompilations for -sm tensor mode, caching patterns per tensor to reduce CPU overhead on decode threads.

Measuring PCIe transfer under dual GPU with pipeline & tensor llama.cpp

AI·A user shares benchmarks of dual RTX 3090 and Titan RTX cards running llama.cpp with Qwen3.6-27B at 180k context, comparing tensor and pipeline parallel modes for PCIe transfer efficiency.

I benched quad 5060Tis for code generation with Qwen3.6-27B so you don't have to (it's really good)

AI·The author benches a quad 5060Ti setup for Qwen3.6-27B code generation using MTP, reporting 608 tokens/s prefill and 52.2 tokens/s decode at 256k context on a cost-effective build.

next best jump from qwen 3.6 27b

AI·Community discusses upgrades from Qwen3.6-27B, recommending options like DeepSeek V4 Flash or GLM-5.2 requiring 100-250GB VRAM for noticeable performance gains.

I feel like I'm not using my hardware efficiently

AI·User with RTX 4090, 3090Ti, and 128GB RAM struggles to run larger models like 122B due to VRAM limits while Qwen3.6-27B runs efficiently without using system RAM.

Running Qwen3 30B A3B at 50 tok/s on RTX 5060 Ti

AI·Achieves 50-54 tok/s for Qwen3 30B A3B float8 on 16GB RTX 5060 Ti using custom CUDA/C++.

Are EPYC CCDs all you need + benchmarks

AI·Reports EPYC 9374F CCD benchmarks using ik_llama.cpp; finds limited decoding gains vs older 9135 at low thread counts.