← sources

llama.cpp Alternatives for GTX 1080

source: reddit (reddit:LocalLLaMA)published 2026-09-14 14:20ingested 2026-09-15 02:40

retained text

I'm just a poor man with a GTX 1080 and a desire to run Local LLMs. I have been running llama.cpp for over a year now, but I want to try other inference backends. The issue is that none of them seem to support sm\_61.

The ones I have tried:

\- SGLang: 20+ Series

\- vLLM: 20+ Series

\- FreeToken: 30+ Series

\- mistral.rs: 20+ Series

Are there any other inference backends that support Pascal cards like mine? Also, I have no plans to upgrade this until 2028 because of my financial situation and because it's working fine otherwise.

document fetched 2026-09-15 00:43 · canonical https://reddit.com/r/LocalLLaMA/comments/1wg4sns/llamacpp_alternatives_for_gtx_1080