geerlingguy/ai-benchmarks issue #39 — RTX 3080 Ti 12GB llama.cpp Vulkan llama-bench (build d82b7a7c1)

Fields

id

sources/geerling-ai-benchmarks-issue-39-rtx3080ti

URL

https://github.com/geerlingguy/ai-benchmarks/issues/39

Publisher

Jeff Geerling

Kind

benchmark

Published

2025-11-28

Accessed

2026-10-17

notes

Same two-host protocol as issue #40. Build d82b7a7c1 (7193), Vulkan, ngl 99. gpt-oss-20b on the 265K PC reduced to pp512/tg128/pp512+tg128 after ErrorOutOfDeviceMemory at a 4096 prompt. Power is whole-system wall draw.

Notes

Same two-host protocol as issue #40. Build d82b7a7c1 (7193), Vulkan, ngl 99. gpt-oss-20b on the 265K PC reduced to pp512/tg128/pp512+tg128 after ErrorOutOfDeviceMemory at a 4096 prompt. Power is whole-system wall draw.

Cross-references

This record references no other records.