llama.cpp discussion #19480 — CPU inference thread (EPYC 9454P / Ryzen)

Fields

id

sources/llama-cpp-discussion-19480-cpu-inference-thread-epyc-9454p-ryzen

URL

https://github.com/ggml-org/llama.cpp/discussions/19480

Kind

forum

Accessed

2026-10-15

notes

GitHub discussion thread on CPU inference: covers EPYC 9454P/9554 12-channel DDR5 (~460 GB/s) vs dual-channel consumer APUs; measured tok/s figures for MoE and dense models cited by the EPYC/Ryzen records that reference this source.

Notes

GitHub discussion thread on CPU inference: covers EPYC 9454P/9554 12-channel DDR5 (~460 GB/s) vs dual-channel consumer APUs; measured tok/s figures for MoE and dense models cited by the EPYC/Ryzen records that reference this source.

Cross-references

This record references no other records.