llamaperf

LFM3 1.7B

on AMD Strix Halo 128GB · llama.cpp

Tone: positiveAMD hardware
Nov 1, 2026
Throughput
78.0 t/s gen · 4100.0 tokens/s prompt processing (prefill)Prompt processing is how fast the input is read, not how long you wait for the first word. How to measure time to first token.
Quant
Q8_0 (GGUF)

Use cases

summarization

Summary

User reports LFM3-1.7B at 4.1k t/s prefill and about 78 t/s decode on a Strix Halo iGPU. Setup is llama.cpp with Q8_0 GGUF weights on ROCm. Tensor-splitting across a 7900 XTX plus XT gives about 7.1k t/s prefill and about 151 t/s decode. User finds it more coherent at 32k-64k prompts than a 1.7B model should be.

Compare this run

Medians of plain runs: one device, one request, no speculative decoding, the whole model in the device's memory. The report above is one setup, so read it beside them and not as one of them.

LFM3 1.7B at 8-bit by hardware

Bars are medians on one scale, 0 to 329.0 tokens/s. Press ▶ to watch text arrive at a row's speed.

Models most run on the AMD Strix Halo 128GB

Bars are medians on one scale, 0 to 77.0 tokens/s. Each dot is one run and opens its report. Press ▶ to watch text arrive at a row's speed.

Everything reported on the AMD Strix Halo 128GB →Every LFM3 report →Estimate LFM3 on this card →