lhl 8060S llama-2-7b.Q4_K_M Vulkan tg1 decode
48.0 tok/s
Fields
id
benchmarks/lhl-8060s-llama-2-7bq4-k-m-vulkan-tg1Metric
decode_tok_s
Value
48
Unit
tok/s
Model
llama-2-7b.Q4_K_M
Methodology
lhl strix-halo-testing llm-bench (github.com/lhl/strix-halo-testing, llama-2-7b.Q4_K_M/results.jsonl). llama.cpp llama.cpp-vulkan, backend Vulkan, no -fa, ngl=99, tg1 batch-1 decode at depth 0 on Ryzen AI Max+ 395 (Radeon 8060S gfx1151). System RAM peak 12682MiB, avg power 56W, max temp 58.0C. Per-request batch-1.
Measured by
third_party
Reproducible
yes