A May 16, 2026 publication reports llama.cpp throughput tests with MTP for Qwen 3.6 models on three GTX 1080 Ti GPUs.
The publication describes inference benchmarks using llama.cpp and MTP for Qwen 3.6 models, run on three GTX 1080 Ti GPUs. The reported setup uses a Q4_0 K/V cache and draft-MTP flags; the text also gives context sizes and token rates, but those values are not specified in the material available here.
The findings may help engineers compare inference on older GPUs. To verify the figures and test conditions, consult the original publication and check its reported setup, flags, and context sizes; do not extrapolate performance to different hardware or workloads without your own measurements.