llama.cpp b7083 Vulkan: Mistral-7B-Instruct-v0.3-Q8_0 decode 128 tok on amd-radeon-rx-9070
72.72 tok/s (single-request decode, 128 generated tokens, batch 1)
Fields
id
benchmarks/phx-vk-b7083-9070-mistral-7b-tg128Metric
decode_tok_s
Value
72.72
Unit
tok/s (single-request decode, 128 generated tokens, batch 1)
Model
Mistral-7B-Instruct-v0.3-Q8_0
Methodology
Phoronix Test Suite run of llama.cpp b7083 with the Vulkan backend, one shared host (Michael Larabel, 2025-12-08, result 2512070-NE-LLAMACPPV90). Intel Arc B580 on Mesa 26.0-dev ANV, AMD Radeon cards on Mesa 26.0-dev RADV, both under Linux 6.18. Geometric mean of N=3 runs. ONE request: the prompt-processing test prefills a single prompt and the text-generation test decodes a single 128-token completion. NOT an aggregate across concurrent requests. Batch size is the Phoronix llama.cpp default of 1 for these two tests. Quantization is Q8_0 as named in the result title. Power, clocks and thermals were not published. Card: amd-radeon-rx-9070. Model: Mistral-7B-Instruct-v0.3-Q8_0. DECODE of one request, text-generation-128 test, not aggregate.
Measured by
third_party
Reproducible
yes
notes
Geometric mean, N=3, read from the published result SVG llamacpp-vulkan-mistral-7b-tg1. Same-day comparison also measured RTX 5090/5080/5070/5060 Ti and RX 9070 XT on this exact build; those cards already have many rows here and are not re-filed. Vulkan, so not comparable to a CUDA or ROCm number for the same card without saying so.
Notes
Geometric mean, N=3, read from the published result SVG llamacpp-vulkan-mistral-7b-tg1. Same-day comparison also measured RTX 5090/5080/5070/5060 Ti and RX 9070 XT on this exact build; those cards already have many rows here and are not re-filed. Vulkan, so not comparable to a CUDA or ROCm number for the same card without saying so.