| model                          |       size |     params | backend    | ngl | type_k | type_v |  fa |            test |                  t/s |
| ------------------------------ | ---------: | ---------: | ---------- | --: | -----: | -----: | --: | --------------: | -------------------: |
| llama 70B Q8_0                 |  69.82 GiB |    70.55 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |           pp512 |       1895.79 ± 3.18 |
| llama 70B Q8_0                 |  69.82 GiB |    70.55 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |          pp4096 |       1806.35 ± 6.88 |
| llama 70B Q8_0                 |  69.82 GiB |    70.55 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |           tg128 |         20.77 ± 0.01 |
| llama 70B Q8_0                 |  69.82 GiB |    70.55 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |  pp512 @ d16384 |      1405.17 ± 10.84 |
| llama 70B Q8_0                 |  69.82 GiB |    70.55 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 | pp4096 @ d16384 |       1307.50 ± 4.00 |
| llama 70B Q8_0                 |  69.82 GiB |    70.55 B | CUDA       | 999 |   q8_0 |   q8_0 |   1 |  tg128 @ d16384 |         18.44 ± 0.00 |

build: d2462f8f7 (9590)