Granite 3.0 3B
30 records across 1 hardware.
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 345.34 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 613.21 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 866.6 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 866.49 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 1093.52 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 1092.78 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 1170.83 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 1171.68 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 1164.62 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 1169.14 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 1314.74 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 1312.95 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 2363.34 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 2364.72 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 3445.3 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | Q8_0 | llama.cpp (ROCm) | – | 3450.53 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 594.71 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 647.84 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 932.02 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 930.0 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 973.78 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 1168.95 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 1135.67 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 1129.44 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 1970.72 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 1974.53 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 3230.53 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 3236.54 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 4070.07 | – | – | 2048 | 2026-10-05 | community | – | source |
| A800M | Granite 3.0 3B | IQ4_XS | llama.cpp (ROCm) | – | 4073.25 | – | – | 2048 | 2026-10-05 | community | – | source |