Just got 100 tps on generation, but in total time it around 45-60 t/s in case of prompt processing waiting. Available memory show: GPU KV cache size: 152,671 tokens Maximum concurrency for 131,072 tokens per request: 1.16x amd-smi monitor for this gpu: GPU XCP POWER GPU_T MEM_T GFX_CLK GFX% MEM% ENC% DEC% VRAM_USAGE 3 0 183 W 82 °C 84 °C 3036 MHz 100 % 5 % N/A 0 % 23.6/ 24.0 GB 5 0 161 W 81 °C 88 °C 3101 MHz 100 % 0 % N/A 0 % 23.7/ 24.0 GB 7 0 165 W 78 °C 86 °C 3095 MHz 100 % 1 % N/A...