TEST #48 RESULTS
03/28/2025 - 7:06 PM
ACCELERATOR
MODEL
52.0
tokens/s
generation
888
ms
time to first token
1482
tokens/s
prompt
443
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 7402P 24-Core Processor (znver2)
RAM
251.6GB
OS
Linux
Kernel Release
5.15.0-86-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #96~20.04.1-Ubuntu SMP Thu Sep 21 13:23:37 UTC 2023
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1731
tokens/s
54.7
tokens/s
610
ms
pp4096+tg256
1380
tokens/s
45.9
tokens/s
2.99
sec
pp2048+tg256
1584
tokens/s
51.0
tokens/s
1.31
sec
pp2048+tg768
1578
tokens/s
50.2
tokens/s
1.32
sec
pp1024+tg1024
1707
tokens/s
52.7
tokens/s
618
ms
pp1280+tg3072
1648
tokens/s
48.8
tokens/s
796
ms
pp384+tg1152
1781
tokens/s
54.2
tokens/s
234
ms
pp64+tg1024
1464
tokens/s
55.5
tokens/s
61
ms
pp16+tg1536
469
tokens/s
54.8
tokens/s
51
ms