TEST #13 RESULTS
03/19/2025 - 11:02 PM
ACCELERATOR
MODEL
25.9
tokens/s
generation
2.48
sec
time to first token
563
tokens/s
prompt
181
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 7352 24-Core Processor (znver2)
RAM
125.64GB
OS
Linux
Kernel Release
6.8.0-52-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #53~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Wed Jan 15 19:18:46 UTC 2
RUNTIME
Name
llamafile
Version
0.9.1
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
631
tokens/s
33.4
tokens/s
1.65
sec
pp4096+tg256
476
tokens/s
9.1
tokens/s
8.71
sec
pp2048+tg256
565
tokens/s
22.0
tokens/s
3.67
sec
pp2048+tg768
564
tokens/s
19.9
tokens/s
3.68
sec
pp1024+tg1024
626
tokens/s
28.6
tokens/s
1.67
sec
pp1280+tg3072
619
tokens/s
15.5
tokens/s
2.10
sec
pp384+tg1152
657
tokens/s
33.5
tokens/s
611
ms
pp64+tg1024
584
tokens/s
36.5
tokens/s
135
ms
pp16+tg1536
347
tokens/s
34.7
tokens/s
71
ms