TEST #144 RESULTS
03/28/2025 - 10:21 PM
ACCELERATOR
MODEL
94.5
tokens/s
generation
291
ms
time to first token
4546
tokens/s
prompt
1139
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
125.7GB
OS
Linux
Kernel Release
5.15.0-118-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #128-Ubuntu SMP Fri Jul 5 09:28:59 UTC 2024
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
5575
tokens/s
97.2
tokens/s
194
ms
pp4096+tg256
4099
tokens/s
87.0
tokens/s
1.01
sec
pp2048+tg256
4990
tokens/s
93.4
tokens/s
421
ms
pp2048+tg768
4990
tokens/s
92.5
tokens/s
421
ms
pp1024+tg1024
5628
tokens/s
95.5
tokens/s
192
ms
pp1280+tg3072
5483
tokens/s
91.3
tokens/s
244
ms
pp384+tg1152
5969
tokens/s
97.1
tokens/s
75
ms
pp64+tg1024
3484
tokens/s
98.6
tokens/s
28
ms
pp16+tg1536
694
tokens/s
97.9
tokens/s
33
ms