TEST #143 RESULTS
03/28/2025 - 10:21 PM
ACCELERATOR
MODEL
49.6
tokens/s
generation
645
ms
time to first token
2130
tokens/s
prompt
547
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 7K62 48-Core Processor (znver2)
RAM
503.5GB
OS
Linux
Kernel Release
5.15.0-122-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #132-Ubuntu SMP Thu Aug 29 13:45:52 UTC 2024
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1317
tokens/s
50.8
tokens/s
797
ms
pp4096+tg256
1949
tokens/s
46.3
tokens/s
2.12
sec
pp2048+tg256
2407
tokens/s
49.0
tokens/s
871
ms
pp2048+tg768
2409
tokens/s
48.7
tokens/s
870
ms
pp1024+tg1024
2722
tokens/s
50.0
tokens/s
396
ms
pp1280+tg3072
2644
tokens/s
48.1
tokens/s
504
ms
pp384+tg1152
3062
tokens/s
50.9
tokens/s
145
ms
pp64+tg1024
2169
tokens/s
51.4
tokens/s
49
ms
pp16+tg1536
491
tokens/s
51.0
tokens/s
52
ms