TEST #142 RESULTS
03/28/2025 - 10:20 PM
ACCELERATOR
MODEL
77.2
tokens/s
generation
341
ms
time to first token
4006
tokens/s
prompt
968
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 7542 32-Core Processor (znver2)
RAM
125.7GB
OS
Linux
Kernel Release
5.15.0-92-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #102-Ubuntu SMP Wed Jan 10 09:33:48 UTC 2024
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
4940
tokens/s
79.0
tokens/s
220
ms
pp4096+tg256
3404
tokens/s
71.5
tokens/s
1.22
sec
pp2048+tg256
4279
tokens/s
76.4
tokens/s
492
ms
pp2048+tg768
4276
tokens/s
75.7
tokens/s
492
ms
pp1024+tg1024
4939
tokens/s
78.2
tokens/s
220
ms
pp1280+tg3072
4829
tokens/s
74.9
tokens/s
278
ms
pp384+tg1152
5512
tokens/s
79.2
tokens/s
82
ms
pp64+tg1024
3208
tokens/s
80.1
tokens/s
32
ms
pp16+tg1536
669
tokens/s
79.7
tokens/s
36
ms