TEST #1204 RESULTS
07/09/2025 - 10:08 PM
ACCELERATOR
MODEL
14.5
tokens/s
generation
12.11
sec
time to first token
115
tokens/s
prompt
52
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen Threadripper 3960X 24-Core Processor (znver2)
RAM
62.7GB
OS
Linux
Kernel Release
6.14.2-300.fc42.x86_64
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #1 SMP PREEMPT_DYNAMIC Thu Apr 10 21:50:55 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
120
tokens/s
15.0
tokens/s
8.63
sec
pp4096+tg256
103
tokens/s
13.0
tokens/s
39.68
sec
pp2048+tg256
112
tokens/s
14.1
tokens/s
18.36
sec
pp2048+tg768
112
tokens/s
14.0
tokens/s
18.33
sec
pp1024+tg1024
118
tokens/s
14.5
tokens/s
8.77
sec
pp1280+tg3072
115
tokens/s
14.2
tokens/s
11.18
sec
pp384+tg1152
120
tokens/s
15.0
tokens/s
3.26
sec
pp64+tg1024
120
tokens/s
15.7
tokens/s
594
ms
pp16+tg1536
114
tokens/s
15.4
tokens/s
202
ms