TEST #36 RESULTS
03/28/2025 - 6:02 PM
ACCELERATOR
MODEL
105
tokens/s
generation
121
ms
time to first token
11674
tokens/s
prompt
2167
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 7352 24-Core Processor (znver2)
RAM
127.9GB
OS
Windows
Kernel Release
10.0
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
16551
tokens/s
115
tokens/s
71
ms
pp4096+tg256
11425
tokens/s
105
tokens/s
375
ms
pp2048+tg256
14479
tokens/s
61.6
tokens/s
159
ms
pp2048+tg768
8963
tokens/s
112
tokens/s
238
ms
pp1024+tg1024
16554
tokens/s
113
tokens/s
71
ms
pp1280+tg3072
15426
tokens/s
115
tokens/s
92
ms
pp384+tg1152
16343
tokens/s
113
tokens/s
33
ms
pp64+tg1024
4471
tokens/s
112
tokens/s
24
ms
pp16+tg1536
851
tokens/s
103
tokens/s
28
ms