TEST #19 RESULTS
03/25/2025 - 6:17 PM
ACCELERATOR
MODEL
82.9
tokens/s
generation
620
ms
time to first token
2592
tokens/s
prompt
703
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
127.87GB
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
3041
tokens/s
85.4
tokens/s
349
ms
pp4096+tg256
1656
tokens/s
48.6
tokens/s
2.49
sec
pp2048+tg256
2367
tokens/s
69.6
tokens/s
880
ms
pp2048+tg768
2375
tokens/s
66.0
tokens/s
876
ms
pp1024+tg1024
2955
tokens/s
83.5
tokens/s
357
ms
pp1280+tg3072
2820
tokens/s
59.8
tokens/s
465
ms
pp384+tg1152
3822
tokens/s
101
tokens/s
109
ms
pp64+tg1024
2836
tokens/s
121
tokens/s
29
ms
pp16+tg1536
1456
tokens/s
112
tokens/s
18
ms