TEST #356 RESULTS
04/06/2025 - 6:07 AM
ACCELERATOR
MODEL
68.2
tokens/s
generation
452
ms
time to first token
2836
tokens/s
prompt
754
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen Threadripper PRO 3955WX 16-Cores (znver2)
RAM
127.8GB
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
3441
tokens/s
68.3
tokens/s
316
ms
pp4096+tg256
2739
tokens/s
65.2
tokens/s
1.51
sec
pp2048+tg256
3176
tokens/s
67.1
tokens/s
661
ms
pp2048+tg768
3153
tokens/s
67.7
tokens/s
663
ms
pp1024+tg1024
3414
tokens/s
68.8
tokens/s
314
ms
pp1280+tg3072
3339
tokens/s
66.8
tokens/s
398
ms
pp384+tg1152
3471
tokens/s
69.8
tokens/s
125
ms
pp64+tg1024
2181
tokens/s
70.6
tokens/s
42
ms
pp16+tg1536
612
tokens/s
69.9
tokens/s
40
ms