TEST #124 RESULTS
03/28/2025 - 10:06 PM
ACCELERATOR
MODEL
61.7
tokens/s
generation
767
ms
time to first token
1729
tokens/s
prompt
518
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon CPU E5-2630 v4 @ 2.20GHz (broadwell)
RAM
31.2GB
OS
Linux
Kernel Release
5.4.0-205-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #225-Ubuntu SMP Fri Jan 10 22:23:35 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
2026
tokens/s
66.2
tokens/s
521
ms
pp4096+tg256
1717
tokens/s
57.6
tokens/s
2.40
sec
pp2048+tg256
1963
tokens/s
45.0
tokens/s
1.06
sec
pp2048+tg768
1525
tokens/s
62.0
tokens/s
1.36
sec
pp1024+tg1024
2123
tokens/s
64.3
tokens/s
497
ms
pp1280+tg3072
2070
tokens/s
60.6
tokens/s
634
ms
pp384+tg1152
2218
tokens/s
66.0
tokens/s
188
ms
pp64+tg1024
1832
tokens/s
67.1
tokens/s
49
ms
pp16+tg1536
90
tokens/s
66.5
tokens/s
191
ms