TEST #152 RESULTS
03/28/2025 - 10:27 PM
ACCELERATOR
MODEL
76.3
tokens/s
generation
415
ms
time to first token
3192
tokens/s
prompt
838
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon CPU E5-2690 v2 @ 3.00GHz (ivybridge)
RAM
62.8GB
OS
Linux
Kernel Release
5.15.0-131-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #141~20.04.1-Ubuntu SMP Thu Jan 16 18:38:51 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
3827
tokens/s
78.5
tokens/s
280
ms
pp4096+tg256
2912
tokens/s
70.6
tokens/s
1.42
sec
pp2048+tg256
3456
tokens/s
75.5
tokens/s
606
ms
pp2048+tg768
3450
tokens/s
74.8
tokens/s
607
ms
pp1024+tg1024
3823
tokens/s
77.2
tokens/s
281
ms
pp1280+tg3072
3706
tokens/s
73.7
tokens/s
358
ms
pp384+tg1152
4156
tokens/s
78.5
tokens/s
105
ms
pp64+tg1024
2796
tokens/s
79.3
tokens/s
35
ms
pp16+tg1536
602
tokens/s
78.9
tokens/s
39
ms