TEST #129 RESULTS
03/28/2025 - 10:11 PM
ACCELERATOR
MODEL
111
tokens/s
generation
310
ms
time to first token
4120
tokens/s
prompt
1138
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon CPU E5-2699 v3 @ 2.30GHz (haswell)
RAM
125.8GB
OS
Linux
Kernel Release
5.15.0-134-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #145-Ubuntu SMP Wed Feb 12 20:08:39 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
4979
tokens/s
114
tokens/s
214
ms
pp4096+tg256
3992
tokens/s
100
tokens/s
1.04
sec
pp2048+tg256
4610
tokens/s
110
tokens/s
453
ms
pp2048+tg768
4608
tokens/s
109
tokens/s
454
ms
pp1024+tg1024
4949
tokens/s
112
tokens/s
216
ms
pp1280+tg3072
4790
tokens/s
106
tokens/s
276
ms
pp384+tg1152
5195
tokens/s
114
tokens/s
82
ms
pp64+tg1024
3215
tokens/s
117
tokens/s
28
ms
pp16+tg1536
739
tokens/s
116
tokens/s
30
ms