TEST #146 RESULTS
03/28/2025 - 10:22 PM
ACCELERATOR
MODEL
122
tokens/s
generation
190
ms
time to first token
6657
tokens/s
prompt
1625
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 7B13 64-Core Processor (znver3)
RAM
2003.9GB
OS
Linux
Kernel Release
5.4.0-174-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #193-Ubuntu SMP Thu Mar 7 14:29:28 UTC 2024
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
7255
tokens/s
125
tokens/s
149
ms
pp4096+tg256
6782
tokens/s
112
tokens/s
613
ms
pp2048+tg256
7998
tokens/s
120
tokens/s
265
ms
pp2048+tg768
7495
tokens/s
119
tokens/s
281
ms
pp1024+tg1024
7942
tokens/s
122
tokens/s
137
ms
pp1280+tg3072
8513
tokens/s
118
tokens/s
158
ms
pp384+tg1152
10026
tokens/s
127
tokens/s
46
ms
pp64+tg1024
3265
tokens/s
131
tokens/s
27
ms
pp16+tg1536
636
tokens/s
128
tokens/s
33
ms