TEST #938 RESULTS
05/24/2025 - 1:41 AM
ACCELERATOR
MODEL
138
tokens/s
generation
146
ms
time to first token
8479
tokens/s
prompt
2000
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen Threadripper 3970X 32-Core Processor (znver2)
RAM
125.6GB
OS
Linux
Kernel Release
6.8.0-60-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #63-Ubuntu SMP PREEMPT_DYNAMIC Tue Apr 15 19:04:15 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
8863
tokens/s
145
tokens/s
122
ms
pp4096+tg256
8578
tokens/s
121
tokens/s
486
ms
pp2048+tg256
10653
tokens/s
139
tokens/s
200
ms
pp2048+tg768
10636
tokens/s
134
tokens/s
200
ms
pp1024+tg1024
11186
tokens/s
141
tokens/s
98
ms
pp1280+tg3072
10619
tokens/s
130
tokens/s
128
ms
pp384+tg1152
9835
tokens/s
143
tokens/s
46
ms
pp64+tg1024
4456
tokens/s
146
tokens/s
21
ms
pp16+tg1536
1480
tokens/s
145
tokens/s
17
ms