TEST #104 RESULTS
03/28/2025 - 9:41 PM
ACCELERATOR
MODEL
54.4
tokens/s
generation
301
ms
time to first token
4461
tokens/s
prompt
931
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 5 7600 6-Core Processor (znver4)
RAM
30.6GB
OS
Linux
Kernel Release
6.8.0-45-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #45~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Wed Sep 11 15:25:05 UTC 2
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
5818
tokens/s
59.5
tokens/s
193
ms
pp4096+tg256
3999
tokens/s
45.1
tokens/s
1.05
sec
pp2048+tg256
4970
tokens/s
52.6
tokens/s
431
ms
pp2048+tg768
4971
tokens/s
51.2
tokens/s
431
ms
pp1024+tg1024
5855
tokens/s
55.2
tokens/s
192
ms
pp1280+tg3072
5549
tokens/s
49.5
tokens/s
248
ms
pp384+tg1152
6089
tokens/s
58.2
tokens/s
80
ms
pp64+tg1024
2216
tokens/s
60.0
tokens/s
44
ms
pp16+tg1536
682
tokens/s
58.7
tokens/s
40
ms