TEST #109 RESULTS
03/28/2025 - 9:48 PM
ACCELERATOR
MODEL
73.3
tokens/s
generation
647
ms
time to first token
1823
tokens/s
prompt
591
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 7K62 48-Core Processor (znver2)
RAM
125.6GB
OS
Linux
Kernel Release
5.15.0-124-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #134-Ubuntu SMP Fri Sep 27 20:20:17 UTC 2024
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
2365
tokens/s
77.6
tokens/s
446
ms
pp4096+tg256
1975
tokens/s
60.8
tokens/s
2.09
sec
pp2048+tg256
2242
tokens/s
70.9
tokens/s
928
ms
pp2048+tg768
2232
tokens/s
69.5
tokens/s
932
ms
pp1024+tg1024
2391
tokens/s
75.1
tokens/s
441
ms
pp1280+tg3072
2178
tokens/s
67.1
tokens/s
603
ms
pp384+tg1152
2040
tokens/s
78.3
tokens/s
201
ms
pp64+tg1024
522
tokens/s
81.1
tokens/s
134
ms
pp16+tg1536
460
tokens/s
79.7
tokens/s
47
ms