TEST #81 RESULTS
03/28/2025 - 9:15 PM
ACCELERATOR
MODEL
37.1
tokens/s
generation
289
ms
time to first token
4299
tokens/s
prompt
820
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 9374F 32-Core Processor (znver4)
RAM
1448.6GB
OS
Linux
Kernel Release
6.8.0-51-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #52~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Mon Dec 9 15:00:52 UTC 2
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
5535
tokens/s
38.8
tokens/s
211
ms
pp4096+tg256
4956
tokens/s
33.1
tokens/s
860
ms
pp2048+tg256
5369
tokens/s
36.0
tokens/s
409
ms
pp2048+tg768
5374
tokens/s
35.6
tokens/s
409
ms
pp1024+tg1024
5570
tokens/s
36.6
tokens/s
211
ms
pp1280+tg3072
5063
tokens/s
34.8
tokens/s
280
ms
pp384+tg1152
4917
tokens/s
38.0
tokens/s
100
ms
pp64+tg1024
1474
tokens/s
40.8
tokens/s
65
ms
pp16+tg1536
432
tokens/s
39.9
tokens/s
59
ms