TEST #68 RESULTS
03/28/2025 - 8:59 PM
ACCELERATOR
MODEL
37.9
tokens/s
generation
1.02
sec
time to first token
1328
tokens/s
prompt
367
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 7443 24-Core Processor (znver3)
RAM
251.5GB
OS
Linux
Kernel Release
6.8.0-49-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #49~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Wed Nov 6 17:42:15 UTC 2
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1601
tokens/s
39.3
tokens/s
665
ms
pp4096+tg256
1231
tokens/s
34.3
tokens/s
3.36
sec
pp2048+tg256
1449
tokens/s
37.4
tokens/s
1.44
sec
pp2048+tg768
1288
tokens/s
36.5
tokens/s
1.62
sec
pp1024+tg1024
1572
tokens/s
38.5
tokens/s
677
ms
pp1280+tg3072
1233
tokens/s
36.3
tokens/s
1.06
sec
pp384+tg1152
1729
tokens/s
39.5
tokens/s
247
ms
pp64+tg1024
1462
tokens/s
39.9
tokens/s
68
ms
pp16+tg1536
386
tokens/s
39.6
tokens/s
66
ms