TEST #106 RESULTS
03/28/2025 - 9:42 PM
ACCELERATOR
MODEL
33.5
tokens/s
generation
2.72
sec
time to first token
427
tokens/s
prompt
174
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon Gold 6138 CPU @ 2.00GHz (skylake-avx512)
RAM
376.5GB
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
520
tokens/s
35.9
tokens/s
2.00
sec
pp4096+tg256
488
tokens/s
26.7
tokens/s
8.44
sec
pp2048+tg256
507
tokens/s
31.9
tokens/s
4.07
sec
pp2048+tg768
507
tokens/s
31.2
tokens/s
4.07
sec
pp1024+tg1024
520
tokens/s
34.1
tokens/s
2.00
sec
pp1280+tg3072
502
tokens/s
30.2
tokens/s
2.58
sec
pp384+tg1152
504
tokens/s
36.3
tokens/s
787
ms
pp64+tg1024
221
tokens/s
37.9
tokens/s
315
ms
pp16+tg1536
75
tokens/s
37.0
tokens/s
238
ms