TEST #105 RESULTS
03/28/2025 - 9:42 PM
ACCELERATOR
MODEL
57.8
tokens/s
generation
880
ms
time to first token
1460
tokens/s
prompt
458
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon CPU E5-2690 0 @ 2.90GHz (sandybridge)
RAM
503.8GB
OS
Linux
Kernel Release
6.5.0-28-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #29~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Thu Apr 4 14:39:20 UTC 2
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1700
tokens/s
61.4
tokens/s
619
ms
pp4096+tg256
1411
tokens/s
48.6
tokens/s
2.92
sec
pp2048+tg256
1593
tokens/s
56.1
tokens/s
1.30
sec
pp2048+tg768
1593
tokens/s
55.0
tokens/s
1.30
sec
pp1024+tg1024
1689
tokens/s
58.8
tokens/s
622
ms
pp1280+tg3072
1644
tokens/s
53.1
tokens/s
795
ms
pp384+tg1152
1686
tokens/s
61.3
tokens/s
243
ms
pp64+tg1024
1370
tokens/s
63.4
tokens/s
61
ms
pp16+tg1536
453
tokens/s
62.3
tokens/s
50
ms