TEST #416 RESULTS
04/06/2025 - 7:57 PM
ACCELERATOR
MODEL
36.5
tokens/s
generation
629
ms
time to first token
2085
tokens/s
prompt
495
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 9554 64-Core Processor (znver4)
RAM
43GB
OS
Linux
Kernel Release
5.15.0-136-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #147-Ubuntu SMP Sat Mar 15 15:53:30 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
2656
tokens/s
40.2
tokens/s
413
ms
pp4096+tg256
1945
tokens/s
29.3
tokens/s
2.15
sec
pp2048+tg256
2320
tokens/s
35.0
tokens/s
914
ms
pp2048+tg768
2316
tokens/s
34.0
tokens/s
916
ms
pp1024+tg1024
2614
tokens/s
37.1
tokens/s
421
ms
pp1280+tg3072
2495
tokens/s
32.6
tokens/s
542
ms
pp384+tg1152
2803
tokens/s
39.6
tokens/s
166
ms
pp64+tg1024
1182
tokens/s
40.8
tokens/s
78
ms
pp16+tg1536
434
tokens/s
39.9
tokens/s
61
ms