TEST #67 RESULTS
03/28/2025 - 8:59 PM
ACCELERATOR
MODEL
37.1
tokens/s
generation
519
ms
time to first token
2601
tokens/s
prompt
571
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
1511.6GB
OS
Linux
Kernel Release
6.8.0-40-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #40~22.04.3-Ubuntu SMP PREEMPT_DYNAMIC Tue Jul 30 17:30:19 UTC 2
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
3645
tokens/s
37.1
tokens/s
308
ms
pp4096+tg256
2320
tokens/s
32.6
tokens/s
1.80
sec
pp2048+tg256
2682
tokens/s
38.1
tokens/s
790
ms
pp2048+tg768
2690
tokens/s
36.7
tokens/s
788
ms
pp1024+tg1024
3570
tokens/s
37.0
tokens/s
315
ms
pp1280+tg3072
3461
tokens/s
35.2
tokens/s
398
ms
pp384+tg1152
3366
tokens/s
38.9
tokens/s
137
ms
pp64+tg1024
1217
tokens/s
39.9
tokens/s
81
ms
pp16+tg1536
455
tokens/s
38.5
tokens/s
60
ms