TEST #87 RESULTS
03/28/2025 - 9:21 PM
ACCELERATOR
MODEL
95.8
tokens/s
generation
90
ms
time to first token
14584
tokens/s
prompt
2498
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon Platinum 8470 (sapphirerapids)
RAM
1007.3GB
OS
Linux
Kernel Release
6.8.0-48-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #48~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Mon Oct 7 11:24:13 UTC 2
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
19746
tokens/s
99.8
tokens/s
62
ms
pp4096+tg256
15680
tokens/s
88.8
tokens/s
277
ms
pp2048+tg256
18969
tokens/s
94.2
tokens/s
119
ms
pp2048+tg768
18967
tokens/s
93.5
tokens/s
119
ms
pp1024+tg1024
19556
tokens/s
94.2
tokens/s
63
ms
pp1280+tg3072
17413
tokens/s
92.2
tokens/s
84
ms
pp384+tg1152
16243
tokens/s
95.7
tokens/s
33
ms
pp64+tg1024
3634
tokens/s
102
tokens/s
26
ms
pp16+tg1536
1043
tokens/s
101
tokens/s
24
ms