A piece of history for future crunchers

For future crunchers, now that old ACEMD Long and Short Work Units are extinct.
When original ACEMD app existed, reported results looked like this:



For comparison, current New version of ACEMD (ACEMD3) results look like this:



Life is continuous evolution.

Hilarious.

113695 is actually one of my hosts.
You can still see the output of some failed CUDA65 and CUDA60 tasks too. This host had two GTX 670s (CUDA60), later a single GTX 980, and after that a single GTX980 Ti. I still have one of those GTX 670s as it got broken (stripes on the screen).
514770 is the same host, but with Linux.

Before this information finally disappears, I found another interesting result from Viktor Svantner's 506552 host:



As it can be seen, it is a 4 GPUs host.
Old results format was useful for taking many interesting conclusions.
- Being a 4 GPUs system, it has to be very well vented, as temperatures are not getting too high.
- It is a multi GPU system, but at beginning of result it could be seen exactly at what GPU model the Work Unit was processed: GeForce GTX 1080 Ti.
- It could be seen both device and memory clocks at the moment of starting WU.
- Also at the end was reported an useful parameter Time per step that permitted quickly estimate the performance (lower is better).
I mean: comparing 1.526 ms of this card with 1.226 ms of yours, it can be appreciated your performance increase due to both slightly higher device overclocking (1721MHz vs 1683MHz) and SWAN_SYNC on.

New result format is missing this information, but I think it's worth it, since now both Linux and Windows systems and modern GPUs can contribute to Project, and from now it is not dependent on an eventually expiring license...

..it has to be very well vented, as temperatures are not getting too high
...

well, anything beyond 70°C should probably be considered "high"

Being a 4 GPUs system, it has to be very well vented, as temperatures are not getting too high.

This host either use PCIe risers to separate the cards to give adequate ventilation for air cooling, or it has liquid cooling.
Judging by that the card temperatures don't change much while crunching, and the individual cards have increasing temperatures from one to the next (GPU 1->0->2->3), the (linked) liquid cooling is more likely.

from now it is not dependent on an eventually expiring license...

You wish, but actually there's no proof or indication of that.

Being a 4 GPUs system, it has to be very well vented, as temperatures are not getting too high.

This host either use PCIe risers to separate the cards to give adequate ventilation for air cooling, or it has liquid cooling.
Judging by that the card temperatures don't change much while crunching, and the individual cards have increasing temperatures from one to the next (GPU 1->0->2->3), the (linked) liquid cooling is more likely.

from now it is not dependent on an eventually expiring license...

You wish, but actually there's no proof or indication of that.

I'm going with risers. Liquid cooling should be lower. I run hybrids and they usually stay in the mid 50s to maybe low 60s during the summer. In winter, when the temp outside is near freezing, that temp will drop to high 40s to mid 50s.

Of course you could just ask him in a PM how he has his system set up.

I'd rather guess the Power Limit has been cut. That's a low clock speed.