Folding Forum

ajm wrote:Thank you Joe_H! But the TPF is a rather fleeting apparition. I have to be in front of the computer during the folding to see it, and it changes constantly. Whereas the points appear in the log and on https://apps.foldingathome.org/cpu
As the machine run 24/7, I would prefer using a value that I can more easily ascertain after the fact. Is the TPF that crucial for such a test?

I would say TPF is a good measure. TPF is just the time it takes to do 1%, and the logs will show that.

Yes, but as Joe_H mentions above, you need to take an average on ten of those 1%, because they vary. I mean, is the work involved really worth it? Doesn't the credit/bonus take that into account already?

TPF will tell us how much faster/slower the folding was. Of course, if you can use the end credit and mathematically work backwards to find the TPF, that would work too: https://foldingathome.org/faqs/points/b ... etermined/

I just don't see why we need that TPF value here, as it is a comparison 1:1 between systems?
If I have a better ratio Credit / Base Credit (that Excel can calculate directly on my sheets) on Linux than on Windows, that will tell us what we need to know just as well, no?

ajm wrote:I just don't see why we need that TPF value here, as it is a comparison 1:1 between systems?
If I have a better ratio Credit / Base Credit (that Excel can calculate directly on my sheets) on Linux than on Windows, that will tell us what we need to know just as well, no?

My opinion is still that TPF is a better straight performance comparison because it's linear, whereas the bonus points are exponential.

Well, measuring only the TPF will give us a better idea of the raw computing performance, as opposed to adding the network performance in the mix when considering the Credit.
OK, I'll stop monitoring the Bonus. And maybe enter the TPF in minutes (eg 48 s = 0.8; 1 m 22s = 1.367) - that will be handier later.

If you look at the WU Stats it has a days indicator which if divided by 100 gives the TPF ... I do this in a horribly mandraulic way to check TPFs for beta team reports (at the moment I have too much time on my hands) ... I guess this could easily be "automated" by someone cleverer than me.

Thank you, but I'm not sure I understand? You mean here, for example: https://apps.foldingathome.org/wu#proje ... 776&gen=41

It gives
User Team CPUID Credit Assigned Returned Credited Days Code
ajm 182116 7B208F5E4478FE20 12 856,2 2020-04-15 10:32:59 2020-04-15 14:02:12 2020-04-15 13:54:43 0,14 Ok

So, 0,14 days, that is 0,14/100 x 1440 (minutes) = TPF in minutes?

I do it (Daysx86,400‬)/100 to get it in seconds but same thing yes .. I then break it into minutes and seconds for my reports adding in other stuff ... "14379 (111, 0, 12) TPF 0m37s 32core avx_256" from https://apps.foldingathome.org/wu#proje ... e=0&gen=12

There are differences with my count (well, just looking at the Estimated TPF in the Advanced Control and hoping it will be accurate enough), but your system is definitely better, as it uses FAH's stats.
Thank you Neil-B!

Several PRCG that I delivered don't appear in the stats. I guess those are "advanced" WUs that got deleted?

May not be ... I know of at least one server which has yet to posts stats from nearly a week ago (I have 80+ of them awaiting credits) - there may be others ... If you post the PRCGs that are missing someone might be able to check.

Please note I am not complaining about the delay in any way (hey it could be a nice big bonus when they turn up), rather to illustrate there are some stats style delays specific to certain servers (or possibly even projects).

Thank you, but a couple WUs more or less are not so important here, as I will run those slots for a week on each system to get a good baseline. For now, I deleted the missing WUs on my sheet, as well as the advanced code.

Neil-B wrote:If you look at the WU Stats it has a days indicator which if divided by 100 gives the TPF ... I do this in a horribly mandraulic way to check TPFs for beta team reports (at the moment I have too much time on my hands) ... I guess this could easily be "automated" by someone cleverer than me.

While that is one way, please note that it would also include the time taken to download/upload (including multiple attempts) as it is the total time, not just time to fold a single frame. Give the current bandwidth limitations, the calculated TPFs might be longer than the actual TPF once the F@H Server limitations are resolved.

True .. I forget I have been lucky with getting a near continuous stream of WUs and very few upload issues .. my bad.