Page 2 of 4
Re: 2012 slump?
Posted: Tue Oct 16, 2012 8:43 pm
by Nathan_P
As 7im says its mostly down to people re sizing their farms coupled with the economic situation and the really hot summer. If you look at team stats in the last week or so people are coming back as the weather cools off. Also lots of people have shut down lots of machines and gone for smaller setups using less power including me. I used to fold on 5 GPU and 2 SMP, now i fold on 1 bigadv and 2 SMP for less powe rand this will go down to 1 Bigadv and 1 SMP in the next month or so. I still have all my GPU's and would run them if i could afford to do so. You only have to look on the forum's of most of the teams to see that lots of people are in a similar situation
Re: 2012 slump?
Posted: Tue Oct 16, 2012 9:04 pm
by 7im
And even though V7 seems to report the number of active clients correctly (my does too), can we be sure the FLOPS counts are being reported correctly?
So even if the active client count is accurate and went down because people are running more concentrated clients (2 or 3 i7s instead of 10 dual cores) that would account for some of the client count dip without that being taken so pessimistically.
And if the FLOPS reporting is wrong, it may not accurately reflect that increase in concentrated power as it should. Can the FLOPS counter even go above a single core? Can it count to 4P? Or know how to read a kepler card? Or a new 7xxx series card? There may be whole chunks of unreported FLOPS...?
Re: 2012 slump?
Posted: Tue Oct 16, 2012 9:50 pm
by mdk777
And if the FLOPS reporting is wrong, it may not accurately reflect that increase in concentrated power as it should. Can the FLOPS counter even go above a single core? Can it count to 4P? Or know how to read a kepler card? Or a new 7xxx series card? There may be whole chunks of unreported FLOPS...?
Well, I asked a similar question around a year ago when there was a huge jump in the number of ATI GPU.
However, if we have so little confidence in the reported calculations, why are they even posted?
Assuming that they can't calculate and report the FLOPS of the project accurately, certainly does nothing for credibility and recruitment.
And hence the self-inflicted slump in participation could occur.
Real or imagined....bad press.
Any answers more than simple environmental, social, or economic issues are better suited for the "What's holding FAH Back" thread, not this one.
Referenced in my first post.
Assuming that a two year downward trend is a statistical aberration,environmental, or economical; presumes the answer to the question rather than investigating it.
I think the downward slump was predicted by many due to the management and structural decisions of the project. You might not agree, but limiting the investigation to preclude obvious factors does nothing to find true causation.
Re: 2012 slump?
Posted: Tue Oct 16, 2012 11:41 pm
by k1wi
I agree with mdk77's previous statement that total FLOPS and total active clients tend to track very closely (even accounting for the inherent 50/7 day lag associated with 'active' clients), though I would expect Moore's law to be more long term.
Re: 2012 slump?
Posted: Wed Oct 17, 2012 12:13 am
by 7im
mdk777 wrote:
However, if we have so little confidence in the reported calculations, why are they even posted?
While flagrant pessimism may be your personal writing style, taking every statement I make to it's logical extreme is a fallacy and is not helpful to the discussion. You're just being argumentative.
I did not say that I had little confidence in the FLOPS numbers. I asked if it was possible that the FLOPS numbers haven't accounted for recent changes. I assumed it was a not large difference, but you assumed it was a large difference. "We" have have a difference in opinion about the confidence level.
Re: 2012 slump?
Posted: Wed Oct 17, 2012 12:44 am
by mdk777
While flagrant pessimism may be your personal writing style, taking every statement I make to it's logical extreme is a fallacy and is not helpful to the discussion. You're just being argumentative.
I just wrote that I think it is possible to have 6 million rather than 6000 participants.
I don't know of any realist growth projection that could be more optimistic.
However, it won't happen looking for rounding errors.
Re: 2012 slump?
Posted: Wed Oct 17, 2012 7:26 pm
by Ivoshiee
Active client count should go down with the V7, it is designed to replace numerous installations on the same system. Different thing is with FLOPS. If it does go up or remain the same then we are still okay.
Re: 2012 slump?
Posted: Wed Oct 17, 2012 10:35 pm
by PinHead
Many people have moved from PC's to PAD's, Phone's or NetBook's. So the number of PC's / Laptop's per household is also reducing and upgrades are not occuring in those homes. Wife's laptop mostly sit's idle these days, she does it all from her phone. The change has been brewing for several years now.
Re: 2012 slump?
Posted: Thu Oct 18, 2012 3:54 am
by bruce
Ivoshiee wrote:Active client count should go down with the V7, it is designed to replace numerous installations on the same system. Different thing is with FLOPS. If it does go up or remain the same then we are still okay.
Going back to the question I asked earlier: Is the number being reported is the number of cores running or something else. If I was running a v6 SMP client and GPU client, that counted as two. When I replaced the V6 software with V7 (with a SMP and GPU slot) does it count as one or two? I think it still counts as two.
Re: 2012 slump?
Posted: Thu Oct 18, 2012 8:25 pm
by Ivoshiee
bruce wrote:Ivoshiee wrote:Active client count should go down with the V7, it is designed to replace numerous installations on the same system. Different thing is with FLOPS. If it does go up or remain the same then we are still okay.
Going back to the question I asked earlier: Is the number being reported is the number of cores running or something else. If I was running a v6 SMP client and GPU client, that counted as two. When I replaced the V6 software with V7 (with a SMP and GPU slot) does it count as one or two? I think it still counts as two.
The number has been as number of the FAH client installations. I do not know if the V7 should change it, it should not.
Re: 2012 slump?
Posted: Thu Oct 18, 2012 9:46 pm
by k1wi
Ivoshiee wrote:bruce wrote:Ivoshiee wrote:Active client count should go down with the V7, it is designed to replace numerous installations on the same system. Different thing is with FLOPS. If it does go up or remain the same then we are still okay.
Going back to the question I asked earlier: Is the number being reported is the number of cores running or something else. If I was running a v6 SMP client and GPU client, that counted as two. When I replaced the V6 software with V7 (with a SMP and GPU slot) does it count as one or two? I think it still counts as two.
The number has been as number of the FAH client installations. I do not know if the V7 should change it, it should not.
Is there anyone who has a stable installation of V7 that can compare their count of active clients to their # of slots? Surely that would be the easiest way to go about it?
I would be willing to give it a go with a fresh & temporary username/team combination if it ends the speculation.
Re: 2012 slump?
Posted: Thu Oct 18, 2012 9:54 pm
by Zagen30
I know for a fact that the number of active clients on the user page is equal to the number of active v7 slots, as for a long time I had 4 slots running across two v7 installations and that showed up as 4 active clients. What I don't know is if the OS stats are just reporting the sum of everyone's active slots (using the definition of "active" specific to each client type, of course) or if it's calculating active clients differently.
Re: 2012 slump?
Posted: Thu Oct 18, 2012 9:57 pm
by bollix47
Yes, I have 11 slots running on 8 computers and the Official stats show 11 Active Clients.
Re: 2012 slump?
Posted: Thu Oct 18, 2012 10:28 pm
by k1wi
Quick replies! So it would certainly appear that individually 1 v7 slot = 1 active 'client'.
Which, as you say leaves your uncertainty Zagen30: Does PG gather total stats by summing all the individual active counts or something along those lines. It also leaves the underlying question "Does the total stats page even include v7 at all?"
According to today's Client statistics by OS page the total number of active 'CPUs' is stated at 259,390. Summing the totals of each client type equals 254,823, which is 4,567 or about 1.8% short. I would imagine there are more than 4,567 v7 clients/slots in the wild? Perhaps the two totals are just measured at different points in time or maybe there is a small number of 'odd' client types still working away?
I don't think there is anyway anyone other than PG could answer these questions, but I would propose that if the answer is "v7 is not reflected in the overall stats and stats by OS" then they might want to get on that pretty quick? The bottom falling out of both total FLOPS & active participation just because the next generation client isn't being measured cannot be good publicity...
---
FWIW the difference between the sum of CPU type totals and total CPUs amongst total CPUs is 14%, I assume reflecting depreciated clients?
Re: 2012 slump?
Posted: Thu Oct 18, 2012 10:42 pm
by 7im
k1wi wrote:
...cannot be good publicity...
Ya, so let's keep posting about it over and over and keep this topic on top of the list for everyone to see.
Let's call this a dead horse.
Let's wait for PG to address said horse.