mdk7777777
Well, tear predicted this very early on in the BA thread...It is very easy to adjust TFLOPS to hide lower participation rates.
 
We have now just witnessed a second miracle in TFLOPS performance.
 
Total active CPU continues to decline. Now at 157K
 
Yet in a couple of days...an additional 13K GPU were "found"(out of a total of only around 20K a few days ago)
 
So, TFOPS increased from 15K to 25K in the same few day period.
 
So in conclusion...total number of machines continue to decline, but total TFLOPS increased(not by a little) but by a whopping  66% due to "found" processing power.;)
 
Now, I don't doubt for a minute that GPU are very powerful, and that this is a possibility.
 
But since the total active participation of GPU were only 20K a few days ago...where did these extra 13K GPU come from?
 
Just another miracle of STATS I guess.
 
 

I will not say this accounts for all the new GPU's but for "some of them" https://folding.stanford.edu/home/update-to-stats-system-accounting-of-ati-gpus/

yeah, while finding the lost AMD, they also picked up a large haul of nvidia.
 
I was just guessing 13k from memory. Rattledagger estimates it higher on the FF:
 
"...before adding 14964 GPU-clients (not counting Amd) and the majorify of these was added the last 48 hours."
 
To put in perspective, the entire EVGA team is around 1000 active, right?
 
I haven't seen any team(s) adding 13-15k members, much less GPU.
 
perhaps their numbers have just been completely wrong since the last update a year ago...which corrected the numbers being wrong for the two entire previous years....
 
oh well, just numbers right? not important to get them correct, or credible, or accountable, or peer reviewed...Its not like we are doing science or anything important...right?

well, after a few days of the FF defending the STATS update, I took a look.
 
 the active fermi cpu does not match the core count as it does for any other gpu.
unless there is a huge number of gpu out there running multiple cores per cpu...this is where the problem is.
 
It would also explain going from 15 to 27 TFLOPS in a few days.

yeah, it is clearly borked despite what the apologists at the FF might say.
 
I figured they might say the increase in Fermi cards was caused by the same accounting error that caused them to loose the AMD cards.
 
However, the core should equal the cpu count on GPU...so something is obviously very wrong.
 
It stands to reason that you are not going to nearly double the entire production of the project overnight without adding any new resources.
 
But the FF has become infected with Clericalism...a religious zeal to defend against any criticism...regardless of how obvious or justified.
..."We assume the numbers are correct...people who doubt them are just no good infidels". LOL..LOL
 

Not very, but I still preferred the FAH Tracker, because it did that wonderful job of detecting when I was launching a game and paused things FOR me.

That would be a big impact on all of the things FAH is trying to do, recognize that we do other things with our computers and make it DETECT them.

Now if only they built a program detection bit into v7, it would be quite simple for us to use. Especially those of us (like me) who run a single computer.