Upgraded Maxwell support for Core17 Note the (“Advanced Methods” setting)
With the newer NVIDIA drivers, it looks like Core17 works well on Maxwell.  We’ve released Core17 to Maxwell on adv (“Advanced Methods” setting).  If you’re having problems, you can set from adv back to the regular fah setting, allowing donors to opt out if they’re having problems.  The latest drivers are recommended.
 
I will wait and get one GTX 990 with the Hyper Copper when it comes out for my first Maxwell.
 

I'm basing this on 24 hours of core17 9201's. GTX970: 240k x 365 = 87,600,000 points/yr. That's one  $350 card.
Core15 and 18 = about 55-70k ppd. 70k x 365 = 25,550,000/yr.
My GTX780 gets 185k x 365 = 67,525,000/yr on the same core17 wu's.
 
bcav, your OP was correct or on the right track. It just took Fah  2 weeks to disperse the wu's properly.
 
To all folders that purchased Maxwell cards, good luck and hope Fah keeps em coming.
 
I'll be testing 780 vs 970 with a kill-o-watt meter in the next few days to add fuel to the stats.
 
btw I'm using the (latest) 344.16 drivers.
 .

        "I'm sure they are working on the issues...I'm just not convinced we won't see shortages of optimized WU again in the near future."

 
It's to be expected. There's always bumps in the road with Fah.
The 780 has only picked up 1 9201 in the same time span (36 hours) as the 970. It's been getting 13000 and the new 104xx core 17's.

My 980 seems to be picking up lots of Core17 now, it makes me happy.

I'm still planning on replacing my PSU, cleaning out the case and adding an extra 980 along with some SSDs in the next year though.

Mmm, pretty computer.
 
Maybe I'll be able to clean the rest of the room some day lol

My 970 has been going since I picked it up from the border yesterday, and I am loving this.  Tonnes of points, less power usage, and it cool and quiet!  Currently getting 232k ppd

Ugh, brutal.

widsss
So a non-maxwell card will fold core 18 better than maxwell?  Or is the gimping going to cover all cards?

I would say more that Core 17 have been fixed to support the new Maxwell Card and FAH is working on Core 18 now to work with the Maxwell card.
 
But Yes.

Just to clarify:
 
Core 18 is an upgrade of core 17 using the latest version of OPEN MM
openMM is the opencl code that they use to build the forces and interactions of the atoms (the core math that communicates with the cards GPU math function capability.
 
The newest open MM version is 6.1 or something.
 
When these use these advanced(and optimized to run quickly on GPU) math functions, they see a speed up of their computations.
However, something in MAxwell is giving errors (only on the new core18)
 
So, rather than figure out where the problem lies (the maxwell hardware..how it handles and accelerates memory...or the drivers that NV has issued, or some problem with the openmm math software) they simply replace the advanced math functions(only on Maxwell cards) with simpler, older software.
 
This allows the cards to function=finish without error, but at a huge slow down=1/4 speed.
 
Since other cards are able to do the math functions without error...they should run as well as or better than the core 17 that they replace.
 
I wrote the following to kyleb regarding my opinion of this strategy:
 
 "Pushing this type of WU is very poor for donor morale. It shows a lack of respect for the donors investment, time and electricity.

This is the worst of all options. In order of preference the options are:

1. continue to supply only the limited working WU to Maxwell cards
2. suspend Maxwell support until the cores/driver issues are resolved
3. send out intentionally crippled WU that run at 1/4 speed and don't inform donors of any plan to resolve the issues."
 
As per usual when PM PG, I had no response.
 

To Date My two R9 290X have only received Core 17, I have yet to see one that is. (advanced)

Well received my first core 15 today on my new 970.  These suck.  seriously.  2 minute 54 second fold time, with only 14093 points, and no bonus!  My ppd has dropped from 230k to 69999.

That's good to hear.  I am worried though because halfway through this core 15 my card started to screech like a banshee, due to coil whine..

2 minute 54 seconds per fold.
And yes I am using client-type = advanced.  
 
it was my first core 15 and my card had some serious coil whine while doing the core 15.  Now that I am back on a core 17, I have no coil whine.  My system is also using about 30 watts less power.
During core 15 I was at 252 watts of total usage, during core 17 I am at 221 watts usage.  My time per fold on core 17 is 2 minutes 3-6 seconds, the seconds fluctuate a bit.
 
And in case you are wondering, yes I have a killawatt type meter hooked up to my computer at all times. http://www.newegg.com/Product/Product.aspx?Item=N82E16812119560 is what I use to monitor my power usage.

Since I have one rig that has 3 780 Ti SC's, would it better to keep this on FAH and then use the four new GTX 980 SC which will be in two different rig for crunching?

mdk7777777
Yup, that sounds like a plan.
Efficiency for folding on  maxwell is now going to 1/4 of the potential.
 

Re: Core18 V3 Feedback

by kyleb » Mon Oct 27, 2014 11:42 amOK, this core version has been moved to adv

What if we take our flag off of advmethods?

Well I heard some more coil whine coming from my PC, and it was another damn core 15.  Except this time my computer is using 290! watts.  What the hell.  These cores are murder on my card and my power bill.  I seriously can not take this core 15 crap.  I can't even be in the same room with my computer when these are going, and that's impossible as it's in my bedroom, and I kinda need to sleep there.