it should DL the new core with the next WU
just don't pause existing ones for any reason - because it will fetch the new core
seems the new core error collection doesn't like the way the old DATA is
 

Update so all will know:
 

"bollix47"FYI: I've sent a PM to proteneer concerning the lack of auto update for core 17 from version 44 to 45.
Until there's a 'fix' for this I advise beta GPU slots running core 17 be set to Finish and the core manually deleted between work units.

Zagen30

New zeta core project 8700 released to beta

As of posting this I don't see 8700 on psummaryB or C, but from the post it's worth 6000 base points with deadlines of 6.5 and 8.8 days, vs. 7663's figures of 1600/4.3/5.5.  Don't know if this is a substantially larger protein, longer timescale/WU, or something else.

 
Largest GPU WU by far, 53k atoms.  Is much harder on GPU compared to 7663, some have had to downclock factory OC cards for stability.

My GTX 660 Ti OC has a TPF of about 5min 22sec (that the average and will vary) and takes just over 9 hours to finish.
 
GTX 650 OC has a TPF of 14min 21sec
GTX 675M Stock has a TPF of 15min 04sec
 
Have you tried restarting the Client and is the GPU Usage 90%+?
 
EDIT -> Are you using FahCore_17 version .44 or .45? I am using .45

I just got a P8900 on my GTX Titan:

 
Decent PPD increase over 7663 as well. It runs a bit hotter, but it's still a good bit less stressful than games. This is at a core clock of 1006MHz (my Titan doesn't OC worth a damn)

also if you have multi gpus on the same system you must finish or restart all gpus for it to update and start the new WU. I lost a 7663 at 63% when it updated after pausing to see what was wrong.

here is a blend of 7663 and 8900. 680s @ 1228 (x79) 7970's @ 1060 (p67)

also the 8900 wu is now on summaryC so you can update your HFM.

yodap

snuddi

Did you have to change anything to get those 8900 WU (other then the "client-type beta" flag)?

I didn't do anything snuddi. It finished a 7663 and started this wu.
Over 20% point increase over 7663.

Great thanks. I will just wait and see, didn't get one now but maybe later.

texinga

"Holy Mother of GPU"...!  Thanks for that screen-shot Jedi.    I think it may be "comin' around again" and great to see some PPD-luv again for some of the graphics cards. 

This will look like nothing if we end up getting equal points for equal work:
 

jedi95
Now, we also know that OpenMM 5.1 has some major performance improvements over the current version. According to the details posted here the performance of the GTX 680 in the PME hbonds test is 2.12 times higher. Taking this performance improvement into account we now get:

(2.12 * 28.5363) / (2.4 * 5.60527) = 4.50 times faster
39 seconds / 4.50 = 8.67 seconds TPF

This now gives a PPD between 428,883 and 511,762.

 
More details:
http://forums.evga.com/tm.aspx?m=1909690
 
I suspect that the reason for keeping the points low at the moment is to avoid a huge rush of people folding these beta WUs. Perhaps we will see more PPD when the WUs move out of beta.

980X SMP 10 VM V6 <-Linux64_FAH
2xGTX480 V7 <-W7 64bit
 
8900 (30, 1, 3)
EC: 19815.26
PPD: 32892.19
TPF: 8 mins 40 secs
 
7663 (23, 26, 84)
EC: 7249.39
PPD: 36205.05
TPF: 2 mins 53 secs
 
P8900 is about 2.5k lower than P7663 on PPD
These are all current projects in the works.
Each GPU assigned its own CPU core.
 
 
Project ID: 8556
 Core: GRO-A3
 Credit: 1673
 
 Name: 980X -SMP 10 in VM
 Number of Frames Observed: 175
 Min. Time / Frame : 00:06:30 - 35,035.2 PPD <-SMP 12
 Avg. Time / Frame : 00:07:23 - 28,939.9 PPD <-SMP 10

I have the shader set to 1480 from stock 1401 clocks.
I would think the 590 would be faster than a 480, you have each GPU locked to a single CPU core of its own?
Are you SMP folding also?

cokeman54

TheWolf

I have the shader set to 1480 from stock 1401 clocks.
I would think the 590 would be faster than a 480, you have each GPU locked to a single CPU core of its own?
Are you SMP folding also?

Not folding on cpu at all. Tried oc and tpf went up so back to stock clocks. I expected better from the 590 also so I suspect it is the Windows environment slowing it down. Like the 76xx series pc lags a lot but temps are much cooler than 76xx series. Looks like this core is a work in progress.

Some are seeing fail for the 8900 project with OC GPU's.
So far I haven't but this is only my 1st WU of this project
plus my GPU memory is not OCed "not sure at this point if that is the reason" it hasn't failed.

 

New jobs for P8900 are in the making, hopefully, once it is done, there will be a total of 9500 WUs available.

178k ppd?

Just finished an 8900 on one of the 680's, had to slightly downclock them for stability, but they are still giving me about 5K PPD more than the 7663's were at higher clocks and 7K more at current clocks. On my 560 the 8900 is running at about 1K less than the 7663.