• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Core_17

HumanHunter said:
Ok so I got the dreaded "missing libglib-2.0.0.dll" error message also. So thinking a had a bad HDD (read old,) I rebuilt the system. And now after completing just 3 WU's on the GPU I have it again! after reading this forum I know that it isn't my system. So here is the big question: What is the fix? Even if it's just a temporary work around. Currently my F@H system is just running on SMP. I new to folding so I need someone to walk me through a solution.

Here are my system specs: Win 7 Pro 64bit, Intel 980X, Asus Rampage II Extreme, GTX470 with 320.18 driver, 6Gig of 1866 RAM, FAH GPU Tracker V2 ver 3.54.

Thanks
HumanHunter
Click to expand...

Are you folding for team 33? ;) That might cure some ills. :D
 
HumanHunter said:
Ok so I got the dreaded "missing libglib-2.0.0.dll" error message also. So thinking a had a bad HDD (read old,) I rebuilt the system. And now after completing just 3 WU's on the GPU I have it again! after reading this forum I know that it isn't my system. So here is the big question: What is the fix? Even if it's just a temporary work around. Currently my F@H system is just running on SMP. I new to folding so I need someone to walk me through a solution.

Here are my system specs: Win 7 Pro 64bit, Intel 980X, Asus Rampage II Extreme, GTX470 with 320.18 driver, 6Gig of 1866 RAM, FAH GPU Tracker V2 ver 3.54.

Thanks
HumanHunter
Click to expand...

Did you get this issue sorted? It might be related to privileges and gaining access to that file. You should find it in the same folder as the F@H client. ie: C:\Program files (x86)\FAHClient. There is a chance that you might need to run the fahclient.exe as admin or that for some reason your install didn't put the files into the folder. The other thing is maybe your shortcut points to a different version that's in another folder and that folder doesn't have the correct .dlls
 
Liger88 said:
One of my cores just went to the Core_15 8074 WU, so I smelled something was no good in my neighborhood.

Low and behold, new Beta units arriving :)

For those who aren't members, BOLO P8901 incoming!
Click to expand...

Yup a couple of my GPUs have started working on Core 15 work units.

Let's hope those 8901's start dropping soon or my eventual mowing may get pushed ahead of schedule by a few days (or weeks). :p

EDIT: My 7970 just grabbed a core 16 unit. So, now this has me wondering if this is another work assignment server glitch like the last time a couple days ago when the 8900's suddenly disappeared or if project 8900 is done.
 
Last edited:
Carbon_Rod said:
Yup a couple of my GPUs have started working on Core 15 work units.

Let's hope those 8901's start dropping soon or my eventual mowing may get pushed ahead of schedule by a few days (or weeks). :p

EDIT: My 7970 just grabbed a core 16 unit. So, now this has me wondering if this is another work assignment server glitch like the last time a couple days ago when the 8900's suddenly disappeared or if project 8900 is done.
Click to expand...




Ugghh they kinda suck too. Went from 7:30 tpf to 9:20-9:30 tpf for both my 660's. 15.5 hours for 25K. Gonna drop the PPD down drastically while taking a lot longer meaning 1 WU per day for each. Dunno where they come up with the figures if the P8900 was 40,000 atoms and the P8901 is 75,000.
 
I just got my first 8901 late last night. I had finally gotten the system with my 780's back up and folding after a motherboard swap. I was surprised to see that one 780 had grabbed another 8900 and the other grabbed my first 8901. But shortly after starting to fold, it had driver issues so I decided to shut it down until this morning in order to diagnose it. It's back up and running now but my actual PPD numbers are skewed because of the downtime.

But just for comparison, avg. TPF on one of my 780's for a 8900 is 3:01. Avg. TPF for a 8901 is 3:51. PPD looks to be about 2/3 the PPD of a 8900.
 
Liger88 said:
Ugghh they kinda suck too. Went from 7:30 tpf to 9:20-9:30 tpf for both my 660's. 15.5 hours for 25K. Gonna drop the PPD down drastically while taking a lot longer meaning 1 WU per day for each. Dunno where they come up with the figures if the P8900 was 40,000 atoms and the P8901 is 75,000.
Click to expand...


*EDIT*

Total PPD for 2 x 660's dropped from a solid 85K PPD (actual) to 75K (most likely 50K actual). TPF jumped up to 9:40 and its taking about 16.5 hours to fold one. Total suckage on these ones. :( Even worse than when it was just both GPU's doing nothing but 7663 units which was 80K PPD actual.
 
Last edited:
Gettin 5:27 TPF on 8901 on my 7970 for 90k PPD vs 110k-ish for 8900. A noticeable drop, but I'm not complaining after chewing on a terrible Core 16 unit all day yesterday for a measly 7k PPD! :rolleyes:
 
Activate: AMD said:
...but I'm not complaining after chewing on a terrible Core 16 unit all day yesterday for a measly 7k PPD! :rolleyes:
Click to expand...

I hear ya on that one! I just looked and the rest of my GPUs currently have 8901's thankfully.

I dunno what's up with HFM because it is reporting that my 7970 is doing a 5:20 TPF but getting 69K PPD?! Tried downloading all of the projects again and it still doesn't update. I know that 8901 appeared in PSummaryC before being actually released, and then they changed a few of its stats just prior to its release into beta.

The FAH client is reporting the correct values though and I've got 5:09 avg. TPF for about 98K on my 7970.

EDIT: FAH Client is reporting a TPF of 3:55 and an estimated PPD of 147K for one of my 780's.
 
Last edited:
Just an FYI, 8901 is approx. 75,000 atoms, 8900 is 45,000 so its a much bigger unit. Longer tpf is to be expected.

As for the points. Unless you are a beta tester you cannot complain;) If PG decide it wants a re evaluation of points they will do so
 
Nathan_P said:
Just an FYI, 8901 is approx. 75,000 atoms, 8900 is 45,000 so its a much bigger unit. Longer tpf is to be expected.
Click to expand...

True, which brings to mind the odd math. Nearly twice the size and lose 30%? :p


As for the points. Unless you are a beta tester you cannot complain [or be heard];) If PG decide it wants a re evaluation of points they will do so
Click to expand...


lol so true. There are quite a few beta testers raising their voice, so lets cross our fingers.


*EDIT*

It appears after the first P8901 WU the second one started off just like the first, with its 9:20 - 9:40 TPF. However, about 4-10% in (wasn't paying attention to know for sure) I looked back and suddenly the TPF dropped down to 3:54. I thought this was a clock skew (which happens ever now and then as we all know), but looking around I'm seeing on FoldingForums a few other people running 660's or 660 Ti's reporting the same with no change hours later (no one completed yet).

Looking at the logs I noticed a difference.

Code: 
00:51:31:WU01:FS01:0x17:Completed 0 out of 1000000 steps (0%)
00:55:49:WU01:FS01:0x17:Completed 10000 out of 1000000 steps (1%)
00:57:26:WU02:FS00:0x17:Completed 1175000 out of 2500000 steps (47%)
00:59:38:WU01:FS01:0x17:Completed 20000 out of 1000000 steps (2%)
01:03:24:WU01:FS01:0x17:Completed 30000 out of 1000000 steps (3%)
01:06:57:WU02:FS00:0x17:Completed 1200000 out of 2500000 steps (48%)
01:07:09:WU01:FS01:0x17:Completed 40000 out of 1000000 steps (4%)
01:11:01:WU01:FS01:0x17:Completed 50000 out of 1000000 steps (5%)
01:15:21:WU01:FS01:0x17:Completed 60000 out of 1000000 steps (6%)
01:16:57:WU02:FS00:0x17:Completed 1225000 out of 2500000 steps (49%)
01:19:08:WU01:FS01:0x17:Completed 70000 out of 1000000 steps (7%)
01:22:58:WU01:FS01:0x17:Completed 80000 out of 1000000 steps (8%)
01:26:28:WU02:FS00:0x17:Completed 1250000 out of 2500000 steps (50%)
01:26:48:WU01:FS01:0x17:Completed 90000 out of 1000000 steps (9%)
01:30:38:WU01:FS01:0x17:Completed 100000 out of 1000000 steps (10%)
01:34:51:WU01:FS01:0x17:Completed 110000 out of 1000000 steps (11%)
01:36:29:WU02:FS00:0x17:Completed 1275000 out of 2500000 steps (51%)
01:38:37:WU01:FS01:0x17:Completed 120000 out of 1000000 steps (12%)
01:42:23:WU01:FS01:0x17:Completed 130000 out of 1000000 steps (13%)
01:46:02:WU02:FS00:0x17:Completed 1300000 out of 2500000 steps (52%)
01:46:07:WU01:FS01:0x17:Completed 140000 out of 1000000 steps (14%)

They both started the P8901 initial one with 2,500,000 steps, the second one is only 1,000,000. Now I'm not an expert on this so maybe someone can break it down, but I think this WU is for real. Who knows if its perm or just a test on there end, but I'm liking this lol. Both my 660's would be pulling 300K PPD and complete each WU in less time than a single 12 hour 8900 WU, let alone the previous 16.5 hour 8901.

Anyone have thoughts or seeing similar results?

First GPU that started picking up 8901 WU's and starting the second.
jbuEsANh2x1xd4.jpg


Second GPU that is still on the first 8901 WU.

jwMLaCi7ap8xN.jpg



*NOTE* lol you can see just in the time it took to take the screenshots the visible % change :D
 
Last edited:
Maybe they changed the project mid stream? I mean it is beta after all.

On a side note, the 8900's are still around. One of my GPUs is working on one right now.
 
I just got one of the new 8901's on one of my GTX680's - frame time: 2:21 for 300k ppd (frame time averaged after work unit was 66% complete)....
 
loudbob said:
I just got one of the new 8901's on one of my GTX680's - frame time: 2:21 for 300k ppd (frame time averaged after work unit was 66% complete)....
Click to expand...



LOL enjoy it while it lasts. DAMN IT I knew something was up but I was hoping they were doing something to offset the 60% bigger WU and 30% less PPD drop, but apparently it was an error in the steps (1 million vs 2.5 million). At least you'll get every point, but the assignment server is shut down til at least Monday so I assume it'll be the more consistent 8900's for now.

I might have to move to advanced so long as it only gives out Core_17 8900 WU's if anyone knows. Can't do no 16 hour 8901 and go from 85K to 50K PPD. Maybe they change that when they bring it up lol. I'll be letting someone else beta test that theory on Monday though. Can't risk failing a WU that takes 16 hours.
 
Had been away for two weeks and checked my systems upon return:

P8901 has usually TPFs on my stock-clocked cards of

GTX Titan: 3:52 - 3:55
GTX 780: 4:23 - 4:27
7970 GHz Ed: 5:17 - 5:23

One of the Titans picked up one of those shorter P8901 units with 1.000.000 time steps and its TPF dropped to 1:27 - 1:30. This unit finished just normal and got its acknowledgement from the server.

Not sure if the client recognises this shorter units in the points they get awarded from the server, but if there is no difference in awarded points to the 2.5m timesteps version, then this shorter P8901 units would deliver 600k ppd. As only one unit showed up, the longer term impact can not be verified if this is indeed the case.
Screenshot

After all it is beta, and things are expected to change all the time.


Andy
 
Last edited:
Liger88 said:
I might have to move to advanced so long as it only gives out Core_17 8900 WU's if anyone knows.
Click to expand...

No, you'll be in the same pool as the core 15 and core 16 units as well. The closest you'll get to doing solid 8900's is to leave your client with the "beta" flag. That is until the 8900's dry up or if the assignment server has another issue, then even the ones with the "beta" flag will start getting the core 15 and core 16 units.
 
On a related note, has anyone analyzed the core_17 startup delay to see if anything can be done to minimize it? It doesn't appear to be specifically GPU related, as the GPU only spikes once before going to steady full load.

EDIT:
I asked the question on the fah IRC channel as well, from the man himself:
12:01 Sun proteneer_osx rhavern_home: hopefully the next major core release will deal with the core_17 startup delay
12:01 Sun proteneer_osx we know whats causing it
 
Last edited:
Just got assigned 2x 8902 WU's on a GTX680 and 670 an hour or so ago. Showing frame times of 2:14 and 2:39. For an estimated PPD of 340k and 264k. :eek:
 
terrabyte89 said:
Just got assigned 2x 8902 WU's on a GTX680 and 670 an hour or so ago. Showing frame times of 2:14 and 2:39. For an estimated PPD of 340k and 264k. :eek:
Click to expand...

P8902 seems to be nice units - if the reported values are true :D
Its a unit with 1mio timesteps (again)
original.jpg
 
Update:
Just lovely :)

Single Titan system
original.jpg


Dual Titan system
original.jpg


Dual GTX780 system
original.jpg


Dual AMD 7970 system A
original.jpg


Dual AMD 7970 system B
original.jpg


Total current ppd forecast for 11 GPUs: approx 6 mio ppd :D

.... I've seen the future (of FAH & GPUs) ....
 
Look out Grandpa, brilong, musky....

Looks like I need to sell off some hardware to invest in core_17-capable GPUs.

If you buy a Radeon 7970 for $330 today, that nets you over 1000 PPD per dollar based on those numbers. The GTX 780 isn't too far behind in terms of PPD/$ either.
 
Ya, I just got up this morning and was greeted with a 8902 on my 7970... I thought that must be a mistake. Guess it's real. Woot!
 
DooKey said:
That's just crazy PPD.
Click to expand...
At this rate DooKey, I would catch you in a few days (instead of the original 3 weeks) :D
There are "only" 28m points to cover.
Too bad, I don't have any spare Titans laying around ;)

Update:
In the last 2 hrs my systems finished 11 WU of P8902 projects.
According to Stanford's stats server 673k points credited
 
Last edited:
lol screw you greedy bastards. I'm happy with my 105K with two 660's on 7810/7811. Still have to pinch myself that 4 months ago I went from 6.8K on a GTX 260 to 105K with 2 x 660's. Best purchase EVAR!!!!!!


/can't wait for 8902 :D
 
I wouldn't say so.
I'd rather express our interest to capture this rare and unique opportunity - may it last a few days ... ;)

BTW, yodap (evga team) reports 254k ppds on his 660ti card.
 
would just like to throw out there that 6950 with 6970 shaders is making 45k ppd on 8902
 
Here are the number of atoms of recent core17 projects.
P7663: 22.000
P7810: 31.119
P7811: 23.055
P8900: 40.000
P8902: 75.000

The unusual high ppd of P8902 WU seems to be based on the 1m/2.5m timestep "glitch".

The current P8902 units are based on 1m timesteps.

With a Titan this gives a TPF of - for instance with a Titan - 1m20s.
TPF for a 2.5m timestep run would be around 3m20s.

According to the bonus point calculator,
a TPF of 1m20s gives 746k ppd
a TPF of 3m20s gives 189k ppd

While the first result is obviously much nicer, I assume the second one is the one as originally intended.

(....but I could be of course completely wrong .....)
 
^^^interesting.

Some times it pays off to beta test new units
 
P8902 WU's should not be given out anymore. Appears the server restarted and accidentally started handing out WU's that were not meant to be worked on yet. Still no confirmation on whether the points were intended or not, but it was made mention that they never did solve the mystery of the 1/2.5 million stepping issue with 8901 units.
 
Liger88 said:
P8902 WU's should not be given out anymore. Appears the server restarted and accidentally started handing out WU's that were not meant to be worked on yet. Still no confirmation on whether the points were intended or not, but it was made mention that they never did solve the mystery of the 1/2.5 million stepping issue with 8901 units.
Click to expand...

That's the first thing I thought of when I saw those crazy points. "How long will this last before they pull the 8902's just like they did the 8901's?" Guess you've answered that question.
 
Carbon_Rod said:
That's the first thing I thought of when I saw those crazy points. "How long will this last before they pull the 8902's just like they did the 8901's?"
Click to expand...
I think it is safe to assume that all of us expected this will not last long.

Anyway it skewed intended results even in the short duration it lasted.

I was trying to get to my first 3m ppd on sunday (without the P8902 effect).
P8902 added a few hundred k on top of it in the few hours it ran.

original.jpg


BTW, just for the record:
The GPUs had a signifcant higher energy consumption while they ran P8902 work units. With the other core17 work units, the Titan's for instance ran at 78-80% of TDP, while the P8902 units triggered a 88-90% level.

To put things in perspective - one example: My dual GTX780 system needed 420 watt (at the plug) for pre-P8902 units, while with P8902 it consumed 460 watt.

Looks like the core17 is maturing fast, gaining with successive iterations/datasets higher efficiency in the GPU (as observed via TDP).

rgds,
Andy
 
You must log in or register to reply here.
Back
Top