GTX660 Problem

MAGIC Quantum Mechanic
MAGIC Quantum M...
Joined: 18 Jan 05
Posts: 1695
Credit: 1043184293
RAC: 1367286

RE: Magic and Nobody316

Quote:

Magic and Nobody316 -

I bought a new power supply with the GTX 660. It a Ultra LSP 650 watt.

The quad core CPU is running at a cool 47C as I write this with 4 CPU WU's and one GPU WU running.

I ordered another 4GB of memory. It only cost 1/2 of what it costs to take my wife out for dinner so I figured that was a good deal.

I will report back after I install it and try Einstein again.

True and it is always the cheapest and easiest way to make thing run better.

Let us know after you run with the ram upgrade for a couple days.

I even did that with the laptop I am on right now making it have 8GB and while I am typing away it runs BRP cuda X2 and T4T X2 and GRP X8 with no problems ever and I have been running it 24/7 since I got it last June (I do have a small fan blowing on it at when I am not using it and have it in the coolest room in the house)

The main thing I always do with new parts is register with the company (PSU's,Ram,and my EVGA cards)

WB8ILI
WB8ILI
Joined: 20 Feb 05
Posts: 45
Credit: 924195877
RAC: 230464

For Magic (and others)- I

For Magic (and others)-

I added 4GB memory to the existing 2GB. Since this is a 32 bit system, Windows only uses 4GB. But still, I doubled my available memory.

No change. The screen locks up after about 10 seconds of running an Einstein WU. It sort of "wakes up" and updates the screen every 30 seconds. I let it run until the system crashed (10 minutes). The System Event Log is filled with NV Error code 14 errors (about one every 30 seconds). The final error is NV4_Disp.dll is in an infinite loop.

As I wrote before, the card works on SETI and Milkyway. I have been cranking out SETI and Milkyway WU's 24/7 for 6 weeks now without any problems.

I think I will just accept the fact I can't run Einstein with my configuration. It is still being put to good use.

Nobody316
Nobody316
Joined: 14 Jan 13
Posts: 141
Credit: 2008126
RAC: 0

RE: For Magic (and others)-

Quote:

For Magic (and others)-

I added 4GB memory to the existing 2GB. Since this is a 32 bit system, Windows only uses 4GB. But still, I doubled my available memory.

No change. The screen locks up after about 10 seconds of running an Einstein WU. It sort of "wakes up" and updates the screen every 30 seconds. I let it run until the system crashed (10 minutes). The System Event Log is filled with NV Error code 14 errors (about one every 30 seconds). The final error is NV4_Disp.dll is in an infinite loop.

As I wrote before, the card works on SETI and Milkyway. I have been cranking out SETI and Milkyway WU's 24/7 for 6 weeks now without any problems.

I think I will just accept the fact I can't run Einstein with my configuration. It is still being put to good use.

Ok if you would try something and see how it goes and post about it...

Try running only E@H and nothing else and see if it still gives error... looking up the error's you posted now and will post back after I find something.

Edit:
Ok I see ya tried that already... ok still looking

Edit:
Ok lets try this... What is your screen resolution and refresh rate... maybe try adjusting them and see if this helps... look into reading minidumps and maybe post it as well

PC setup MSI-970A-G46 AMD FX-8350 8 core OC'd 4.45GHz 16GB ram PC3-10700 Geforce GTX 650Ti Windows 7 x64 Einstein@Home

WB8ILI
WB8ILI
Joined: 20 Feb 05
Posts: 45
Credit: 924195877
RAC: 230464

Nobody316 - I have tried

Nobody316 -

I have tried different screen resolutions in the past but to no avail. I can't say I tried every one. I know I tried 1280x720. I am currently using 800x600.

Per your suggestion, I tried (besides the 60Hz I am using), 72 Hz, 85 Hz, and 100 Hz. I didn't try 56, 70, or 75 Hz as I got tired re-booting the computer. The screen still "locks up" as I have described previously.

I also tried slowing down the GPU. The default is 1058 MHz. I tried 744 MHz and 550 MHz but the results are the same (bad).

Still cranking away on SETI and Milkyway.

Horacio
Horacio
Joined: 3 Oct 11
Posts: 205
Credit: 80557243
RAC: 0

I've seen something similar

I've seen something similar as the OP issue on one of my hosts...
Its an I7-3570 with a gtx-690 on a Gigabyte MB with PCI3.0 8Gb RAM (but only 4 used due to Win7 Pro 32b)... the 690 is lightly OCed to match the speeds of a GTX680, there is no PhysX installed and neither any of the optional drivers for 3D monitors and the windows is configured to not use any of the fancy effects.

Ive not noticed this until recently because I use that host almost allways remotelly with VNC and the issue doesnt affect the mirror video drive used by VNC...

But using the computer locally, the video gets freezed for several seconds every time a new BRP WU is finished. Im not sure if it gets freezed also on another stages of the process, but I think it does.

This issue only happens with Einstein's BRPs and it doesnt matter how much WUs per GPU are running.

The whole computer keeps working, in fact the mouse cursor keeps moving on the screen while the screen is freezed and it keeps responding to clicks... Its just the video that doesnt get refreshed... Ocassionally this freeze seems to produce a driver failure and then windows reset the driver (ruining the current WUs beeing crunched) but after the drive is restarted everything works as ussual.

Im using the driver 310.70. In another host on which Im using 306.97 and 2 GTX560Tis I dont have this issue. (I dont know if it is due to the driver or because they are Fermi GPUs instead of Kepler)

Nobody316
Nobody316
Joined: 14 Jan 13
Posts: 141
Credit: 2008126
RAC: 0

WB8ILI: last option i can

WB8ILI:

last option i can think of... Try changing hardware... ram and GPU... only thing I can think of is hardware... If you can change out ram and test if no go then try changing GPU to another computer. If still no go then I am out of ideas... atleast for now...

PC setup MSI-970A-G46 AMD FX-8350 8 core OC'd 4.45GHz 16GB ram PC3-10700 Geforce GTX 650Ti Windows 7 x64 Einstein@Home

WB8ILI
WB8ILI
Joined: 20 Feb 05
Posts: 45
Credit: 924195877
RAC: 230464

Horacio - What you wrote is

Horacio - What you wrote is consistent with my situation. The mouse moves around OK. And after 30 seconds the computer realizes what I have clicked on. Then the computer goes to "sleep" for another 30 seconds. Also, the computer (except for the screen) is running. I can access it over my local network with no delays in response. I wonder if you see any errors is the Windows Error Event log.

Nobody316 - I started with a 2GB memory module. Then I added a 4GB giving me 2GB+4GB. I am not sure what part of that 6GB is the 4GB Windows uses. Then, I pulled out the 2GB module, leaving me with the new 4GB module. So, I essentially have tried at least two different memory chips. I could move the GPU plus 650w power supply to a 2nd computer but the 2nd computer has exactly the same MB and processor. Somehow, that seems like a low probability of success.

As I wrote before, sometimes one just has to accept the fact my combination of H/W, S/W, etc. and Einstein aren't going to get along.

Nobody316
Nobody316
Joined: 14 Jan 13
Posts: 141
Credit: 2008126
RAC: 0

If it is only 1 of your

If it is only 1 of your computers that are running the same specs then I really think it might be hardware problem. Just because you don't have problems with other sites don't really mean anything as E@H seems to run different than other places and more so for the GPU work units. Also stress test don't seem to be good enough to count on fully. In the end it's up to you if you want to test the idea or not. Buying bad parts is not uncommon and goes unknown until something like Boinc runs and pushes the hardware to it's max which is more than stress test. I hope everything works out for ya but as of now I don't have any other ideas beyond hardware.

PC setup MSI-970A-G46 AMD FX-8350 8 core OC'd 4.45GHz 16GB ram PC3-10700 Geforce GTX 650Ti Windows 7 x64 Einstein@Home

Horacio
Horacio
Joined: 3 Oct 11
Posts: 205
Credit: 80557243
RAC: 0

RE: Horacio - ... I wonder

Quote:
Horacio - ... I wonder if you see any errors is the Windows Error Event log.
As I wrote before, sometimes one just has to accept the fact my combination of H/W, S/W, etc. and Einstein aren't going to get along.

I have not found the error 14 on the event log, but that host is mainly a SETI cruncher that only does Einstein as backup when SETI fails to give me work...

And Ive noticed this because my mother was here using that host to play some light internet flash games and in one particular game with a timer she was having big issues so she told me that something was going wrong...
As this issue doesnt affect my normal usage of that host I didnt made too much research, but seeing that there is somebody else with the same issue I thougt it was worth to share my case also...

I agree with you that there are combinations of SW and HW that sometimes fail and is very hard to find why... and when the issue arises only using a certain and specific app, then its rarelly just a hardware problem... (but that doesnt imply neither an app bug...)

Sadly, figuring out what it is requires too much work as it could be anything, from a bug in the code of the apps, a bug in the compilers used to make the apps, a bug in the drivers related, a bug in the firmware, but also it could be a completely unrelated app or piece of hardware which interferes... and to make everything worse, it could be more than one bug on different things which only appears when those things are mixed in certain way...
So, checking all this requires a combined work between the user, the programmers of the apps, the makers of the compilers, the makers of the drivers and the builders of the hardware...we know, that kind of cooperation wont happen anytime soon... specially if the issue only appears sporadically or in a very few hosts.

Cliff Harding
Cliff Harding
Joined: 23 Jun 05
Posts: 14
Credit: 98412536
RAC: 470741

WB8ILI -- I may be too

WB8ILI --

I may be too late for this conversation, but it will be here for anyone else that has a similar problem. I also have an EVGA GTX660 card and it seems that your problem could be combination of things.

1) If you haven't installed EVGA's Precision X from their site, do so as soon as possible. In the upper right corner of the app, you'll see a set of gears, click it. Under both the General properties & Fan tabs make sure that both boxes are checked. Close the properties and make sure that the Windows startup and the Auto boxes checked at the bottom of the app. This will ensure that the fan control is automatic and will kick in upon Windows startup. Finish by minimizing the app to the notification area -- DO NOT CLOSE THE APP!! If you close the app you will lose auto fan control and the card WILL HEAT UP. It took me about a week to figure this out.

2) Make sure that you DO NOT RUN ON ALL CORES. You must reserve at least 1 core to feed the GPU. I run all of projects on 6 cores out of 8 (SETI, Einstein, Milkway) with each tasks taking what it needs to feed the GPUs. You are running on an AMD board, so someone else will have to help you there, but I think I heard either here or on SETI that you have to have 2.

I don't buy computers, I build them!

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.