Video Card Recommendations?

Status
Not open for further replies.

Coputernewbie

Distinguished
Sep 25, 2011
310
0
18,790
So Ive finally discovered why my gaming computer kept saying "Display driver has stopped responding and has recovered" but I need a new card closely similar too or better than my current R6790. It needs to be around $140 and needs to support an 8pin PCI-E cable or have two 6pin PCI-E ports.

That is the design flaw in my current card. It only has 1x6pin port when it needs 8pins to draw the appropriate 150watts it needs.
Please keep in mind I have an Antec Earth Watts 650W PSU.

Thanks for all the advice in advance!
 
Solution
For drivers, there are a few utilities that will work. The last one I used with success was this:
http://www.guru3d.com/content_page/guru3d_driver_sweeper.html

When you go to add/remove in control panel and remove a program, it just removes "surface" files but leaves of lot of things behind. Something like this, when ran, will go through you system directories and registry and remove all old trace files. It should be pretty straight forward. After removing the drivers, hunt down the latest catalyst drivers and install for your OS version.

Voltage control may not be available for all cards. What you will have is the Catalyst software for your card and it should look like this under the performance tab...

matt_b

Distinguished
Jan 8, 2009
653
0
19,010
I think I'm a bit lost here. What problem have you been having, and what does the pin arrangement have to do ith fixing a problem here? Also, the graphics card draws 75 watts off of the motherboard, and gets another 75 watts from a 6-pin power connecter = 150 watts??? So what does the appropriate 150 watts mean?

Dollar range, the 6850 is as strong as you'll get for that $140. You'll have to catch a good sale to find 6870's for close to that price, stock is getting very limited on both though.
 

Coputernewbie

Distinguished
Sep 25, 2011
310
0
18,790
So like a lot of people I have the problem of my computer freezes for a second then comes back saying that the display driver has stopped responding and has successfully recovered. So for a long time ive looked up posts and read thing about the issue. There all kinds of fixes for it. So I changed the motherboard, and ran a check on all the components and everything came back functional.

Friday I spoke to a tech friend of mine who found it strange that my video card only had one 6pin power connector and said that and 8pin or 2x6pin connectors to give it the proper 150watts my card uses. This was later reassured here:
http://www.tomshardware.com/forum/372829-33-converter-problems#t2807819

So I was fairky certain in my problem until now. No offense of course! I was unaware that the card drew power from the Motherboard too. The person in the other forum post didnt mention this at all.
 

matt_b

Distinguished
Jan 8, 2009
653
0
19,010
I find this very puzzling about your card situation. On one hand you have a rare XFX 6790 with 1 six-pin power connecter, and most other manufactures chose the twin six-pin connecter route. Where this gets stranger is that the 6850 (which is a bit more powerful) has by default 1 six-pin power connecter AND uses a few watts more of electricity than the 6790 - yet there appears to be no problem as many have this card (more owners than the 6790 for sure). So taking into account what KO888 told you in the other post, this has to be a screw up on XFX's part. If he is putting the blame on the power connecter, then have you tried contacting XFX to see what their response is? If this is the underlying cause, then they should step up and take responsibility on what looks to be a design issue with their card only. I can tell you for a fact though that this shouldn't be a power draw issue given what this card is capable of but either a design or manufacturing flaw with XFX - how else can the 6850 and it's 1 six-pin connecter be explained?

The "crashed and recovered" thing is going to be the most common problem with Radeon cards. I see you have an Earthwatts 650 PSU in your sig, which version do you have btw, (they make 2 versions but they're quite different in design)?
 

Coputernewbie

Distinguished
Sep 25, 2011
310
0
18,790
When I ran test on my parts everything came back fine but I couldnt check the CPU or card itself. A tech friend said if there was anything wrong with the CPU the the computer wouldnt even work to begin with so I spoke to XFX about the card. I got it sent in and got a new one back (a 6790) but the "Crashed and Recovered" still occurred. I was stumped until, like I said, talked to a friend of mine on friday. And I dont any other card to test if its a flaw in that cards design in anyway.

On kind of a side note, when I was first building my computer, I didnt put the spacers between the board and case but later realized thats why it didnt start when ever I tried to turn it on. Could that have effected anything?

My PSU is:
http://www.newegg.com/Product/Product.aspx?Item=N82E16817371044

OR

Model Number: DT1107DZE0005217

According to this website, the 6850 actually uses less power then the 6790:
http://www.hwcompare.com/9995/radeon-hd-6790-vs-radeon-hd-6850/

if you have a better way or site for comparing parts let me know
 

matt_b

Distinguished
Jan 8, 2009
653
0
19,010
I've seen a few sites have the 6850 at basically a tie with the 6790, but I do in fact see more with the 6790 using a few watts more. This is still too close to one another (going by actual and not TDP rating) that the extra power connecter shouldn't be that important. A thing about TDP though is that it is designed power usage - basically meaning the product really shouldn't use any more than this number, it isn't actual power consumption.

About the PSU, if it was the other one Earthwatts unit, then that type would require load balancing and I've seen people load too much on only 1 of the rails and experience horrible stability because of it.

Yeah the no spacers are a bad deal as the entire board is exposed and the back panel on a case is metal. Don't feel bad though, I know someone that forgot a standoff and after a year it stopped working due to the board warping and solder breaking contact from so much stress in that spot - not sure how the screw reached though.

I would assume this computer has had a clean driver sweep and all drivers up up to date then? If nothing else and as a last resort, have you tried to under-clock it a bit or add a little voltage (if possible) to see if it still behaves this way in case power is suspect? I've had cards in the past that would behave this way, one was a card pushed too hard without voltage compensated, one was where the 12volts from the PSU would drop as low as 11.5 under load (plenty of rated watts but crappy brand), another time I found drivers for another card in the system folders, and the last time it was something I had no idea until one day it just simply went away (must have been drivers on this one). Maybe some of these other experiences may give you some new ideas. I'd hate to have to replace something for working but flawed with a company not willing to back it up or just give me my money back in return.

 

Coputernewbie

Distinguished
Sep 25, 2011
310
0
18,790
well to start off, I dont have a standoff in one spot but figured that the screws that are meant to support the GPU would compensate...

As for the drivers, they were clean at one point. Im not so sure now because I now have "auto detect and install" as trying to fix this problem sometimes requires me to switch between the on board and my actual card.

Ive never messed with clock or voltage and am not sure how to even do that. Ive had a recommendation on how to fix this problem bring something like that up. I believe he wanted me to mess with the idle clock speed (which I didnt wind up doing). And at some point I changed some TDR value to 8 or something. (the amount of time the PSU gives the GPU to respond before resetting, which was also thought to be the problem)

How would I go about checking the driver files or under clocking? would that void anything?
 

matt_b

Distinguished
Jan 8, 2009
653
0
19,010
For drivers, there are a few utilities that will work. The last one I used with success was this:
http://www.guru3d.com/content_page/guru3d_driver_sweeper.html

When you go to add/remove in control panel and remove a program, it just removes "surface" files but leaves of lot of things behind. Something like this, when ran, will go through you system directories and registry and remove all old trace files. It should be pretty straight forward. After removing the drivers, hunt down the latest catalyst drivers and install for your OS version.

Voltage control may not be available for all cards. What you will have is the Catalyst software for your card and it should look like this under the performance tab:
http://www.legitreviews.com/images/reviews/1585/ccc-6790-idle.jpg
I think the default clocks for your card are 840 MHz for the core, and 1050 MHZ for the memory. Click the padlock icon to unlock, and take the slider for the GPU core and move it left and try lowering it from 840 to 820/810 and click apply in the lower right - done. Overclocking would obviously be the opposite. This does not harm/void anything if under-clocking, it actually puts less stress and demand for power consumption on the card. For more advanced tweaking and possible voltage control, most would use something like MSI afterburner. If you have voltage control and Catalyst didn't list it, it would look like the slider at the top of Afterburner like this:
http://i1034.photobucket.com/albums/a427/MrDomRocks/GPUOverclockto580specs.png

Checking for stability, try this program. It's fairly common and is a good test because it isn't a synthetic test but rather a real-world kind of test. Just run it a few times and see how your system runs and note any behavior like you have been having.
http://unigine.com/products/heaven/

I can tell you for a fact that if the drivers are indeed good, when I overclock a card in stages slowly at a time, I will encounter a driver stopped responding message if I do not encounter a lockup first. What fixes this almost every time is adding voltage (within reason) because the GPU/memory cannot sustain itself on such little power supplied. So maybe this will at least give you some ideas and hopefully a free solution for now.

*Edit* - of course any diagnosing is trial and error so do things in stages instead of all at once.
 
Solution

Coputernewbie

Distinguished
Sep 25, 2011
310
0
18,790
I finally had the time to mess with the underclocking.

So I moved it up and down from 840MHz to 810MHz but it only helped a small portion. It didnt stop the message completely. Through all the settings it hadnt crashed only once or twice during each test.

I noticed during the Windows Media Player test it seemed to bounce back and forth between 100MHz to whatever I had the setting as. Its worth mentioning that this didnt happen every time. (It always crashed on the longer video files but not always the smaller ones)

During the MiceCraft test the clock speed sat at max whatever the setting was almost the whole time! And eventually carshed as it almost always has.

On the YouTube test, the clock speed would spike to max setting during the load time then sit a 300MHz

The Temps didnt go above 50 C or so during these tests.

When I first built the computer, it never crashed. If I missed anything youd like to know plz let me know.

Note: It crashed while I was typing this and the clock was at max... not sure why.

Its been awhile since I did a clean re install of the GPU drivers. Should I Take the Card out, wipe the drivers reinstall them and put the card in? How should I go about it?
 

matt_b

Distinguished
Jan 8, 2009
653
0
19,010
When I first built the computer, it never crashed. If I missed anything youd like to know plz let me know.
This is probably an important clue. If it's always done it and you've just finally had enough, this is one thing. If it came out of nowhere and/or got progressively worse, then it is either hardware that is failing (not likely on this one yet) or probably something wrong on the software side. So trying to rule-out software here:
Its been awhile since I did a clean re install of the GPU drivers. Should I Take the Card out, wipe the drivers reinstall them and put the card in? How should I go about it?
You do not have to physically remove the card. An OS has default display drivers to at least give you a picture.
1) Try the link I gave you at the top of the previous post to the Guru3D site. Download it
2) Find the newest driver for your card and the OS you're using on AMD's site.
3) remove the Catalyst and ATI Radeon driver package(s) in the add/remove section of Windows
4) Reboot (do so anyway if not prompted).
5) Run The driver sweeper program, it should find the ATI/AMD drivers and leftover files by default, and remove them.
6) Install newest Catalyst Driver suite.

Run in to an issue, just ask.....


You could try this also. It's rare, but sometimes demand exceeds the timeframe that a card can upclock itself to full operation capacity and it crashes.
 

Coputernewbie

Distinguished
Sep 25, 2011
310
0
18,790
Well Ill be damned!!

The last driver sweeper I tried didnt seem to work but Im not having any troubles now!

Thank you so much matt_b!! It doesnt seem like we stayed on topic exactly but Im really grateful for the help! This problem has driven me inane for so long! Thank you thank you. :)
 
Status
Not open for further replies.