Sign in with
Sign up | Sign in
Your question

Artifacts with Graphics card HD4850

Last response: in Graphics & Displays
Share
a b U Graphics card
January 6, 2011 2:01:26 AM

Alright well I upgraded the power supply on one of my machines to support an HD 4850 but I am still getting artifacts and crashes.

I originally put the card into the Dell and got artifacts, and my conclusions was that the original Dell PSU wasn't powerful enough. So I purchased a OCZ ModXStream Pro OCZ500MXSP 500W ( http://www.newegg.com/Product/Product.aspx?Item=N82E168... ) and installed it.

It allows for playing 1v1 games in SC2, but if we try for a 3v3, we see artifacts and it blue screens. TF2 isn't playable at all.

I just did a OCCT test and the system had the artifacts and crashed again.

What could be wrong? The power supply is for sure good enough.

I have no clue at this point, please help! :) 

Edit: Please see this post for latest updated: http://www.tomshardware.com/forum/forum2.php?config=tom...
January 6, 2011 3:55:28 AM

Whats the card temp idel and under load
a b U Graphics card
January 6, 2011 12:19:09 PM

The card runs about 60C idle, under intense load I don't know since it ends up giving up artifacts and then a blue screen, but I am pretty sure it isn't overheating.

I used to use this card in another computer for over a year and it worked great. But when I upgraded, I moved the card to the current machine and ever since then I get artifacts and blue screen.

This is Windows 7 32-bit with the latest CCC.
Related resources
a b U Graphics card
January 6, 2011 2:31:51 PM

I'm wondering if the problem is because the PSU has two separate 12V rails? Does this mean that if I connect just one cable from the 12V output into my GFX card then I'm only getting the 18A? Each 12v rail has 18A on this PSU...
a b U Graphics card
January 6, 2011 2:33:39 PM

Btw, this is the PSU's back, the RED inputs are for the 12V I believe. There comes with two cords, one that is a 6pin and another that has a 6+2pin.





a b U Graphics card
January 6, 2011 3:18:43 PM

It sounds like the vid memory is going bad. You may not notice unless the game is utilizing more vid memory in which case it is running into the bad part of the ram.

Also, your temps seem high. I have an MSI HD4850 OC and it idles around 36°C and gets into the high 50's under load.
a b U Graphics card
January 6, 2011 3:25:47 PM

Great :(  I'll take the card out and test on another machine to be sure. Good thing XFX has a lifetime warranty, hopefully can RMA it for another one.

Is there any way to test the vid memory on the card?

Also, if anyone has other suggestions let me know.

Also, was I totally wrong about the 12v rail?
a b U Graphics card
January 6, 2011 5:15:29 PM

Awesome Hawkeye22, I'll give it a shot tonight and post back. I'm kind of hoping this is the problem, much easier to deal with since I'd assume this is covered in warranty.
a b ) Power supply
a c 165 U Graphics card
January 6, 2011 5:30:47 PM

You really need to eliminate temperarures as a issue first, the HD4850 was well known for running hot, especially with the stock single slot cooler, and this is going to be more of a problem in a case with limited airflow.
Download and run Furmark, it'll stress the card and should show the GPU temperatures while it's running, if it shows >100C under load, just ESC to end the test.
You might also want to check the CPU temps as well, Speedfan and Coretemp are free and seem reliable.
a b U Graphics card
January 6, 2011 5:36:26 PM

Yeah, that is a good suggest, the only problem is as I stress test, I get major artifacts and then crazy random that covers the screen and then the blue screen. So I haven't been able to see the temp after a few seconds of serious stress testing. The computer is still running, we even had skype working and talking for a few minutes while the screen was all rainbows, but then eventually we have to restart because it is stuck.

That is what is annoying, I can't see anything to test anything and even when the stress test ends, the screen doesn't come back to normal, just scattered rainbow colors everywhere.
a b ) Power supply
a c 165 U Graphics card
January 6, 2011 5:45:34 PM

Ah. Right. I see.
Go into the CCC and unlock Overdrive, then LOWER the clocks and retest, a rule of thumb is to lower the memory 2MHz for evey 1 MHz of core speed and start by dropping the core in 25MHz increments.
I suspect this may help but that the other posters and you are correct: the card is on its way out.
January 7, 2011 12:03:50 AM

Good thread.... Yes you were mistaken about the +12V rails, single or multiple rails dont matter its all from the same source. 60C is high, I have used 4850's for a couple of years now and 60c is a load temp usually.
a b U Graphics card
January 8, 2011 7:43:11 PM

I ran the check on the memory and no errors came up, so it looks like that isn't the problem...

I'm going to bring the card here and test it on my newer machine that was running it fine for a year. If it works, then it must be the new PSU has some problems or something else we haven't figured out yet.
a b U Graphics card
January 8, 2011 9:53:59 PM

Also, it looks like the card is idling at 42C. I think it was on load getting to 60c, so that is better now.
a b U Graphics card
January 9, 2011 3:52:06 AM

Well, an update. We can run OCCT without it crashing now. The temp gets to maybe 70C max with running full screen at 1920x1080 with OCCT.

We were happy so we went into TF2, played for about 2 minutes, but then the problem came back, artifacts and then a black screen. No blue screens, just a black screen and the DVI turning on and off (monitor losing signal and getting it back). No way to get out of it except to shut off the comp, but mouse and keyboard are responsive.

Sees so weird that OCCT doesn't crash it, but TF2 does. We reinstalled Windows 7, got a new HDD, and everything...

The last thing to try is swapping it into my newer system and see what happens. Gonna do that now.

Any idea why OCCT isn't triggering the issue? It is stressing my card, putting up 180FPS but I get no crash. What are the games doing differently that is making the card stop working?
a b U Graphics card
January 9, 2011 5:58:58 AM

I put the card into my newer system and it worked perfectly fine like it used to. So I have ruled out 100% that it is something wrong with the card itself.

I'm going back to saying it is the PSU. I realize when OCCT is running the GPU test, the CPU isn't drawing power so there is enough power and it works. To test this theory, I did the Power Supply test, that taxes the GPU and CPU, and BAM! I had the issue again.

So this new power supply is much better, we can play a few minutes of TF2 but then the load seems like it is too much.

Do you think my PSU is defective? I just did an Antec PSU calculator and it recommended 420W for my system, even this card only asks for 450w, so this OCZ one should be enough.... its brand new.

Should I RMA the supply? Really need some advice.

Thanks!
a b U Graphics card
January 10, 2011 11:09:10 AM

It's hard to say. It's strange that it works fine in your newer system and not in the old. Maybe there is a problem with the PCIe slot on the dell and it's not getting enough voltage. It's rare to find a PSU that's faulty from the get go. 500W is plenty for the vid card depending on what else you have in your system.

Anyhow, if you are going to try a new PSU, I'd stick with a 500W or slightly larger rather than following the PSU calculator. You want to at least stick with the 450W recommended by the vid card.
a b U Graphics card
January 24, 2011 12:49:09 AM

I'm bring this back because we are still having the issue. I RMAed the power supply and have a new 500W one.

If I run OCCT for the power supply test, I am able to do it without any issues. I run OCCT for graphics card test and I have no issues either.

When I try to play games like TF2 or SC2 I get the issue again. Basically the screen gets strange colored thin vertical lines then turns black. Then I can see my cursor and even move it but I can't get the game to show up or anything else (even with alt+ctrl+del or alt+tab). This continues like this and even the monitor loses signal and gets it back continuously. I have to hard restart to get my system back. I noticed that the game is still playing, I can hear the sound, but the monitor can't display anything.

I thought maybe it was the system RAM but I did memtest and it turned out fine.

This is what I've tried:
1) New 500W power supply
2) Another new 500W power supply
3) Put the card in another computer to test (worked fine)
4) Graphics card memory test (passed)
5) Memtest (passed)
6) Fresh install of Windows 7 (on a new HDD even)
7) OCCT power supply test (passed)
8) OCCT graphics card test (passed)
9) Older Radeon x300 in computer (works fine)

So I'm totally stumped as to what the problem can be. I think it can't be the power supply anymore. The only thing left is the PCI-E socket, but it can run an x300 without any issues at all. PLUS OCCT can get the card to 100% activity without any problems at all.

So... what the heck? Why is this happening? Any clues?
a b U Graphics card
January 24, 2011 6:49:13 PM

Any Ideas?
a b U Graphics card
January 25, 2011 2:06:20 PM

Experts, any advice? :) 

Maybe this thread is too long, make a new one?
a b U Graphics card
January 25, 2011 2:37:36 PM

I know you said you tested the video ram, but it still seems like a vid card issue to me. Is it still under warranty? can you RMA it? Other than that, I'm out of ideas.
a b U Graphics card
January 25, 2011 2:48:28 PM

I tested the video card on another computer and it worked without any issues at all. So with that and testing the video RAM, I can't imagine the problem is with the card.

It just seems like everything is working but it still isn't working...

Anyone have experience with OCCT that can tell why it would work fine with it but not for games?
a b U Graphics card
January 25, 2011 4:07:53 PM

I read through all the posts again.

1. You swapped power supplies.
2. You tried the card in a different computer and it worked.

Have you tried another video card in this computer and if so, did it work?

If you ruled out the video card and power supply and you didn't try another card (of equal or higher wattage) in this machine, then perhaps the PEG slot is going bad or maybe even part of the motherboard that supplies power to the PEG.
a b U Graphics card
January 25, 2011 4:34:39 PM

I only tried a Radeon x300 in the PCI-E slot and it worked fine, but that card doesn't require extra power (no 6 pin required).

The only other card I have is a 6850, which I could give a shot.

If the PCI-E slot is bad would the x300 be able to run without any issues? And would I be able to get the 4850 to run at 100% load in OCCT? Both of these are the case right now. I'm not totally clear what it means to have a bad PEG slot.
a b U Graphics card
January 25, 2011 6:06:20 PM

The PCIe verson 1.1 slot can provide up to 75 watts via via the slot itself. This is in addition to what the PSU directly supplies the card with. Your x300 is drawing power directly from the slot. I'm not sure how much the PCIe 2.0 spec allows for. Anyhow, if power to the PEG slot is unstable, the vid card won't work properly.

By bad PEG (PCI Express Graphics) slot, I mean it has bad contacts or is shorted or the motherboard blew a cap or something that would cause the slot to get less power or fluctuating power.

Try the 6850 and see if it does the same thing.

Also, I know you said it's not a heat problem, but have you tried playing games with the case open and a fan blowing on the vid card?
a b U Graphics card
January 25, 2011 7:19:06 PM

Thanks for the replies.

It is looking like probably the PEG is the issue but two things I'm still not clear about is how the card can be pushed to 100% via OCCT and not have any issues, if there was a power issue with the PEG socket. I'm going to try some other programs to test and see if they run stable as well.

I'll also try the 6850 but have a feeling it will fail to work as well.

As for the fan blowing on the card, I haven't tried that. The case is open and the fan is pumping air out without any issues. The temp doesn't even reach 70C on full load.

I'll run some more tests tonight and see what I get back. Hopefully running ATiTools and furmark will help me figure this out better. I'll also run OCCT for a little longer. If these three don't throw me errors though, I'm really stumped.
a b U Graphics card
January 26, 2011 10:39:56 AM

It's definately a tricky situation, especially since some of your tests result in conflicting information. Best of luck to you.
January 26, 2011 11:55:01 AM

In CCC turn the clocks back like someone else suggested or cool it somehow this will fix it as I believe it is a heat issue I had the same problem when I OCed my 4870 while playing Crysis.
a b U Graphics card
January 30, 2011 2:47:55 PM

I've been testing the computer and card and these are my results.

Prime95 - Passed
FurMark Stability test - Passed
OCCT 1 hour - Passed

Play the original Day of Defeat doesn't case any problems (I played for 20 minutes online). Playing an indie game called Aurora doesn't cause any problems. Playing Peggle works as well.

SC2 is still causing the problem. Sometimes it happens on the loading screen, sometimes after I login, and sometimes after I join a game.

TF2 has the problem still, after I've joined a game.

Portal has the problem, on the loading screen.

I know for sure this isn't a heat issue. I inspected the motherboard and it doesn't appear damaged at all. The PCI-E slot is firmly attached with no visible damage.

I'm going to keep testing different games but this is really weird. How can OCCT, Primr95 and Furmark all pass, and even older games work, but as soon as I try a newer game the display messed up?

Also, just to make it clear, the computer doesn't shut down. Usually I see a few weird color dotted vertical lines (but sometimes I don't) then the screen goes black. The system is still running but the I/O cuts out and in intermittently. I can never get the display back and it turns on and off (loses signal and gets it back), but my keyboard and mouse do the same thing.

I just did a sweep of the gpu drivers and got the latest ccc. One thing is it doesn't open, which is a new issue, but I think most likely unrelated since I've had this issue when the older version of ccc used to open.

This is so frustrating, does anyone know a case like this where you would have all 3 of those GPU testing programs pass but still can't play some games?

Edit: Just tried out Torchlight and the game works as well on full graphics... :pt1cable: 
a b U Graphics card
January 30, 2011 2:51:48 PM

Also, when I used to have this working, it was with Windows Vista 32-bit about a year ago with the original power supply. That configuration worked with this graphics card for more than a month. Now I'm running Windows 7 32-bit and have a new HDD and new PSU but the system is the same.
January 30, 2011 8:01:04 PM

have you tried different drivers or are you only testing with the lastest?

I have an HIS HD4850 it was working fine till i got BFBC2 . I had to downgrade all the way back to the 10.4 to get it working again. My machine was hard locking up as well when i launched the game.
a b U Graphics card
January 31, 2011 3:00:31 AM

Thanks for the tip, I've now tried about 5 different versions of CCC but to no avail. One good thing is I can finally open CCC though.

I tried playing SC2 as a guest account, not online, and was able to play a whole mission and keep the gaming running for more than an hour without any issues. When I quit the game though, to go back to the home screen (login main screen) I got the problem and now my screen is black and like described above.

I just can't understand. I will be trying another PSU tomorrow, one that I know works for sure to see if that could be it, maybe the voltages aren't stable...
a b U Graphics card
January 31, 2011 9:52:42 PM

An Update:

I've now tested out a bunch of combinations and in the end the 4850 just doesn't work with this computer.

Dell + OCZ PSU + 4850 = failed
Dell + Cooler Master PSU + 4850 = failed

Dell + OCZ PSU + 6850 = passed!
Dell + Cooler Master PSU + 6850 = passed!

New Computer + Cooler Master PSU + 4850 = passed!
New Computer + Cooler Master PSU + 6850 = passed!

As you can see from the above, the graphic cards are fine and the PSUs are fine. The PCI-E in the Dell is good too because the 6850 works! And the 4850 is good because it works fine in the newer computer.

I reapplied thermal paste to the 4850, dropped the temp down by 10C!! but still that hasn't helped. I reapplied the thermal paste on the Dell's CPU, but I can't tell the temp on it though.

The only thing I can think is that it an airflow issue. The 6850 has dual fans on it and so the air flow is quite a bit. The 4850 has the stock cooler on it... The only thing is the 4850 never gets hot enough to warrant a shutdown. I can't tell what the CPU temp is though, maybe it is getting too hot?

So I could just put the 6850 into the old dell and the 4850 into the new comp. It seems so silly to do it since the Dell can't utilize even the max 4850 (CPU bottleneck).

Am I just out of luck. The Dell is a Dimension 9150, here is an example (this is with a x600, mine has the 4850 in its place) of what it looks like inside:



As you can see, the MOBO is on the opposite side, so the graphics card points up instead of down, and half of the GPU's fan is under the plastic heatsink cover... but still I want to stress the point that the GPU temp isn't getting that high, but I have no way to read the CPU temp.

Any more advice would be GREATLY appreciated.

Thanks :) 
a b U Graphics card
February 1, 2011 1:00:33 AM

Another Update, an Important one.

I decided to plug in two monitors with the 4850 in the Dell and it worked without any issues at all. I was able to play TF2 for as long as I wanted with no black screen or problems.

So I decided to minimize, take out one monitor and then maximize, and as soon as the game loaded, the problem came. I tried a bunch of different combinations of this, switching the display ports being used and it was consistent.

So, the 4850 only works with two monitors connect to it. If only one monitor is connected, I cannot play games. (note: this is only the case on the Dell, on the other computer it works regardless)

Anyone experienced this problem? What could be causing this?
a b U Graphics card
February 2, 2011 6:33:34 PM

Any ideas at this point? I feel like this symptom may be the key to figuring out the whole problem.
February 2, 2011 7:46:54 PM

Pallimud said:
Another Update, an Important one.

I decided to plug in two monitors with the 4850 in the Dell and it worked without any issues at all. I was able to play TF2 for as long as I wanted with no black screen or problems.

So I decided to minimize, take out one monitor and then maximize, and as soon as the game loaded, the problem came. I tried a bunch of different combinations of this, switching the display ports being used and it was consistent.

So, the 4850 only works with two monitors connect to it. If only one monitor is connected, I cannot play games. (note: this is only the case on the Dell, on the other computer it works regardless)

Anyone experienced this problem? What could be causing this?



It sounds to me that there might be an issue with the Catalyst software/drivers if the dual monitor setup allows your card to work. There really isn't anything about plugging in a second monitor that would change the way your card works or interacts with the board. I know you stated that you have tried different catalyst versions, but perhaps go back to some of the original ones from the 4850, and make sure that they are set up properly for the single monitor setup. This certainly wouldn't be the first driver/compatibility issue with the 4850.

On a side note, as an owner of two of these cards in a crossfire setup, under load my temp never exceeds 58 degrees, definetly make sure that you have very good cooling for this card. When I put both cards in with only one case 120mm fan they ran above 70 degrees frequently, after adding two additional 120mms they stay 45-58 degrees.
!