**UPDATE** CARD IS DEAD Problem with Sapphire hd 4870 1GB GPU

o3srt4me

Member
Feb 5, 2009
36
0
0
Hello all,

*** So to all of you who have been helping me I really appreciate it and your are fairly familiar with my problem, well as of last night I stock clocked the GPU and the Processor, did some TF2 max everything for about 3 hours and called it a night, this afternoon I downloaded the Furmark and ran the Stability test at 1280x1024 with no AA and the card shot up past 93c in seconds, I immediately stopped the test. Thinking nothing of it I started call of duty 5 to play a few matches and within 45 seconds the screen went to gray and I began to smell a burning from my case. I assumed the worst that my 500w powersupply had gave way and fried everything...., tried rebooting and got nothing, not even a DOS screen. I immediately pinpointed that it was the card that was smelling so I pulled it out and put in my old 8800gt and the computer started up fine....

I must say that though my powersupply is the smallest recommended, the key thing to me at least is that it is recommended by ATI. I obviously can return the card for a new sapphire, but im wondering.....was it a faulty card from the beginning, and thats why it inevitably died 3 days later, or did my setup destroy the card.... I guess I will begin contacting newegg about where to go from here :(***


New to the forums, also new to PC building. I just recently built my gaming computer and ran into my first real problem after purchasing a Sapphire HD 4970 1GB. My components are as follows

Antec 500w Earthwatt powersupply
E8500 OC to 3.6Ghz
Arctic cooler CPU fan
7200rpm 500GB HDD
4gb ddr ram
Sapphire 4870 1gb GPU OC (790/1000)mhz
Asus P5Q Pro 775 LGA

After installing I downloaded driver set 9.1 from ATI and ran the auto tune. The ATI auto set me at 820 for a gpu clock and 1000 for my memory clock. During this test a yellow icon with a exclamation mark in the center flashed twice telling me that the GPU had crashed and recovered successfully. After the autotune completed I attempted playing Rainbow 6 vegas 2, I was able to play for about 5 seconds before the entire comptuer would freeze. The same occurance happened in Call of duty 4. I was rather confused and did some back research here on the forums and found that people were having similar problems with the Sapphire 4870 512mb cards. I wasnt sure if anyone knows if the same problem that was fixed by flashing the Vbios is also a problem affecting there new 1GB models?? My other guess is that my powersupply is being maxxed out and the card is not getting sufficient power, or possibly I dont have sufficient power to OC the 4870???
The highest temp I noted during the Autotunes' load test was 74c, anyone seen a similar problem with their 1GB sapphire??? Any help will be much appreciated!!

Thanks guys

P.S. this is my first thread. From doing research I found the Anandtech forums to be by far the most helpful!!!

 

o3srt4me

Member
Feb 5, 2009
36
0
0
I did that and it definately was a main part of the issue. I pulled it back and the games ran no artifacts, but I noticed after gaming that at some point the GPU crash warning was triggered, so there must still be too much OC...

So then is it somewhat safe to say if I had a bigger power supply I could potentially run a higher overclock????
 

nOOky

Diamond Member
Aug 17, 2004
3,245
2,320
136
Then obviously I would return the card to stock clocks, and if the problem goes away you simply have not-very-overclockable-card. That does seem like a low end power supply for your set up, it probably isn't the issue though.
 

o3srt4me

Member
Feb 5, 2009
36
0
0
Thanks, yea I began to figure the same.....I ran crysis and it crashed the game, after returning it to normal everything seemed to run ok. Hmmm kind of dissapointing, is it worth asking for a new 4870, because I do know alot of people who had no problems overclocking theirs.

And another question, how much difference does overclocking the 4870s actually make? any experts out there????
 

FerrelGeek

Diamond Member
Jan 22, 2009
4,669
266
126
One other question, what is you fan set to? If your overclocking and want to do heavy-duty gaming, I'd make sure that your fan is running at least at 35%.
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
Originally posted by: o3srt4me
I did that and it definately was a main part of the issue. I pulled it back and the games ran no artifacts, but I noticed after gaming that at some point the GPU crash warning was triggered, so there must still be too much OC...

So then is it somewhat safe to say if I had a bigger power supply I could potentially run a higher overclock????

you first mistake is to set any overclock before testing your new card and your new system at stock everything
- i hope you did not do the same thing with your CPU :p

as to a larger PS, it depends. If your current once supplies sufficient stable voltage at your GPU's overclock, then a bigger one will make no difference to it
- doesn't AMD recommend a more powerful PS?

if you are going to OC, monitor your temps closely and adjust your fan speeds if necessary
rose.gif
 

o3srt4me

Member
Feb 5, 2009
36
0
0
Lol I definately played games with the card stock clocked and they ran perfectly stable, but I was getting frame drop as expected at 4xaa and 1920x res so I figured I would start pushing the card. So my brilliant friend (mind you I am new to pc building lol) says hey use that auto tune program it will dial in the setting for your system. Well it pushed the OC waaay up yo about an 820 core and 1000 for Mem (but from my understanding the mem wasnt that high as compared to other users)


Here is my question and Im sure its dumb, where can I find a safe program that will allow me to monitor my gpu temps while in game....



btw the fan speed is set to 60, it was set to 50 right out of the box???

it idles around 45c and the highest it went in during the auto tune was about 74c
 

nOOky

Diamond Member
Aug 17, 2004
3,245
2,320
136
Try oc'ing the memory first, you'll probably see more headroom with that than the core clock. I don't trust any settings unless they allow me to play for hours stably. Most stressing or automatic overclock tools are optimistic.
Your temps sound fairly normal for that card, you could remove the side of your case and see if they drop at all. Methinks case airflow has quite a bit to do with the video card temps.

 

o3srt4me

Member
Feb 5, 2009
36
0
0
Haha ok well I am running crysis in a windowed mode right now and after about 10 mins of gaming the card is already at 76c lol, and sitting there. Just how much heat can these things take!


my settings during this were

Core clock: 760mhz

Mem clock: 940mhz

Fan speed: 60%

temps- 73-76c


Also does running the fan at a higher percentage cause the card to deteriorate or wear out faster as the warning said in the ATI overdrive????
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
monitoring temps in game?


if you have a 2nd LCD, you can use that
- or alt+tab out of the most intense firefights to see your peak temps

you can also log your temps, i believe
rose.gif


well, if you run your fan 100% of the time, clearly it will wear out faster
- it depends


you can see temps into the 80s in a warm room without panicking

-you should see the FURNACE that is my case with CrossfireX-3
:sun:

awesome for Winter
 

o3srt4me

Member
Feb 5, 2009
36
0
0
Lol woah! thats a pretty insane GPU setup!

Hmmm well In that case I suppose im safe where Im at!
 

o3srt4me

Member
Feb 5, 2009
36
0
0
I just tried something interesting, I turned off the manual fan control and let the driver controlled fan map run....and it seems to be working quite well...The highest fan speed I noted was about 67% but at the same time it backed off when necessary, this to me seems more effective at managing temps than just a constant speed of 60%

I know some people have had issues with the driver set fan speeds, but mine still seems to idle at 45c with a fan speed of only 10%


any opinions??
 

evolucion8

Platinum Member
Jun 17, 2005
2,867
3
81
Which cooler does your Sapphire card have? ATi red fan reference? Toxic looking cooler? Toxic VaporX cooler?
 

o3srt4me

Member
Feb 5, 2009
36
0
0
Originally posted by: evolucion8
Which cooler does your Sapphire card have? ATi red fan reference? Toxic looking cooler? Toxic VaporX cooler?



http://www.newegg.com/Product/...x?Item=N82E16814102801
its the toxic looking cooler, just the black fan in the middle....

Lol Im assuming the red style fan is better and toxic is best seeing as how all people with the red style fan seem to run their 4870s at 35% OCed and here I am up at 60% trying to keep mine cool...



Honestly I kinda wanna return it and get the power color for 9 dollars more with the core clock set at 800mhz stock....:(
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
there is really not so much of a performance difference
- i prefer the stock cooler that exhausts the air our of my case .. the other one heats up my X2, q9550S and roasts the inside of my case
 

Cookie Monster

Diamond Member
May 7, 2005
5,161
32
86
The powercolor might make 1 fps difference to your current card.

Id stay away from autotune, and rather manually clock the core/mem and then test its stability using furmark. Keep in mind that the GPU will throttle itself when the temperature is high, so try upping the fan speed which will typically result in lower temps but higher noise. This wont really wear out the card but maybe the fan itself will deteriorate since your increasing its rpm. (This however should not concern you as fans last for a very long time)
 

o3srt4me

Member
Feb 5, 2009
36
0
0
Originally posted by: Cookie Monster
The powercolor might make 1 fps difference to your current card.

Id stay away from autotune, and rather manually clock the core/mem and then test its stability using furmark. Keep in mind that the GPU will throttle itself when the temperature is high, so try upping the fan speed which will typically result in lower temps but higher noise. This wont really wear out the card but maybe the fan itself will deteriorate since your increasing its rpm. (This however should not concern you as fans last for a very long time)

Hmm well I am now questioning whether or not my card is busted or if there is another problem plaguing my PC....Im on a 1920x1080 res and I decided to back up and remove the AA from Cod 4 and 5 and see the differences in FPS from 4xaa to 2xaa to none. Assuming I would be fine at 2xaa (since I never bothered trying to play in anything under 4 since I bought the card 3 days ago), and to my surprise the same frame drops occured....sometimes as low as 45 during heavy shootouts....So I took the AA Completely off and it still couldnt hold a solid 60 through big bombs detonating next to me! I was really confused... I would assume that my GPU is plenty sufficient to not be the culprate..and since my old 8800gt is dated, the gameplay was no better before so I cant for sure blame it on the card....


if this helps my latest 3d mark06 was 14,800-with the card at 780mhz core and 940mhz mem

I accidentally clicked off the subscores but the 3 scores were something like

SM 1-6xxx

SM 2- 7xxx

CPU- 32xx

anything seem out of the norm here, lol I really just wanna game with good frames one way or another!!!!
 

evolucion8

Platinum Member
Jun 17, 2005
2,867
3
81
That's the same card that I own. I suspect that it has something to do with the Power Supply, it's too low and risky for the HD 4870, 3DMark06 is quite CPU bottlenecked.
 

o3srt4me

Member
Feb 5, 2009
36
0
0
Originally posted by: evolucion8
That's the same card that I own. I suspect that it has something to do with the Power Supply, it's too low and risky for the HD 4870, 3DMark06 is quite CPU bottlenecked.

Well I am an idiot, you are totally right and as for the CPU as well. The reason is that the power supply is WAY too low. I had my oc at 3.66 for the processor but never bothered to run the Prime95 for more than 10 minutes to check temps(which were fine, but I got an error with hardware notice after about 10 minutes!!)....well turns out that my setup has been unstable as all hell with the new card, which is probably why my OC is showing no gains/crashing....I took the card back to totally stock clock and the CPU to only a .15GHZ overclock and finally got prime95 to run for about an hour and ill do more extensive tests tomorrow.....obviously I have no idea what im doing... lol I need to research. So everything is gunna stay stock clocked until next month when I can afford a new power supply....other than that everything seems to be running fine now...

My question is, since I was running such an unstable setup for a few weeks, is it possible I partially damaged any of my components???? is that possible or when things fry do they go from all to nothing????
 

evolucion8

Platinum Member
Jun 17, 2005
2,867
3
81
I don't think so, computer hardware usually when it's broken, it stop working inmediately, specially the motherboard and CPU parts, videocards are bit harder to find out. Put a new Power Supply and do stability test at stock speeds to find out, like running Prime95 and Furmark at the same time, if it does fine, then overclock like you did last time and do the same again.
 

o3srt4me

Member
Feb 5, 2009
36
0
0
Originally posted by: evolucion8
I don't think so, computer hardware usually when it's broken, it stop working inmediately, specially the motherboard and CPU parts, videocards are bit harder to find out. Put a new Power Supply and do stability test at stock speeds to find out, like running Prime95 and Furmark at the same time, if it does fine, then overclock like you did last time and do the same again.

Ok so I downloaded furmark and ran it at 1280x1024 and within nearly 30 seconds the card was at 92c and rising...I stopped the test immediately assuming that this was going to destroy my graphics card....

Not sure what that means, I ran the stability test.....

Ive played several games for long hours and never seen temps like that even at 99% efficiency...what does this mean????


 

evolucion8

Platinum Member
Jun 17, 2005
2,867
3
81
Wow, that's a bit too high, specially that I own the same card and it never goes beyond 70C in full load with Furmark. The GPU can reach between 105C to 125C before it burns out, but still is not an ideal condition for longetivity. A nice stability test is running Prime95 and Furmark at the same time. Furmark is one of the few D3D applications which can max out all the GPU's execution resources, hence, increased temperatures.

I don't know exactly how the GPU utilization works because all games uses the GPU differently, yes, there are a lot of games which can use 99% of the GPU and never goes that high in temperatures, but probably is because that game is maximizing the usage of the GPU's texture units, or shader units, but not both at the same time etc, the ATi's Screenspace demo created for the ATi X1900 series also maxes the GPU usage and can push the card very hard, I used it and my GPU usage never dipped below 99.9% usage with 8x Anti Aliasing. You may want to download it and use it as an alternative test once you swap your PowerSupply.
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
FurMark has been known to burn out ATi GPUs

in fact, ATi wrote it into their driver - since cat 8.8 - to downclock their GPUs when it encounters Furmark.exe
- it acts somewhat like a driver bug that exploits their architecture to overheat it

i would RMA ... Sapphire is no better or worse than anything else .. except their warranty SUCKS
--you have to pay $15 EXTRA for a RMA "application"

VisionTek offers a lifetime warranty