Problem with Sapphire X1900XT - Overeating, Artifacts & Crashes!

[cocksy_boy]

Hi All,

I am having problem with my graphics card and have been scouring the net for a couple of days trying to sort it out. Basically when gaming (Mainly COD4 as an example), after a while playing (sometimes only a few minutes), the game will start artifacting (some parts of the screen flashing, random lines and areas flashing quickly), and eventually it will crash and go to a black screen with the sound still playing or on loop. The only way to get out is to hit the reset button. Not good! Sometimes on reboot Windows will say that the crash was caused by the graphics driver, and is searching for a resolution. Obviously it never finds one!

My PC specs are:

Intel Core 2 Duo 6300 - Stock Cooler
Gigabyte GA-965P-DS3, Rev 1.0, F12 BIOS
2 x 1 GB Geil DDR2-SDRAM PC2-5300 (333 MHz)
Sapphire X1900 XT - With Accelero 2
Nothing Overclocked
IBM/HGST SATA 160Gig & Diamond Max 10 SATA 160Gig
Jeantech 450 Watt ATX PSU with 120mm fan
3 x 120mm fans in Antec P180 Case

Vista Ultimate (Updated)
All up to date motherboard drivers
Catalyst 7.12 Drivers (Previous drivers removed / cleaned with DriverCleaner.Net)

After a lot of searching, and playing with Riva Tuner, ATiTool, Speedfan, Coretemp, and CPU-Z I think I have narrowed the problem to the Voltage Regulator overheating, and/or possible a dodgy PSU.

Basically if I leave my system to its own devices when gaming, the following happens (data gathered from RivaTuner):
GPU temps gradually goes up to about 85 Deg C
The fan duty cycle only goes up to about 36% (about 1300 rpm)
Ambient temp rises to about 65-70 deg C
Voltage regulator output goes to 1.45v
Voltage regulator current hovers around 33A
Voltage regulator temp steadily climbs up to 90-odd deg C

I don’t think most of these figures are too bad, except the voltage regulator temp. So I tried setting the fan to 100% with ATiTool, and the situation is much better:
GPU temps gradually goes up to the mid 60’s C
The fan remains at 100% (just over 2000 rpm)
Ambient temp steadies at about 45 deg C
Voltage regulator output goes to 1.45v
Voltage regulator current hovers around 33A
Voltage regulator temp hovers around 70 deg C

One other thing I have noticed is that the -12V line from the PSU seems to be at approx -17V & the -5V line –seems to fluctuate between -8V & -5V. This doesn’t seem right, but I couldn’t find anything about this anywhere!

(If someone tells me how, I can post some pics of the artifacts in COD4, and the .hml RivaTuner log files if they are any use in helping diagnose the fault! Also, is there a plug in to monitor my motherboard voltage settings with RivaTuner? I think it would be very helpful to have all the data collected by the same app!!!)

So, can anyone make anything from this? If it is the voltage regulator over temping, what can I do about it – is there a decent heatsink I can buy to help reduce the temps. Or could this be fixed with a better PSU – I’m thinking of perhaps the Corsair HX Series 520W Modular PSU - ATX12V v2.2 APFC (http://www.ebuyer.com/product/114945). It should be plenty powerful enough!

Any comments & help greatly appreciated! TIA,

Cocksy.

[cocksy_boy]

Right, I received my brand new Corsair HX620W modular PSU (yeah, I went for the bigger one!!), along with a 120mm case fan (noctua nf-s12) and a new hard disk (WD Caviar sE15 500GB) installed them all, and.............

I'm still getting the artifacts and crashing poblems!!! :( I think my problems are very similar to this guys problems (http://www.sapphiretech.com/en/forums/showthread.php?t=17291)(expect mine is with the PCI-E version, and his is AGP, and his is the 1950, but which is identical to my 1900xt except for clock & mem speeds), but I can't seem to get the sign in / up to the sapphire forums to work, so can't post there to ask for help!!

Now I think the problem has got wose since I installed the new PSU, I had artifacts and a crashes whilst playing COD4 with the fan on 100%, and different artifacts and crashes in Crysis, also with the fan at 100%.

I did notice after one of the crashes, that there was artifacts on the POST screen when booting (a couple of coloured dots that flashed a couple of times), so Ipowered off and left the sstem to cool before I rebooted 10 mins later, and everything was fine in vista again.

Please can anyone help - is it bad RAM on the card - is there any way to check the RAM on the card? I'm running out of options and the card is 18 months old so I prob cant replace it under warrantee now, which is a royal pain in the back side.....!!!!

[Namslas90]

It appears as if your card is overheating. You should pull it out and clean it well, inside the fan housing. You may even want to redo the thermal paste on the GPU, or add an aftermarket cooler. Make sure you have good airflow into the case those cards get hot. Also you can use ATI TOOL (http://www.techpowerup.com/atitool/) to controll the fan speed of the card fan to increase the volume of airflow through the card fan by increasing the fan % or overall speed using the Fan controll feature.

I have a 1900XTX that ran fine (with increased fan speed) and extra case fans installed for about 21 months until the thermal paste "wore out" and had to be redone. After redoing the thermal paste the card runs cooler than it did with stock thermal paste.

:toast:

[cocksy_boy]

It appears as if your card is overheating. You should pull it out and clean it well, inside the fan housing. You may even want to redo the thermal paste on the GPU, or add an aftermarket cooler. Make sure you have good airflow into the case those cards get hot. Also you can use ATI TOOL (http://www.techpowerup.com/atitool/) to controll the fan speed of the card fan to increase the volume of airflow through the card fan by increasing the fan % or overall speed using the Fan controll feature.

I have a 1900XTX that ran fine (with increased fan speed) and extra case fans installed for about 21 months until the thermal paste "wore out" and had to be redone. After redoing the thermal paste the card runs cooler than it did with stock thermal paste.

:toast:

I have recently re-applied artic silver 5 paste to the GPU (having cleaned the GPU & heat sink properly with lint free cloth and isopropanol), and the fan and heatskin are clean from dust & build up.

Also, the RAM chips on the 1900XT are not at the same 'height' above the circuit board as the main graphics core. So on both the stock cooler, and the accellero 2 (which I have fitted), there is a small 'shim' of spongy heat conducting material to help keep the mem chips in contact with the heatsink and cool.

Now I'm pretty sure I have fitted them properly, there isn't a lot that you can really do wrong when fitting them (just peel off the 'covers' on both sides and apply to the chips), and the problems did occur before I fitted the accellero 2 too (too many two's :) ), so I'm 99% convinced I have fitted the cooler correctly.

But that doesn't mean the memory isn't getting too hot (is there any way to tell?), or isn't already heat-damaged (is there any way to test the graphics memory?). I have pretty good air flow in my case - I have the 2 standard P180 exhaust fans going, and the new noctua nf-s12 12cm 'sucker' in fan sucking air through the front grille & filter. So case temps should be OK, I hope!

I am already using ati-tool to change the fan speed to 100% before game playing, and still having the artifact & crashes.

But there aren't any other x1900xt owners with or without accellero 2 coolers complaining about this sort of fault, as far as I can see!!! Maybe it is dead / dying, but at 18 months old, and with a different cooler to stock, I may have problems returning the card now! I do have the original cooler around somewhere to fix back on, but I dont think it will fool anyone!!!

Any thoughts?

[Namslas90]

Well, aparently you have done almost everything you can. You can "Scan for Artifacts" with ATI TOOL, but we allready know you have them. There is a chance the GPU ram is allready damaged, it sure sounds like it. Check this thread it shows how to add Ramsinks to the Power Mosfets, and has proven to lower temps overall.

http://forums.techpowerup.com/showthread.php?t=8385&page=1

Edit; also check the in-case airflow make sure the card fan intake is not blocked by anything and that another fan is not pushing hot air into or at the card. I had that problem once, where the Chipset fan was throwing the heat off the CPU cooler directly into the GPU fan intake. Reversing the direction of the chipset fan solved the problem.

:toast:

[trog100]

can u slow it done below stock speeds.. ???

i think its dieing.. but slowing it down some might prolong its life..

trog

[DOM]

re-applied artic silver 5 paste to the GPU

So did it do this before you put it on the 1st time ?

And why didnt you use MX-1 or 2 or arctic silver ceramique which are nonconductive ?

[cocksy_boy]

So did it do this before you put it on the 1st time ?

And why didnt you use MX-1 or 2 or arctic silver ceramique which are nonconductive ?

Actually, you are right - the first time I put the accellero 2 on, i just used the paste that was on the heatsink... and yes it did artifact and crash!

As for why i used AS5 - it was the best that I had to hand!! I haven't checked up on the best paste for a while, but didn't think there was a major problem using the AS5 - am i wrong there?? I am unaware that the fact that it is conductive is a major problem.... is it?

I may order some non conductive paste if there is a chance thta it is causing the faults.
Any thoughts?

[DOM]

cuz when did it start crashing ? was it before you did anything to the card ?

did you put as5 on the stock hs ?

did you put as5 anywhere else on the card besides the gpu ?

well its better for gpu's non conductive paste cuz is it gets on the card nothing well happen ;)

[cocksy_boy]

It started crashing befire i added the AS5 or changed the cooler. They were ideas I had to try and help!
I haven't put AS5 on the stock cooler.
Only put AS5 on the gpu, and then only a thin layer of it.
Why is a conductive paste any worse on a GPU / graphics card than on a CPU? I'm not being rude, I just didn't know there was an issue there!! :)

[erocker]

Your gpu is a bare chip, while most modern cpu's have an integrated heat spreader. In my opinion the AS5 should be fine as long as it's on the gpu only and not on any of the surrounding resistors or anything. I would try re-applying the paste anyways. Try using MX-2, it isn't conductive and there is no break-in time. The results are immediate. Good luck.

[cocksy_boy]

Hmm, I'll have a look at getting some ceramique or MX-2 then. I'm pretty sure I kept the paste just on the GPU, but I'll have a look when I get a chance to open up my case next!

[DOM]

also have youy tryed running atitool scan for artifact in 2d mode ?

[mrw1986]

Don't know if anyone noticed, but his thread title says "Overeating" lol. Put it on a diet.

[cocksy_boy]

also have youy tryed running atitool scan for artifact in 2d mode ?

how do i do that, and what will that help me find out? Cheers!

[cocksy_boy]

Don't know if anyone noticed, but his thread title says "Overeating" lol. Put it on a diet.

D'Oh!!!

[Kammster]

i remember many threads in this forum about putting heatsinks on the voltage regulators

[jaxxxon]

It's not overheating 85c is fine for this card, and mid 60s with fan 100% is great whilst gaming. Nothing to worry about there, something else is wrong.

[cocksy_boy]

It's not overheating 85c is fine for this card, and mid 60s with fan 100% is great whilst gaming. Nothing to worry about there, something else is wrong.

Thats what I think.

Right, I have just taken the card out and thoroughly cleaned the face of the heatsink and GPU. There was a little more AS5 than I would have liked, but not loads floating around the place.

There was no dust or stuff blocking up the fan or heatsink in any way at all. I put a small amount of AS5 on the GPU and replaced the heatsink. I left it for 20 mins, then took it off to check I hadn't put on too much and take a photo. I think the amount is fine - a thin layer and you can even read ATI through the AS5 on the GPU and a 'reflected' image in the AS5 on the heatsink!!!

(if someone can tell me how to post or link to an image, I'll put it up here)

So I put it all back together and the temps have dropped a bit (core went to mid 70's, atmoshperic to low 60's, voltage reg to mid 70's too while gaming COD4), but still getting artifacting after a couple of minutes gaming. I didn't play for too long before I quit back to windows to post this!!

What the hell am I doing wrong?!!! Is my Sapphire dead, cos I really dont want to have to replace it - its only 18 months old, and should still be able to keep up with games for a while yet!!

Can Anyone help me?!!!! Please?!!!!!

[DOM]

how do i do that, and what will that help me find out? Cheers!

open atitool > settings > X1000 Overclocking > uncheck "If the card has 2D/3D..." close atitool open it again and it should be in 2D

[cocksy_boy]

open atitool > settings > X1000 Overclocking > uncheck "If the card has 2D/3D..." close atitool open it again and it should be in 2D

OK, I ran the test for 45 mins with no errors at all in '2D mode' - that is basically just slower clock speeds, right?

[DOM]

OK, I ran the test for 45 mins with no errors at all in '2D mode' - that is basically just slower clock speeds, right?

yes and lower volts on the VGPU, VDDCI

cuz do you always have atitool running ? and did you make profiles ? so it can switch to 2D/3D speed and volts ?

im thinking no, like you did to make it change to 2D, where it shows you the core and mem at the top it has Profiles click New and I would name it 2D if it 2D speed, then click settings and go to Votage control and click save at the bottom and put it on 2D and to get the 3D speeds go to the X1000 Overclocking and click the "If card" and close and open should load with 3D speeds and volts click New 3D then do the same for the volts

then Settings > Startup look at pic
http://img.techpowerup.org/080110/untitled3.png

then go to 3D Detection look at pic
http://img.techpowerup.org/080110/untitled4.png

that should do it

[cocksy_boy]

I'm gonna have a look at those ATi Tool options later tonight, but in the mean time I'm thinking I may have to buy a really cheap card to test to make sure nothing else in the system is wrong. Can anyone recommend a very very cheap graphics card to buy in the UK & install to test my system, or while mine is away being RMA'd?

It needs to be PCI-E, DVI (dual if possible), ATI (so I have less issues removing and reinstalling the drivers when I chop & change between the cards) and if possible a DX10 card (but not sure if they exist really cheap yet!).

Any ideas? Cheers!

[DOM]

well heres a review on a Diamond Viper X1900XT should be the same as your card and has the volts if you want to know them http://www.techpowerup.com/reviews/Diamond/X1900XT/8.html

and whats cheap ? max

[cocksy_boy]

Cheap..... hmm, well bearing in mind that UK prices are massive compared to the USA, I'm talking about under £50. thats about $100. But it wont buy as much in the UK as it will in the states!

[cocksy_boy]

OK, I think I'm going to RMA the card, so I need to put the stcok cooler back on.

So, does anyone have any close up high detail pictures of the back of their X1900XT 512MB cards so I can see exactly what screws and washers were attached to the card?!!!

I think the Sapphire is the same as the reference ATi card, but I'm not sure of the other manufacturers!
Cheers!

[cocksy_boy]

OK, I have just had a new 'semi-crash'!

I was gaming with Oblivion trying some stuff out, when it semi-crashed: the screen went black / to stand by temporarily, then came back after a couple of seconds. The game continued to work, but I quit to windows as RivaTuner was running and I wanted to see what had happened.
Windows came up with a window saying the display driver had crashed and that it had recovered from the error. I looked at the RT monitor file and took a snapshot:
http://www.geocities.com/anthonycocks/RTOblivionCrash.JPG
and here is the link (http://www.geocities.com/anthonycocks/OblivionSemiCrash.zip) to the zipped RivaTuner file if anyone wants to look at it!

Basically it looks like when it crashed, the fan went off and the temp shot up (obviously). Before the crash, there was a small amount of artifacting, but you can see that the temp of the card wasn't actually very high at all - low 70's. So maybe its nott he temp of the card causing the crashes?

I'm confused!

So can this help anyone diagnose the problem at all?!!!

[Namslas90]

Well with the Vreg temp and Core temp stable at time of crash, and the Vreg output and Fan dropping so much, I would tend to think its your PSU.

I could not even get my rig to boot with an Antec Neo 550 watt PSU and a 1900XTX.

Could just be the PSU showing signs of dieing.

:toast:

[cocksy_boy]

The PSU is a brand new Corsair HX 620W, so should be powerful enough. Also, the Vreg doesn't go to 0, it goes to 1.2 V, which is the standard 2D voltage when in goes to the slower clock speed.

But, I dont know whether it is the drop in voltage causing the clocks to drop (which could be an underpowered PSU), or the drop in clocks causing the voltage to drop (as in a differnet fault).

Any ideas how to test this?

[Namslas90]

Yeah thats a good PSU. No overheating so the card must just be going bad. The drop down to 2d clocks could be a driver issue, maybe cleaning drivers and re-installing after updating Motherboard drivers might help.

:toast:

[cocksy_boy]

Yeah thats a good PSU. No overheating so the card must just be going bad. The drop down to 2d clocks could be a driver issue, maybe cleaning drivers and re-installing after updating Motherboard drivers might help.

:toast:

I have the most up to date motherboard drivers, and cleaned the catalyst drivers when I installed 7.12. I have just recently got a new HDD, so I may try again with a complete re-install to see if that does anything. I may also have a quick dable with XP Pro again to see if that has a better / worse effect!

[DOM]

:wtf: the core went up to 110C after the fan went down

so it has to the card or the fan on the Accelero 2 well since your going to put the stock back on try setting the fan speed on a fixed % in atitool in 3D see if it goes down

[cocksy_boy]

Yeah, I'll try that after I have replaced the original heatsink - not going to happen straight away though!

[Nick89]

OK heres my take on his cards issue, I read through most of this thread and my conclusion for all these crashs is that his card is overheating(DUH), the reason for this is that he's using a stock ARTIC Accelero X2 cooler. I used this cooler on my 1900XT before it died from overheating last summer. and I am useing it AGAIN NOW on my 1950XT but this time I took that crappy fan off and tied a 120mm fan onto the cooling fins of the Accelero X2 and the temps went from 51C idle to 38 idle and from 78C load to 55C load with the 120mm fan on it.

My bottom line is that ARTIC Accelero X2's suck and should never be used on a graphics card unless its just sitting there in the corner of your room and your desperate to get your rig running again because your water pump died:banghead: in your water cooling system :mad:and your GPU waterblock melted...:shadedshu <- That BTW just happened to me and is the reason I am using my Accelero again, I decided to take the stock fan and shroud off my Accelero X2 and stick a 120mm fan on it(which came from my radiator :cry:) so that it would cool better and it does!

Personal issues aside I recommend you get a better card cooler ASAP otherwise you'll be buying a new graphics card.

ALSO to the guy(jaxxxon) that said "85c is fine for this card" NO NO NO NO NO.... Yea it will run at that temp but its far from ok for the card and will shortan the lifespan of the card. :slap:

[cocksy_boy]

Hi Nick - cheers for yout input!!! I'm going to put the stock cooler back on and see what effect it has. I'll let yuou konw how it goes, but it wont be straight away, as I'm quite busy with work at the moment! Thanks again!!!

[cocksy_boy]

Well, I replaced the Accellero 2 with the orignial heatsink with some AS5, and the PC still crashes during gaming. If I set the fan to 100%, some games run OK (like Oblivion), but I still get a bit of artifacting and 'stretching to infinity' in Crysis and COD4. In crysis with the fan at 100%, i get the following:

Core temp is 60°C - not very hot at all, but still artifacting
Ambient temp 51°C
Voltage reg temp, about 60°C

So it seems that the temps aren't very hot, but the problems are still there.

Anyone have any ideas how to fix it?! If not, I'm gonna try and RMA it. Not ideal!

[DOM]

Rma I guess those temps dont look bad at all :ohwell:

[RottnJP]

...So, How did this story end??

I just built a new rig with A64x25600+ and X1900XTX that a friend gave me, and I have the exact same symptoms when running any "real" game (battlefield series). Neverwinter nights gives me artifacts but doesn't crash. BF2142 crashes every time, although the time before BSOD seems variable. It seems like the video signal just croaks- The monitor LED goes yellow, but I can still hear the sound. Freaky...

I've got a ton of case flow- Coolermaster 690 case with 4 140mm fans (2 intake, 2 exhaust) and 2 120's (1/1) plus the 120mm scythe on the TR 120 extreme. Stock cooler on the GPU, but remounted with AS5. Trying to figure out if the card is crap, or I should try a new cooler. From all the above, I'm thinking new card, but I'd love to hear how this story ended...

[cocksy_boy]

I'm in the process of RMAing the card. Unfortunately I have been really busy at work, and also I can't afford a decent replacement card right now. So I am deciding whether to save up for a decent DX10 card and then sell the replacement 1900XT (assuming they do replace it), or to buy a really cheap card and hope that they get the replacement back to me quickly. Either way, I can't take the card out of my PC striaght away, which is a pain in the proverbial!!!

[VroomBang]

Hi All,

I am having problem with my graphics card and have been scouring the net for a couple of days trying to sort it out. Basically when gaming (Mainly COD4 as an example), after a while playing (sometimes only a few minutes), the game will start artifacting (some parts of the screen flashing, random lines and areas flashing quickly), and eventually it will crash and go to a black screen with the sound still playing or on loop. The only way to get out is to hit the reset button. Not good! Sometimes on reboot Windows will say that the crash was caused by the graphics driver, and is searching for a resolution. Obviously it never finds one!

My PC specs are:

Intel Core 2 Duo 6300 - Stock Cooler
Gigabyte GA-965P-DS3, Rev 1.0, F12 BIOS
2 x 1 GB Geil DDR2-SDRAM PC2-5300 (333 MHz)
Sapphire X1900 XT - With Accelero 2
Nothing Overclocked
IBM/HGST SATA 160Gig & Diamond Max 10 SATA 160Gig
Jeantech 450 Watt ATX PSU with 120mm fan
3 x 120mm fans in Antec P180 Case

Vista Ultimate (Updated)
All up to date motherboard drivers
Catalyst 7.12 Drivers (Previous drivers removed / cleaned with DriverCleaner.Net)

After a lot of searching, and playing with Riva Tuner, ATiTool, Speedfan, Coretemp, and CPU-Z I think I have narrowed the problem to the Voltage Regulator overheating, and/or possible a dodgy PSU.

Basically if I leave my system to its own devices when gaming, the following happens (data gathered from RivaTuner):
GPU temps gradually goes up to about 85 Deg C
The fan duty cycle only goes up to about 36% (about 1300 rpm)
Ambient temp rises to about 65-70 deg C
Voltage regulator output goes to 1.45v
Voltage regulator current hovers around 33A
Voltage regulator temp steadily climbs up to 90-odd deg C

I don’t think most of these figures are too bad, except the voltage regulator temp. So I tried setting the fan to 100% with ATiTool, and the situation is much better:
GPU temps gradually goes up to the mid 60’s C
The fan remains at 100% (just over 2000 rpm)
Ambient temp steadies at about 45 deg C
Voltage regulator output goes to 1.45v
Voltage regulator current hovers around 33A
Voltage regulator temp hovers around 70 deg C

One other thing I have noticed is that the -12V line from the PSU seems to be at approx -17V & the -5V line –seems to fluctuate between -8V & -5V. This doesn’t seem right, but I couldn’t find anything about this anywhere!

(If someone tells me how, I can post some pics of the artifacts in COD4, and the .hml RivaTuner log files if they are any use in helping diagnose the fault! Also, is there a plug in to monitor my motherboard voltage settings with RivaTuner? I think it would be very helpful to have all the data collected by the same app!!!)

So, can anyone make anything from this? If it is the voltage regulator over temping, what can I do about it – is there a decent heatsink I can buy to help reduce the temps. Or could this be fixed with a better PSU – I’m thinking of perhaps the Corsair HX Series 520W Modular PSU - ATX12V v2.2 APFC (http://www.ebuyer.com/product/114945). It should be plenty powerful enough!

Any comments & help greatly appreciated! TIA,

Cocksy.


My gut feeling is your psu is not good enough. What's Sapphire X1900 XT min requirement in terms of PSU and does the Jeantech meet that?

I've got a jeantech psu myself, and I suspect it doesn't give enough Amps to my power hungry X1950 pro. Problem is jeantech don't specify the continuous Amps on the 12v rails, but only peak values which is useless.

[cocksy_boy]

Hi VroomBang,

I have upgraded my PSU since the first post to a corsair 620W unit, and the fault still persists. Some people think damage may have been cuased by the under powered psu initially, and there is no going back now! Am still planning on RMA-ing it when i have a chance!

[VroomBang]

cocksy boy,

My x1950pro overheats a lot now (bought it 5.5 months ago) and shop is ok for RMA, thank God. I'm keeping the psu as I'm changing to a less power hungry graphics card (7950GT), which according to the manual only needs a 350w psu, and is fanless (heatpipes), which is nice.

[cocksy_boy]

Probably a good idea - I'm trying to save up got an 8800GT, which isn't so power hungry. and quite good too!!

[Megasty]

Great choice. My bro kept on having overheating issues with his x1900xtx after about 8 months. The gpu reved up to 95C, the fan would max out 20 sec later then freeze up. He has a ultra x-pro 750. It did the same with my psu too. It did it so many times I got mad, ripped it & threw it out the window in the snow & it steamed the snow to the ground :wtf:

I gave him a x2 to play with...I think he's happy now :pimp: