*** The Official X1950 Pro Thread ***

[mart981]

The 6.12 drivers didn't have any effect unfortunately.

The system has 2 IDE 80gig HDD's and a CD/DVD burner plus 2 120mm fans.

Could these symptoms really be due to power though? 2 people have mentioned RAM, one saying he get's similar problems when o/cing RAM too much and the other about temp's.

I don't really want to waste any more money on my old system so if I do spend any more money (instead of just selling the 1950 Pro and sticking my old 6800 back in) I want to be almost positive it will fix the problem.

 

[Dutch777]

I think some of these HIS cards are borderline on the factory overclock, try dropping the memory speed by 20 mhz and see how it is.

---------------------------

 

[Vogon]

Have to say that env temp is toasty, so overheating could be the cause, try case side off and big room fan blowing cold air in as others suggested.
I would also "underclock" it to the stock settings of 580/700(1400) and see if that helps.
Is it only getting the anomolies doing that test, or while playing games and/or surfing etc too ?

 

[honestjohn]

The 6.12 drivers didn't have any effect unfortunately.

The system has 2 IDE 80gig HDD's and a CD/DVD burner plus 2 120mm fans.

Could these symptoms really be due to power though? 2 people have mentioned RAM, one saying he get's similar problems when o/cing RAM too much and the other about temp's.

I don't really want to waste any more money on my old system so if I do spend any more money (instead of just selling the 1950 Pro and sticking my old 6800 back in) I want to be almost positive it will fix the problem.

I hear you, if you don't have a friend with one lying around, you could always get one from an e-tailer with a very good return policy. I believe you can return almost anything to Amazon.com.

 

[Dutch777]

Memory overclocked to far on my Sapphire:

 

[Pobman]

Nice card

Installed my Sapphire X1950PRO AGP in my long in the tooth Asus P4P800SE MB and it just worked, ie booted windows inserted CD and installed drivers etc.

Running with a PIV 3.2 (Prescott) gets 7449 in 3Dmark2005 at stock everything, will have an go at OCing when I put it all in my Kandalf case (air cooling the PIV at the moment rather than the usual watercool hence no overclock).

PSU is an old 350W Enermax unit.
Running 6.12 drivers as thats what was on the CD.

 

[mart981]

Is it only getting the anomolies doing that test, or while playing games and/or surfing etc too ?

They show up in that ATI Tray Tools artifact scan test, 3dMark 05 + 06 also Vanguard and Company of Heroes.

but not straight away, after about 60 secs or so in ATI TT, the second gfx test in 3dMark 05 and after a few mins playing games. So it seems to point towards something overheating but it doesn't seem to matter whether the side panel is off and both fans turned up to max or not.

Thanks for all the help btw.

I can underclock the GPU to 580.50 but when I try to underclock the memory I get black screen in low level mode. When I try to underclock in driver lvl mode (as recommended in the ATi 19x0 overclocking thread) I click apply and the GPU and MEM settings go back to defaults.

edit: some success at last, it managed to run a cycle of 3dMark05 when it had the gpu underclocked to 580.50 with the mem still at 742.50 - 6990 points (about 2000 more than my old o/ced nvidia 6800 card)

The environment temps got to 50oC and the cards temp didn't go above 50oC.

So I think I need to do more testing to determine whether the temp is the significant factor - ie the 2 or 3 degree difference when the side panel is on or off or if it's due to underclocking the GPU.

I hope it is down to the temp otherwise surely that points to the card itself?

The one OcUK returned to me saying it didn't have a fault

 

[aircooled]

I have a idea if you got the know how and the balls to try this,if done right there should be no problems. Flash the bios ,try using the
HIS X1950PRO512AGP NON overclocked bios if you can find it! I really do not want say to much, this is only for someone who knows what their doing and will be able to return it back to stock if all fails. Or maybe you can go into the bios and tweak it alittle. be sure to make back-ups and I found that a floppy works best, do not use a flash drive!
Also try these:
Sapphire X1950PRO512AGP bios
DIAMOND X1950PRO512AGP bios
You get the idea
Good luck!

 

[mart981]

Well that's not really the point, I paid for an OC HIS card. If I wanted one at stock speeds I could've paid £20 less and got the Sapphire one. I wish I had done now.

 

[honestjohn]

Well that's not really the point, I paid for an OC HIS card. If I wanted one at stock speeds I could've paid £20 less and got the Sapphire one. I wish I had done now.

I agree with you mart981, which is why I RMA'd mine. I paid for those clocks and it was killing me having to run the card at stock. Initially, mine worked OK with the HIS default clocks as long as I had the fan set to always on, but after a couple of days I had to drop the card to stock speeds. It also didn't matter with mine either if I ran the system with the case cover on or off. My case temps weren't bad though, usually between 38c-40c.

Did Overclockers just send the card back to you, or did they give you the option to return it with their restock fee of 10 pounds + vat I believe? I read in the returns section here, they are suppose to ask you what you want to do first. Thanks.

 

[spookywillow]

mart981
if your clocks are defaulting when you hit apply then you need to use increments of 8mhz or more.

 

[Vogon]

You still have 50 enviroment with the case side off and a household fan blowing in ?
Either you house is like the tropics, or somethings not right
Your mobo system reading should read pretty much the same as enviroment give or take a few (as most sensors are not that accurate), if its not something is wrong with the card sensor, which may indicate other issues with it.

The reason I mentioned downclocking is to determine if it is because its clocked to high, if that cures it you can mention this to OC and get refunded/replaced as it is not doing as advertised.

But that enviroment temp just seems rather steep for me, so that would be the first thing I would look into

 

[mart981]

Well I just ran 3dmark again with the clock speed back at what it's supposed to be (620) and the artifacts appear again. GPU and environment temps were not higher than when it completed 3dmark 05 cleanly when it was underclocked to 580.

Honestjohn, no I didn't get any option, when I first sent a request for tech support including detials of my sytem and describing the problem all I got back was an RMA number no suggestions or tech support - obviously I assumed that meant they thought it was a fualt with the card too, so i sent it back. The next time I heard from them it was an email saying that the RMA item had been despatched on it's way back to me.

 

[Vogon]

What you posted looks like memory artifacts to me, not GPU.
You will have to downclock the memory, 702 is my default so it shouldn't black screen if you put it there using ATT ?

*Edit* just noticed you said it appeared fine with 580 GPU, so it could be that, but I would be tempted to test the memory with the GPU @ both speeds.
Also still concerned as to why you are in an oven

 

[honestjohn]

Honestjohn, no I didn't get any option, when I first sent a request for tech support including detials of my sytem and describing the problem all I got back was an RMA number no suggestions or tech support - obviously I assumed that meant they thought it was a fualt with the card too, so i sent it back. The next time I heard from them it was an email saying that the RMA item had been despatched on it's way back to me.

Thanks mart981, looks like I misread the return/RMA policy where it says " You will be notified by e-mail before any non-faulty goods are returned to you." I took this to mean they would notify you, so that you could have the option of just retruning it with restock fee. Guess all it is, is a head-up.

By the way, the card I RMA'd, could run OK at 600.75 core / 742 mem. That was suggested as a work-a-round by another user here. So you may be able to bump up the core a little. I wasn't getting artifacts however. My card would just shut down and I would lose signal to the monitor.

 

[mart981]

Here's temps resting and when I aborted 3dmark05 after artifacts appear. This was with the case side off and windows in the room open and it's going to snow tonight, so it was pretty damn cold

Resting temps

Corruption temps

So the only thing that is remotely warm is my cpu, and at 48oC that's not what I would describe as hot.

ATT is showing GPU temp as 42 and Env at 44

Just ran 3dMark05 again but this time with the GPU underclocked to 580. 3dmark completed

When 3dmark went back to desktop after each stage I could see the temps, after the 3rd graphics test ATT was showing the GPU temp as 45 and the Env temp at 46, speedfan was showing my CPU at 49.

So apparently my cards temps were higher during the successfully running 3dMark05 test when the GPU was underclocked than when the problems begin to occur when the card was at it's 'normal overclocked' settings.

Surely that rules out the problem is caused by temps?

edit: cleaned up, spelling.

 

[Vogon]

The fact your envir temp is A: higher than GPU and B: considerably higher than system (which I am assuming is temp1 or 3 ?) is something that doesn't sit right.
I am wondering is something damaged around that area, and causing the probs ?
I take it the memory is still giving you problems when you try underclocking to 700 ish ?
Either way, it doesn't look like an actual heat issue.
I would email Gibbo, see if he can find out how they tested, etc.
From all the things you have tried, it points very much to a faulty card

 

[mart981]

Think I must have been editing in that last screenie and text whilst you posted.

Anyway, yeah I think it does rule out temp being the problem and maybe driver issues aswell?

I still haven't been able to underclock the mem, if I try 8mhz (or any) increments in low level mode I get a black screen of death, and if I try 8mhz (or any) increments in Driver level mode the settings spring back to defaults when I click apply. (Using 8.321.0.0 drivers, no Catalyst Control installed)

The problem solving GPU underclock was done in low level mode.

I can't face any more artifacts, restarts or crashes tonight. But I'm not really sure if it's worth the effort playing with the memory now anyway? I think I've pretty much identifed the cause in the GPU clock setting.

Who is Gibbo and what's his email?

edit: I just saw on the faq not to email mods with ocuk shop related enquiries and I don't have permission to PM him anyway

Maybe you could point him in this direction?

 

[Vogon]

I thought his email was on the contacts page, but it seems not now.
He is one of the management team, and usually very helpfull

 

[mart981]

Well the people I'm receiving messages from using the webnote system are being far from helpful, I'm getting one line replies with no tech support at all.

I have done many tests with different drivers, PSU's, case temps etc and found the only way to cure the problem I described in my initial tech support request and RMA enquiry (which I recieved no tech support, only an automated RMA code) is to underclock the GPU to a 'normal' X1950 speed of 580GHz.

Only then can I run gfx benchmarking programs such as 3dmark05 and run 3D intensive games for significant periods of time.

Surely this points to a fault with the card? It won't run at advertised speeds.

Can you provide any after sales tech support?

Also exactly how and what extensive and extended tests did you run on the card to judge it was not faulty?

Please forward the relevant info to tech support also.

Thanks for your help.

Reply

It was run through loops of 3d mark 06 with settings at full.