X1950 Pro lockups/freezes

DarksideEE7

I use ATI Tray Tools. It's located in the hardware>overclocking section .
You can also use Rivatuner.

Looking at the pic of the VRMs, it looks like they are either painted gold, or they have some sort of small copper coating for cooling.

The reference ATI photo that you posted had different colored VRMs.

Here is a pic of the X1950XT. I wonder if anyone sells a heatsink like this.
I also wonder if my HIS cooler makes contact with the VRMs when the core cooler is on. I see that the ICs next to them are cooled with their own heatsink, so why the hell are the digital VRMs not cooler? I'm pretty sure that they're what's holding me back on my core OC, because when I turn up the PCI-E bus speed I can set higher clocks, but the screen goes black after 150s or so.
I see on the X1950 XT that has PWM and VRM circuitry...or at least they look like VRMs on a motherboard. WTF is up with that? Can they have both digital and PWM/VRM circuitry?

DarksideEE7

I contacted HIS and they said that the card does in fact have heatsinks of the VRMs.

Eeastcoasthandle

Not according to this pic unless they PCB is simply configured differently. I am not sure what's underneath the HS on the far right side. Did you ask specifically for the pro?

Here is a pic with a Right side heatsink removed. It appears to be vrm on the far right though.




Here is a pic of the HIS with the right side heatsink on. Again, I am not sure what's underneath it.

Aim2009

Does ANYONE have an X1950 Pro that is working properly??? My system came out of the box yesterday and the first thing it did was crash the 3DBenchmark. I tried several different drivers, with and without Catalyst, and I tried the Omegas. The temps are less than 50c. The card also crashes on anything on the 3D Menu in the Catalyst Control Center (where it looks like a car driving) simply clicking (not changing any settings) just clicking crashes to a black screen, monitor goes on standby. I'm RMAing the card. Just wanted to know if you guys agreed I got a dud (seeing as how it won't Benchmark) or is this par for the course of the x1950? I've read so many horror stories. Can this card really work? It's got awards all over, but it seems to come with twice the nightmares.... I think I owned a car like this once. Sure was pretty to look at.....


I don't know if this will be a clue to anyone but I just ran the DirectX tests and it crashes instantly (black screen, card just cuts out) when I run Direct Draw. Direct3D passes fine. What gives? Still a hardware issue? I have the ATI 8.321 drivers right now.

temeteus82

I have one working just fine. Works fine after BIOS update from Sapphire.

Aim2009

Glad to hear it CAN work. Forgive me for sounding like a complete n00b. But Sapphire released a BIOS update? Is there a way to update the BIOS on the card? Or are you talking about the mobo BIOS? I have an Asus board, should I be looking to update that? Or going to the manufacturer of my card Connect3D ?

Thanks Temeteus

temeteus82

It's a card BIOS update. Yes the BIOS on GFX card can be updated. I used ATIWinflash to update my X1950 Pro's BIOS.

Athanis

Maybe is a stupid question, but as i am having similar porblems: how do you get the last BIOS for the card?
I had a look at the Sapphire web site and I could not find it.
Thanks.

temeteus82

You got the X1950 Pro Ultimate? It has it's own place in the pull down menu.

Athanis

No, is not the ultimate, just a normal Ati Radeon X1950 Pro with 256 Mb of memory.
I found in the web the utils to update the BIOS, but not the BIOS files.
I just checked and my BIOS is version 009.013.001.010 with date 2006/09/25. Is this the last version?

temeteus82

Then you might want to take a look on Sapphires own forum and make a support request.

Athanis

Just one question: up to which temperature is running your card?
Thanks.

temeteus82

in stock:
idle gpu : 49C mem : 42C
load gpu: 68C mem: 50C

getting lower temp when using att to adjust fan speed.

Athanis

Thanks, i found now the problem (i think). The temperature of the cpu when idle was close to 60 C :-( . I open the pc, cleaned all the fans and dust, and now is down to 47 C.
I hope that would do the trick.

Thanks again.

billorama

ok iv just figured its nothing to do with over heating. i have the agp HIS 1950pro 512mb. i was haveing the same problems as everyone. card crashing when creeeping over 60-64. so i lowered the gpu clock and memory to the actualy ati reference settings using the new ati tray tools and now the card can hit 80C without crashing. alot of people seem to be having these problems and its definatley the last HIS card i buy. it was the most expensive, as well, and it doesnt even operate properly. if it werent for these utilities... this card wouldnt work.

Tipstaff

Same case for me. For some reason HIS clocked their cards waaaay above spec, even more than what they say they do on the box. In my case the video card was clocked at 634 for the GPU (14 above what the box says), and 742.50 (or 1485 DDR2) for the memory. Every time the card went above 60C it'd lock up. Once you clock it back it'll run perfectly. However, since I have 2 cards in Crossfire I was getting tired of having to always underclock both cards and adjusting the fans, so I decided to flash the cards with a new bios that had lower clock speeds (to the actual box speed of 620). While I was at it I also adjusted the fan speeds and the temps they would kick in at (40C = 50% fan, 50C 75% fan, 60C = 100% fan, but I tell you, the card never goes above 60C anymore). Been working perfectly ever since.

>GSXR<mrbusa

These issues are well documented and have been fixed by HIS. They just didn't QC the reference boards at the high temps that some people run them at. It isn't the core or the memory overheating, it is the voltage regulator chips. I have the revision 1 ICQ AGP card and never had one problem. I even overclocked it to 675/805 and it never had one artifact or crash. The card never went above mid fifties on the eighty degree days. One thing I did notice is the HS/F assembly loosened up after burn in of the thermal pads on the Vram and VRM's. As far as the cards being clocked above advertised speeds go, that is a reference design flaw. These boards act weird when the multiplier is messed with. It is not just the x1950 series cards. My 8800 is advertised as 500/800 and when I monitor the clocks they read 513/792. I checked the BIOS on both cards and they were both indeed programed at there advertised speeds. If more people cared about there ambient temps and thought and configured there case airflow then there wouldn't be so many RMA's and maybe graphics card prices wouldn't be so high.
I am no fanboy of any company and hate corporate world with a passion. When I hear people saying that they can't figure out why there hardware is crashing at 65c, 75c, 80c, 90c +...then I just have to say something.

graycat

Well I'm glad to see I'm not the only one having problems with this card!

This thread has proved very useful to me so far and given me lots of places to start looking. However, unlike some people's reported problem my card seems to die at about 42C when under load so I think I can discount that side of things. Am I right in thinking that the other main cuplrits are: insufficient power to the card and drivers?

Apologies for any n00b questions / comments / habits / smells but I'm new to the whole hardware side of things lol

>GSXR<mrbusa

What is your card doing?
Is it doing it on a consistent basis under the same circumstances at the same or about the same temp threshold?
What are the rest of your system specs, especially your PSU, and what GPU drivers are you running?
Are there any hot spots on the card? The best way to test this is to have a laser targeting thermometer. If you use your fingers to feel around the backside of the card then be careful. If something is abnormally hot then you will definitely be able to tell.
Are you overclocking any part of your system? If so then please list what.

graycat

sorry, should have realised I'd no posted my kit etc.

Mobo: asus p4s33
CPU: P4 2.4 Ghz
RAM: 3Gb
GPU: X1950 pro 512 AGP
Drivers: latest from the AMD site - 7.5 I believe
PSU: oem one that came with the PC. 350W

What's happening is that I'll be playing Joint Ops (no laughing at the back!) and it'll go for anyway between 2 and 20 mins with all being well. then suddenly it's just freeze, the sound will stay with the last "note" the screen freezes completely and the only way to unstick it is with a hard restart. Up to that point they'll be no artifacts that I can tell, game play will be smooth and all goes well. I've installed ATT and whacked the fan up to 100%, underclocked it as much as possible and got teh GPU temp up in top right. Usualy it's mid to low 30's going up to the 40's during game play. I've tried different graphic levels in the game to see if that makes and difference and it doesn't so far. I've also popped the side off the PC to be on the safe side, but all to no effect so far.

when just tooling around with movies etc it all works fine with no lagging or freezing.

I've got a Q-Tec 550W PSU ( http://www.qtec.info/products/product.htm?artnr=13492 ) that's being *cough* retired *cough* from work to try tonight and I'll also be taking the monitor off onto it's own socket rather than through the PSU.

On the hotspot side, I've not had a feel around it yet but will do tonight. It's going to be tough though as one side has a HUGE heatsink and fan on it but the other should be free.

No overclocking at all other than the new card

>GSXR<mrbusa

These cards need a good PSU. Look on the side of the 350w and see how many amps are spread across the 12v rail or rails. The Q-tec you are talking about is pretty old and as far as I can tell only has 20a on one 12v rail. The card requires around 30a.
The latest drivers have no debilitating problems with these cards anymore so you are fine there.
I am aware these cards have a huge HS/F. All you can do is feel anything you can get at and see if something feels extremely hot to the touch. That is why I said that a laser thermometer is best.
Underclocking the core or vram to much and then trying to run a 3d game or benchmark can have bad results and cause crashes also.
Your first step is to get a PSU that has the required power to run the card and the rest of your system with power to spare. Anything over 400w with the required amps should be fine for your system as long as it is a quality unit and the rails do not droop to much under load. If you have ten harddrives, ten fans, ten lights, and ten USB devices then you will need more power. Ten is allot but you know what I mean. You have to remember that if a PSU has multiple rails the you will need over 30a combined to be within recommended specs. Each PSU is different in that regard and the easiest way to make sure is to get a decent unit. If it has more than one 12v rail then make sure the combined 12v rails has at least 38a to 40a or more. There is no real science without being able to test the PSU for stability and output. Manufacturers of power supplies like to throw up smoke and mirrors and make there units seem more powerful than they really are. That is why most enthusiasts do not cut corners when it comes to powering there systems. There is a point of overkill for an older system if you do not plan on upgrading any time soon.
I would say that the odds of this being a power problem is above 95% as long as the rest of your system and BIOS are configured properly and the drivers had a clean install.

graycat