DXRenderer.dll error

[VMFA117_Poko]

No. This happens also in GTAIV when I don't use TIR. I did stability test with FurMark and with GPU fan set to 100% I get 100 degrees of Celsius on GPUin less than 20 minutes.

[Ramstein]

No. This happens also in GTAIV when I don't use TIR. I did stability test with FurMark and with GPU fan set to 100% I get 100 degrees of Celsius on GPUin less than 20 minutes.

 

 

that's crazy hot! 100 C degrees!

 

no wonder why your computer is having problems..

:doh:

[VMFA117_Poko]

Yes, I know. I can prepare a scramble eggs on it. No wonder games crashes after 20 minutes of play...

 

Interesting thing is it didn't happen before installing 9.1 catalyst. Now both cards reach such temperatures even when I rolled back to earlier catalyst. Well done ATI! You are real Websters! Bravo!

Edited April 7, 2009 by Poko24

[Kuky]

Poko, your VGA cadrs are HD3850 right? Do they have full cover with fan blowing air OUT of the case to the rear (where you screw card to case)?

 

If your cards are getting that hot it's definitely not good... my HD4870 were 80°C on default fan profile and that's hot already (and they say normal for these cards). If your cards have just fan/heatsink combo that doesn't even blow the air out of the case.... well, that case will get hot pretty soon... which is why I think you get crashes after aprohimately regular time of gameplay.

 

First, get your cards out (and leave only 1 inside to test) and you might wanna take the heatsink off and make sure there is propper contact between the GPU and heatsink. Also you might wanna look into flashing video card BIOS with better fan profile (which is what I did) but then again if you say that even with 100% fan speed you still get hot cards...well that's not healthy... and if it's BOTH of them... it can't be just faulty card.

 

So I gotta drive back home... will check on your update(s)

[VMFA117_Poko]

My VGA cards are PowerColor HD4850.

 

 

Their fan/heatsink combo doesn't blow hot air out but into the case (Cooler Master Stacker) wich is really airy. I have additional fans inside of the case too. I did test with both cards inside, one, then the other. Same temperatures under FurMark Stability test (1920x1200 and MSAA x0). I tried Fault-Safe and Optimised BIOS settings and now I only increased my RAM voltage to 2.24V, cuz as I chcecked default was too low.

 

I don't have any graphic disstortions - just crashes to desktop after while or longer. Error logs always point on ATI drivers or DirectX dlls. I have never overclocked anything and problems started about 2 months ago after installing 9.1 catalyst.

[Infiniti3d]

@Poko24, Can you post a pic of your rig? We can get a better view of what we're dealing with. I'am not sure what causes your problem but it might be a combination of a bit overheating and drivers. What utillity do you use for monitoring the temps? Also consider for the time the problem keeps occuring to leave only on card in, and one stick of ram. To make absolutely sure that its not RAM related you might run that one stick on 800mhz and 2.2V -2.4V and piont a big fan towards the GFX card with the side panel removed. It sounds radical, but it wouldbe dissapointing if you throw in a new Nvidia Card and then noticeing the problem is still there!

Edited April 7, 2009 by Infiniti3d

[Kuky]

Yeah, those heatsinks do not work very well to be honest because these cards do get very hot and all this hot air (remember we are talking about 80°C or more when gaming) is blown into the case and trust me... unless you keep your case completely open it'll get very warm in there.

 

Now, if you're running E8500 at default 3.16GHz your FSB should be 333MHz. Set RAm to 666MHz and that will keep your CPU:RAM to 1:1 ratio which is best and most stable and everything else (PCi-E and PCI buss will be in sync).

 

Also let RAM timings be automatic and set by SPD (or auto) in BIOS and leave all voltages on auto so you eliminate chances of your system having wrong settings in BIOS. If it still crashes and your CPU/RAM/VGA are all at default, you must have hardware issues... and possibly just due to video cards being little too hot... usually if you see artifacts it's your memory on the video card but since you don't have that your video card RAM should not be causing this.

 

One other thing is that since you have CF setup, your NorthBridge voltage might need little boost as this is what's controlling FSB, PCI-E and standard PCI buss. If you run two cards it means double the data going through PCI-E buss and double the work for NB and this can cause instability.

 

Here's one more tip... search for and download Intel Burn Test and use this (as it is definitely the most reliable tool for testing system stability) instead of ORTHOS. This utility will get your CPU to get even hoter then for example runnig two instances of ORTHOS and if after runnig this and you see you have no errors even after 20min of testing your system beside your video cards, should be stable completely... this would then leave only video cards as the cause... if you leave either of them alone and you still get crashes you could try connecting PCI-E power line, if it makes no difference, then try upping NB voltage a bit (do little steps please and read BIOS description caefully as to what default/recomended voltage for NB is... give it only 0.1 or maybe 0.2V more over default).

 

There are really many posibilities in BIOS settings when you overclock and any can cause instability but since you are not overclocking this should not matter and everything should be on auto values in BIOS. If even with 1 card and no overclocking at all and you still get crashes after some time, I think it's gotta be video card getting to hot or sadly maybe even your motherboard is not very good one and in this case I might wanna return it and change for another one.

 

If you still see your video card is over 80°C then you could try BIOS mod and flush video card BIOS with different FAN profile. I have done same with my two cards (now one is sold few weeks ago because I feel CF is not very good choice and not very good value for money). To check how to do BIOS mod go to TechPowerup website (http://www.techpowerup.com/rbe/) and follow instructions there... it's not very hard at all... one advice though, if you decide to flush your video card BIOS, do it for 1 card only. If in case flashing does not work well and your card doesn't boot, you can put other one in and flush original BIOS again to the second card.

Edited April 7, 2009 by Kuky

[Kuky]

Oh by the way... if you use 4 stick of RAM try taking 2 out and have only 1 pair working in dual channel because running 2 video cards in CF and all 4 channels for RAM at the same time is LOT of data for NB and it can cause instability... in which case again you might need to up the NB voltage... see how you go...

[Yskonyn]

I had an ATi X1950 Pro in my old computer and I got the atixxx.dll errors as well with a few games. Command and Conquer 3 was the most notoriously title crashing constantly after about an hour with those errors.

At first I thought disabling Catalyst AI had solved the problem, but it didn't. The only way I kept the system from crashing was to reduce details in the game options. It would run fine that way, eventhough I exceeded recommended system specs for the game.

 

I never really figured out what the problem was, but it seems related to stress on the card in combination with (possible) driver issues.

Hence this looks quite familiar to your problem. I hope you can figure out what causes the problem.

Edited April 7, 2009 by Yskonyn

[Infiniti3d]

@Kuky: All you mentioned Poko has allready tryed with no succes, read the threat. He also mentioned his Ram modules are rated 2.2V-2.4V. Increasing the nothbridge Volts is unneccesary becouse he has the same problems with a single card installed, doing this might get him into more trouble he allready is. He has 2X2GB as he mentioned earlyer not 4X1GB and he tryed one stick already. Reduceing the memory speed to 800mhz he might get more stabillty, running manufacturers recommended settings and setting them manually, at 2.2 or 2.3V wich is the recommended by the vendor.

 

Edit:Why do you think his FSB is 533 at default? A E8500 is a 1333FSB CPU with a multi of 9.5 so 1333:4= 333X9.5=3.16 GHZ. So all he has to try is unlink FSB/RAM ratio and leave the FSB at 333.00 and set the RAM to 800MHZ.

 

I suggest Poko24 you go here http://forums.guru3d.com/ and explain you issue again, there are people there that had a pc when you and i not even were born.

Edited April 7, 2009 by Infiniti3d

[Kuky]

^^^ ah yeah...my mistake there Infinity3d... his default FSB should be 333 not 533 (I kept thinking I was using 533 when I was playing with overclock settings)... need to edit my post... by the way if he leaves FSB to 333 and CPU:RAM in 1:1 ratio he can run his RAM at even 1.8V

 

Anyway, I couldn't remember all that was said in the previous posts so that's why I posted all the things I think he could try... I know he tried already most, if not all... maybe something I mentioned he did not try... still I'm just trying to help and give my feedback... then again I'm not having a go at you... its still a fair point.

[Ramstein]

Two things I see here that grab my attention:

 

1)the heat is way too high, get that heat out and temp down!

(if you need more and larger fans use them!.. even take the side of the case off and run it to see how it runs...)

2)The power supply must have enough clean power for 2 video cards and other parts... (what brand and voltage, and how many rails does it have?)

 

Not sure if you ever said you have motherboard monitor software which records temps and voltages...

 

Using auto settings on the motherboard Bios should be enough to run the computer with no glitches.. (IMHO).

[Slayer]

Have you ever considered that your power supply may be too small? You are using a 600watt PSU with a lot of high end hardware. If this were an Nvidia based system I could say with some assurance youd probably need about a 750 watt PSU minimal.

 

Your system probably works ok in windows but when you start loadiing the components and your power requirement jumps up things get shaky. Alao does your motherboard have an auxillary power hookup for supplying extra power to the PCI-E ? If so is it hooked up?

 

The PSU is one of the most overlooked components by people and its one of the most important.

 

For Nvidia Systems heres a link of interest:

 

http://www.slizone.com/object/slizone_build_psu.html

Edited April 8, 2009 by Slayer

[Kuky]

I think if Poko takes out 2 stick of RAM and runs his PC with default CPU speed and only 2 stick of RAM and single video card, that 600W PSU should be good enough, provided it's a good quality PSU and it's power is not overrated. I think if it was the PSU he would get crashes very randomly... not just about 20min into playing 3D games. Then again if the video cards were to blame or were faulty... chances of both of them being faulty is very slim and still they should be designed to run default speeds even if the fan is not that good if at least you have a case that is ventilated decently... so I'm really out of ideas here... Poko, can you get someone elses PSU that's more powerfull that yours just to eliminate this option.

 

Also I do think if he runs 4 sticks of RAM and 2 video cards the motherboard NB voltage should be increased a bit because with full number of devices working on all major BUS-es NB does have hard time keeping up.

[VMFA117_Poko]

Hi guys. I did not build my rig by myself. I've only chosen elements. The rig was built by proffesionals (at least they should be) at (work)shop. And yes, I considered I have too small power supply but when I said in shop "I want 1000W PSU" they laught and said that this 600W Tagan is more than enough for these two power saving Radeons and power saving E8500. And indeed brochures says that my two Radeons in CF needs 550W, so maybe 600W is not much more than enough but minimum. However, remember I didn't overclocked anything and my rig worked fine for 5 months til I installed cat9.1. Isn't it strange? I did not mess in BIOS before the problems.

 

PSU is connected to motherboard via 24pin plug and 4pin plug with each gcard.

 

My Radeons temps at idle are 57 degrees (showing in CCC override page). I used FurMark to stress the gcards and this program shows 100 degrees after 20 min of testing. Games crash after about 20 minutes too.

 

I will borrow older nVidia from friend and will check if it works ok.

Edited April 8, 2009 by Poko24

[Kuky]

Poko, I think your video cards are getting to hot. 100°C is too hot for the GPU. If on my HD4870 (big brother of HD4850) goes to 80°C under load on deafault fan settings 100°C for your one is too high.

 

PS: did you clean the registry also before you installed new driver? If not do that also. Here's how I do it:

 

- uninstall driver from windows add/remove programs

- restart and boot into safe mode then I do search for all files starting with ATI (ie ati*) then delete them all from C: drive (I keep other files on D: drive so this way I dont' accidentaly delete files I want to keep)

- open registry (start/run menu and type "regedit" and hit enter). Look into all these and delete what ever registries are in ...../Video folder

 

a) HKEY_LOCAL_MACHINE/SYSTEM/ControlSet001/Control/Video

b) HKEY_LOCAL_MACHINE/SYSTEM/ControlSet003/Control/Video

c) HKEY_LOCAL_MACHINE/SYSTEM/CurrentControlSet/Control/Video

d) HKEY_CURRENT_CONFIG/System/CurrentControlSet/Control/VIDEO

*if folder is already empty that's fine

 

- restart and boot normally... then install new driver

Edited April 8, 2009 by Kuky

[VMFA117_Poko]

Yeah, I know... 100 degress is reached only in FurMark. In games top temp was 85-90 degrees. I just cant believe both cards have same problem and I don't get any graphic disstortions . This is probably shi**y PowerColor product... I need to give back them to the store and hope those pros will take them. Remeber, I had to ripp off the stickies from heatsink case 'cuz they unwrapped under heat.

 

Yes, Kuky I always clean up old drivers and registry. Same thing happened after full format.

Edited April 8, 2009 by Poko24

[Kuky]

You know what's funny... I was thinking exact same thing.. my cards are Powercolor also but eventually I got my issues sorted and turned out it's not the cards that was doing it, rather my overclocking settings and also running 2 high end video cards does stress the motherboard a lot and even after I got 1200W PSU it was stil unstable untill I decided to downclock a little (from what I wanted 4GHz to 3.8GHz) Haven't had issues like that since. Then again I got rid of CF setup... single card does work better.

[VMFA117_Poko]

Remeber your problem Kuky. We spoke about it about half a year earlier when I planned to buy new pc. I was asking on the forum did I choosed proper parts for a rig before buying. You said I should think about bigger PSU if I go for CF setup. That is why I said in store I want 1000W PSU :).

[Kuky]

yeah.. and probably you should have gone with it anyway... I have found in many, many shops sale people don't have a clue about what they sell. Unless people selling this stuf are not PC freaks like some of us.. they are clueless... and also, they just want to sell what they have... if they don't have 1000W PSU then that shop is not that good as it means they don't seel high end hardware and if they don't sell it they wno't have a clue about it.

Edited April 8, 2009 by Kuky

[Kuky]

Also you know what's the problem with many PSU... even though they might have total 600W for example... it depends how this power is distributed, or how many rails it has. If the rail on which the PCI-E connector does not provide enough power to the video card, it doesn't matter if the rest of the system in on good power... sure the motherboard and CPU are all well "fed" but if one componenet is not provided with enough power it'll cause issues. See I bought single 12V rail on mine because of exactly this, I wanted to make sure I never get overloaded on 12V rail because some part is drawing too much power and others have to much power provided but on;y use little.

 

Now, what is your PSU model... and how many PCI-E rails (separate cable) does it have?

[VMFA117_Poko]

yeah.. and probably you should have gone with it anyway... I have found in many, many shops sale people don't have a clue about what they sell. Unless people selling this stuf are not PC freaks like some of us.. they are clueless... and also, they just want to sell what they have... if they don't have 1000W PSU then that shop is not that good as it means they don't seel high end hardware and if they don't sell it they wno't have a clue about it.

I thought they are geeks. They had big glasses on the nose :D

 

PSU is made by TAGAN but can't tell you more untill I get back home after work.

Edited April 8, 2009 by Poko24

[Yskonyn]

Now, what is your PSU model... and how many PCI-E rails (separate cable) does it have?

 

Important note here.

Many PSU's give you a rating of a power output. But with high-end hardware you need to make sure you have a steady rating on all rails.

If you're using a weak PSU it can translate to problems you describe with your videocard.

 

I had a Sapphire card once and the forums over there a full of people using inadequate PSU's which cause trouble for the vid card.

[VMFA117_Poko]

Ok, my PSU is Tagan TG600-BZ. It has one main rail deviding on 2 rails: one supplies motherboard via 20+4pin connector and the other 4+4pin connected to motherboard near CPU. Additionaly there is 2 rails for SATA HDDs + 2 rails with 6pin connectors for PCI-Es. Also on PSU front there are 4 slots free.

[Yskonyn]

The hard part is to figure out wether is has a contant power output or just a max power output with fluctuating power average.

But that should be in the manual at the specifications table.

If its the latter I would invest in a steady power unit.