Jump to content
Corsair Logo Corsair Logo
Corsair Community
CORSAIR COMMUNITY Survey, we want to hear your feedback!

Instability with uppgraded graphicscard

gluffse

By gluffse
in Memory

Recommended Posts

gluffse

gluffse

Posted
Posted
Hi, I have some trouble with my computer, namely instability. Problems begun when I bought a new graphicscard, radeon x1950pro, and changed from my old geforce 6800. Because of lack of power from my old PSU, a 420W, I also bought a new 560W. The computer has always been stable before but now random crashes, reboots. Mainly when I play games but windows crashes from time to time. I have changed motherboard, changed from XP to Vista, RMA my new radeon x1950pro but to no avail. Finally I did memtest http://hcidesign.com/memtest/ and had 3 errors in 10 minutes when the memory was set to 2-3-3-6 at 2,7 V. Changed to 3-3-3-8 still errors but much less frequently. Ran memtestx86 at 2-3-3-6 for 8 hours no errors. Somehow it seems as if errors become more frequently during load maybe bad PSU? Or maybe the memtest isn't compatible with vista? But with lower timings I have also lower incidence of crashes and have been able to play a little. I should also point out that I changed back to my old graphicscard and everything worked as it should. I am wondering if it is a hardware fault or some kind of incompatability? This is my first post in a forum ever and when I registered I filled in my system hardware specs so I hope it shows in my post else I have to post it.
Link to comment
Share on other sites
Garvin

Garvin

Posted
Posted

Your system specs showed up just fine. I hate to be the one to tell you this but after looking up the specs on that Chieftec (+12volt rails: 14A, 23A, 15A; 28A peak max combined with +12V1 and +12V3 not exceeding 300watts of draw) it looks like it doesn't meet the recommended 30A combined +12volt spec of the 1950pro 512MB card (standard edition, the extreme version probably has even higher requirements). You might try swapping to the other six pin PCI-E power connector to see if that might get the card hooked to the 23A rail. You might also try removing power from two or three of those hard drives to see if that resolves the issues. From your description, this does look like a power problem.

 

ps: all versions of memtest are not created equal, mainly because most of them are not kept up to date to support the widest range of hardware. The industry approved version (not just Corsair) is named Memtest86+ (latest version is 1.70) and may be found here-

 

http://www.memtest.org/

 

In my opinion, running any memory testing utility from within any version of Windows (or any full bore OS for that matter) is kind of pointless, because OS's (operating systems) reserve areas in memory that will remain untested by the utility.

Link to comment
Share on other sites
Garvin

Garvin

Posted
Posted
Yep, I think I misread the 28A spec, if you take 440watts and divide by 12 you get 36.6 amps. The problem is getting the card on the the high 23 amp rail so it isn't starved by the limitations of the 14 and 15 amp rails. I'd sure like to see the circuit diagram to figure out how they came up with those three rail amp specs with a 300watt limit on the combined +12V1 and +12V3.
Link to comment
Share on other sites
gluffse

gluffse

Posted
  • Author
Posted

For my graphicscard they recommend a 450W PSU. My PSU is designed to support SLI/Crossfire with two 6 pins molex connectors. I think the PSU is more than enough capable to support my system. But if it is a bad PSU maybe it is having trouble delivering enough power at a steady rate. I think I will test by switch to my old 420W even if it is below the specs and only connect the hardware that I really need.

I was wondering if the RAM might be bad. I mean if the much faster new graphicscard is applying pressure on the RAM which it can't cope with, hence the errors. My old one wasn't so fast so maybe the errors didn't show. Just a thought don't know so much about memory.

 

 

All these problems is really annoying and it is difficult to track them down. Reminds me of a problem with a CD-player I had 12 years ago. Never solved that one but it didn't matter, uppgraded my computer soon after and the CD-player worked fine with the new hardware. It still is installed in an old system but hasn't been used for years. I hate to throw away things that work even if I guess I will never use them again. Problem is I have 7 systems left since I bought my first computer in 1990 and they are beginning to take up some space :sigh!:

The oldest thing I still use is my old keyboard which came with my first computer an IBM PS/2 Model 30286. A friend of mine said that he would laugh if my computer problems could be related to this keyboard because I like it so much. Hope not, typing on it right now.:[pouts:

Link to comment
Share on other sites
Keepcostdown

Keepcostdown

Posted
Posted

This gives me an idea that I can think about while cutting the grass. RAM GUY, I have posted an issue in a thread below called "PC2700 Compatiability". It was a follow up to a previous problem. However reading this thread got me looking and thinking I might have stumbled across the answer to my problems. I too have occasionally experienced problems where the system would freeze momentarily, think and then shut down. To restart I have to reset the PSU then it will fire back up to only fail again. At first I thought memory problems after doing Memtest then I suspected virus. Now I suspect PSU and this is why.

First, If there is power fluctuation then it will effect memory and cause errors ( clearly stated on this forum. Hence bad test results and shut downs during the test.

Second, Having to reset the PSU after failure as if the PSU circuit protector had failed.

Third, Antec 550W with only 24A on the 12v rail which does not meet the minimum requirement of 30A on the 12V.

Fourth, Computer loaded with extra goodies.

Finally, after running a quick test the noticable slow down of the fans and dimming of the MB LED just prior to failure. Coincedence? I never noticed the slow down in the fan speed when all the fans are running. They are loud and drowned each other out however, after unplugging all the fans except the CPU and the PSU fan you can clearly here a decrease in fan speed shortly before the computer quits.

 

Is my thought process logical? Is there a easy way to verify short of replacing the PSU?

 

I have been wanting to upgrade the MB, CPU and RAM. Didnt necessarily want to buy a PSU but if I purchase smartly the new setup should hold me for at least another year or more. Then I can make it a work platform and build me a screaming machine.

 

Thoughts?

Robert

 

Disclaimer: Did not mean to hijack the thread but after reading this it made me think and investigate. It helped me think about what I was dealing with.

Thanks

Link to comment
Share on other sites
  • Corsair Employee
RAM GUY

RAM GUY

Posted
  • Corsair Employee
Posted
Yes you may be right, a weak or noisy PSU will or can cause these type of symptoms. PSU's are like the engine in your car and after time they may still work but if you run them at peak power all the time eventually they will start to loose power and or just get noisy.
Link to comment
Share on other sites
gluffse

gluffse

Posted
  • Author
Posted

the EZ plug is connected but with a standard harddisk 4 pins connector. I don't think it is the issue here because it sounds as if its only necessary if you are running two cards in crossfire mode. My other board a MSI K8N Neo4-FI didn't have one. I changed it because I thought the problem had something to do with either hardware incompatibility or that the motherboard had problems feeding power to the graphicscard. But the problems remained.

 

It is much focus on the 30A over one 12V rail. I don't think that one radeon X1950pro uses 360W, so I guess that is the whole system. I think that my PSU's rail with the lowest spec, 14A*12V=168W is quite enough if it is only powering my graphicscard.

Link to comment
Share on other sites
Garvin

Garvin

Posted
Posted

I only mentioned the EZ-plug because I've seen a system with one 7800GT in an SLI board with a lesser supply than the one you have magically stabilize after plugging in a four wire molex into it. You're right, the plugs there mainly for dual card setups to prevent tempory localized bus voltage dips. The only other possibility I can think of is that maybe after using Add/Remove programs to remove the NVidia video drivers there may be pieces left that are giving the Ati drivers fits. I've seen this happen frequently when going from an NVidea based video system to an Ati based video system or vice versa. To get those remnants off a system, I use the registry cleaner of a free utility called CCleaner slim (no yahoo toolbar) available here:

 

http://majorgeeks.com/CCleaner_Slim_No_Yahoo_Toolbar_English_d4191.html

 

First manualy set a System Restore point then after running the Registry scan of CCleaner, manually select from the items listed only those that specifically mention BOTH NVidia and video for removal. Do NOT use the Clean All option, doing so can leave your system unable to boot. If in doubt as to whether an item is a NVidea video driver piece (say a DDL file) or an Nvidea chipset driver piece, don't have CCleaner remove it.

 

ps: the download links for CCleaner slim are centered on the Majorgeeks page and have little country flags next to them.

Link to comment
Share on other sites
gluffse

gluffse

Posted
  • Author
Posted
I thought so too but that was the case with my old operative system, Win XP. This is a clean install with windows vista and the HDD was formated prior to install. The old graphicscard has never been installed. I have done some tests earlier today, will post them tomorrow, now I shall get ready for work.
Link to comment
Share on other sites
gluffse

gluffse

Posted
  • Author
Posted

Ok, so here we go. Disconnected my new PSU and used my old one 420W also a chieftec. Only two HDDs, motherboard, graphicscard, memory, CPU connected. Memory failures with 2-3-3-6 2,70V using http://hcidesign.com/memtest/ . Tested with both modules alone and both times errors, indicating that the modules are OK because it is not likely that two modules are faulty. I told a friend of mine to test his computer using the same memtest and he passed with no errors. He is also running vista. I know that this memtest isn't the one you usually run but it is only under windows I get problems so I think I can use it under these circumstances. Seems to correlate with my problems also, I mean less errors during testing = game testing more stable. I shall later this week lend my graphicscard to my friend so he can test it in his system. My old card will be used in my system and the memory tested just to make sure that there were no problems before I inserted my new card.

 

Thanks for the tip I will check out the manufacturers homepage. I am beginning to give up because this takes a lot of time and it seems like I don't get nowhere. Thanks for all the advice!

Link to comment
Share on other sites
gluffse

gluffse

Posted
  • Author
Posted

I gave my radeon X1950pro to my friend and it has worked fine in his system. Installed my old 6800 and ran memtest with memory timings 2-3-3-6 2,70V and actually had some errors though not as common as before. This is not good. It seems as when the system is under higher load the errors become more common. I have no extra computer in which I can test my RAM, but I can say that errors only occur under windows, memtestx86 ran fine with these timings for 8 hours.

Maybe I have to low voltage?

Maybe it is the processor?

PSU?

But I can't buy new things just to test if it is the culprit.

Link to comment
Share on other sites
gluffse

gluffse

Posted
  • Author
Posted

This test:

http://hcidesign.com/memtest/

 

I get no error code just that something like copying from XXXXXX to XXXXXXX did not generate a correct copy meaning RAM error. Interesting enough, after changing back to my old graphicscard I am experiencing hangs, something I never had before with my old PSU (everything is exactly the same as before just new PSU...well exept for my new motherboard but I had the same problem with my old one after upgrading). Sadly I never ran this memtest before so I can't say anything about that.

But problems seem to increase during higher load. I will replace my PSU with my old one tomorrow and run memtest to see if it works better. If it does then I think there is something wrong with my PSU and will replace it. I mean problems begun after replacing both graphicscard and PSU. If there is problems then I will direct my efforts towards the RAM.

This is very hard to pinpoint, new tests you do seems to contradict earlier tests you have done.

 

The thing I said before about that I don't have another computer to test the RAM is not quite true. I have and older Athlon XP 2400+ with a Biostar motherboard with Nforce 2 chipset capable of running DDR@400 MHz but the processors FSB only supports 266 MHz so I don't think it would be comparable...or am I wrong?

Link to comment
Share on other sites
gluffse

gluffse

Posted
  • Author
Posted
I am running it in windows vista. I have ran http://www.memtest.org with the same timings for 8 hours with no errors. In that case I don't know what the problem is. I occasioally have BSODs while doing things and now I have my original setup except for my PSU and before there were no problems for one year. A friend of mine however ran the test that says I have errors and he came out fine. Changing timings also changed the frequency of errors and generally the stability of the system in all. Will test tomorrow with my old PSU to see if there is a problem with the new one.
Link to comment
Share on other sites
  • 1 month later...
gluffse

gluffse

Posted
  • Author
Posted

Hi, after a long time I have finally brought my memory sticks to my friend and we got errors when running http://hcidesign.com/memtest/ so I guess I have bad RAM. Every component I have changed hasn't made the errors go away. I also ran prime95 and the blended test which tests a lot of ram hangs because of errors after 3-4 hours. The other tests in prime95 worked fine for 12 hours. The only strange thing is that both the sticks seem to be bad because I also get errors in single setup.

 

What should I do?

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

Go to topic listing
×
×
  • Create New...