The Corsair User Forums

The Corsair User Forums (https://forum.corsair.com/forums/index.php)
-   General Questions (https://forum.corsair.com/forums/forumdisplay.php?f=129)
-   -   TR3X6G1600C8D issues... (https://forum.corsair.com/forums/showthread.php?t=78856)

Davyboy 05-24-2009 09:45 AM

TR3X6G1600C8D issues...
 
Hi, I recently built myself an i7 system, and I have been having nothing but problems with it since I put it together...

I am running Vista Ultimate x64, and the problems happen whenever I try stress testing with Prime95, OCCT, Memtest, and pretty much any other application that puts a heavy load onto the system...

When I have the BIOS on Optimal Defaults (with the RAM running at a mere 1066MHz) I can leave the rig Priming large in place FFT's for an entire night and day without an issue, but as soon as I try running it @ 1333MHz or if I enable the XMP profile for 1600MHz, the system either BSOD's, freezes up, or resets itself...

I have tried setting the RAM timings/frequency manually in the BIOS for both 1333 and 1600MHz but the same issues keep happening...

I can sit and play games like Left 4 Dead, and Battlefield 2 for hours without any issues, and I have also used the rig for converting AVI files to DVD and vice versa without any problems as well...

But I just can't get it Prime stable with the RAM running at its rated speed, the only time it is Prime stable is when the RAM is @ 1066MHz...

And the CPU is running @ stock frequency as well, as I dont want to try overclocking it until I get it running stable first...

I have contacted the people where I purchased the hardware from, but they weren't very helpful at all...

I don't really understand which piece of hardware it is that is failing, but I am suspecting it is a RAM issue because the problems only happen when I try running the RAM at 1333MHz and 1600MHz...

I have talked to loads of other people about this problem on various online forums, and some folk tried to say it was my PSU, but I have even tried running the rig with only 1 HDD connected, and an old 8500GT, and I still got the same issues...

Also tried running each stick individually to see whether the problem was being caused by one module, but it did the exact same thing with each stick...

I have tried absolutely everything I can with the hardware that I have, unforunately I dont have any other DDR3 modules to try out on my mobo to see if it is this specific RAM that is causing the problems and not another piece of hardware...

There is nothing else that I can do at this end, and the people who supplied the hardware to me are hopeless as well, so I thought that I might find someone on here who can help me figure out what the problem is with my system...

Thanks in advance

Davyboy 05-25-2009 08:28 AM

Surely there is someone on here who can help me out??

I have been running memtest since yesterday testing each module 1 at a time, in the second white slot on my mobo (slot number 3 according to gigabyte manual, even though it is the 4th slot from the CPU socket), I am letting each stick run memtest for 7 hours at the rated frequency of my modules, and so far the first stick ran without an issue for just over 7 hours before I stopped it manually..

The second stick has now been running along nicely for just over 3 hours without any issues...

I will report back here once I have finished testing each stick... However things seem to be going good so far... the first time I tried each module 1 at a time, I had it in the first white slot (slot 1) and it would crash after about 1 hour of memtest, and after about 10 minutes of Prime95...

But that time I just left all voltages alone, this time I raised QPI/VTT voltage up to 1.355v and I have DRAM voltage @ 1.66v instead of auto...

If all modules prove to be running fine in single channel mode, then why on earth wont they all run along nicely at the same time?

I read somewhere on these forums where RAMGUY said to another i7 user having trouble with these modules to up the QPI/VTT voltage to 1.6v, but I was researching into this last night and apparantly Intel say that the maximum amount of voltage for QPI should only be 1.35v... But then I noticed that some high performance RAM actually set the QPI/VTT voltage right up to 1.6v...

So I would like to know if anyone here that owns an i7 rig runs their QPI/VTT voltage at 1.6v all the time, and also if this is safe to do with my modules??

I really dont want to kill my CPU because I had to much volts going through it, and to me 1.6v does seem kinda high...

I also read somewhere last night, that the difference between Vcore and QPI/VTT voltage should only be .5v, is this correct?


::edit::

Well, I am now testing the last module in memtest86+, I will try to run it for 5 hours and see how things are, I just hope that by raising the QPI/VTT voltage to 1.355v has stablized my rig..

If this last module is fine, I will install all three modules again, and try running memtest again for 5 hours with three sticks installed... if that runs without any problems, I will try Prime95 for the entire night to check if things are stable...

I really hope that I have nailed this, as I just want this new rig to run stable... I have never had any problems like this before, and I have been building my own systems for about 9 years now...

This forum doesn't seem to be very active at all,,, I know loads of people use Corsair hardware, so I thought there would have been lots of people here willing to help, but it looks like I was wrong...

Anyway, I will report back here with my results...

xpatar 05-25-2009 03:45 PM

First of all, did you put the modules at the white slots creating a tri channel configuration? I'm referring to the white slots 1,3,5 according to your manual.
As long as you're able to keep your system stable at stock speeds it's probably ram settings. If I were you I would do the following:
Enter the bios and manually set your timings according to Corsair. Particularly I would set: Vdimm = 1.65V (if there isn't try 1.64v just to be on the safe side), timings 8-8-8-24 @ 1600MHz and leave qpi/dram @ auto. That would let the system to choose the necessary voltage for your qpi/dram automatically. Save your settings and enter os. If you're able to enter os watch your easytune utility to check what is the voltage that the system applied to your qpi/dram. That way you will avoid giving high values from the beginning. Qpi/dram voltage isn't a standard and that's dependable on the memory controller which for a corei7 is integrated to the cpu.
"I also read somewhere last night, that the difference between Vcore and QPI/VTT voltage should only be .5v, is this correct?" Well, that's unofficial, I would't take it for granted.
Also check if you have the latest bios version. You can check that link for beta bioses.
http://forums.tweaktown.com/f69/giga...st-bios-28441/

Wired 05-25-2009 04:45 PM

Quote:

Originally Posted by Davyboy (Post 402252)
Surely there is someone on here who can help me out??

As it states at the top of the forum, Corsair is currently closed today. Also, both of your posts were still before their normal hours of operation, so they wouldn't have seen either post for at least another 3 hours after the last post.

Normally 3 loops of Memtest86+ v2.11 is enough.

Quote:

If all modules prove to be running fine in single channel mode, then why on earth wont they all run along nicely at the same time?
Could be a setting, or just could be a bad memory controller (in the CPU).

Quote:

This forum doesn't seem to be very active at all,,, I know loads of people use Corsair hardware, so I thought there would have been lots of people here willing to help, but it looks like I was wrong...
Memorial Day weekend's always slow. Most people go outside and stuff :)

Davyboy 05-25-2009 08:33 PM

OK, running each module one at a time runs memtest for 5 hours no problem at all, but when I have them all installed (yes in the white slots!) it runs memtest for about 3 hours without a problem, but at some point after that it crashed and the bottom half of the screen was all corrupted with red bits,, I guess it was probably errors, but the system crashed before it got the chance to display them properly...

I took pictures of the memtest runs with 1 module at a time, and also took a pic of the screen corruption after 3 hours with all 3 modules installed running at 1600MHz with the XMP profile...

I am currently running large in place FFT's in Prime95 with all 3 sticks installed, and it has been going for about 4 hours now without any problems as well...

The problem seems to come and go, as I have managed to get it stable before, but after I turned it off, and then went back to use it the next day, I did somemore testing for stability and it started crashing again!!!

I am starting to suspect my PSU is probably the cause of this and not my RAM...

Does anyone else have any suggestions?

@xpatar,

I am running the latest official BIOS for my board, I tried the F8b beta BIOS but it caused my rig to crash even more than it already does!!!

And I have tried everything you have suggested already... I have spent almost 2 weeks trying to source the problem in this rig!!!

I will give it til the end of this week coming to see if I can get it 100% stable, and if I am still experiencing problems after next weekend, I will look into RMAing everything back to the suppliers....

xpatar 05-26-2009 01:33 AM

As you mentioned in the beginning everything works fine @ stock speed.
The system becomes unstable when you change your memory settings.
You can try also lntel Burn Test program:
http://www.ultimate-filez.com/files/IntelBurnTest.zip
Check it out for proper cooling. It stresses quite a lot cpu/ram, more than prime does.
As the author of the program says:
"Benefits of using Linpack:
1. More accurate than Prime95 Small FFTs/Blend (under x64 OS).
2. Takes less time to tell if your CPU/RAM is unstable than Prime95 (usually
something like 8 minutes Linpack vs 40 hours under Prime95).
3. Use the same stress-testing engine that Intel uses to test their products
before they are packed and put on shelves for sale."
Also before pressing the start button use 16 threads for single Core i7.
In that way you will have quick and accurate results without the need of waiting hour as prime 95 does.
Check it with stock and rated speeds. See what happens. I suppose you have changed the order of the modules. If not try doing that and start testing again. See if can you use other modules. Does this happen again?
About the psu? Well, I'm not sure about that. It has 4 independent 12v rails giving each 18Amps. That sounds cool.

Davyboy 05-26-2009 07:05 AM

~Hi again,

About the PSU xatar, it wasn't so much that I thought I was underpowered, I think that the PSU might be starting to fail, as the problems are intermittent...

I have been running Large FFT's in Prime95 for almost 14 hours now without any issues at all, with the XMP profile enabled in the BIOS, and the QPI voltage set to 1.355v...

I am just going to leave it running the whole of today, and if it is stable, I will try overclocking the system to 3GHz and see if I can get it stable up there, if everything is ok @ 3GHz I will be happy for now...

Also, I will be buying myself another 6GB of RAM so that I have extra modules that I can test with as well...

I was wondering, would having the RAM modules placed in a certain order affect the stability??

Because I have moved them around from the way I had them before, they are in a different order, but I can't see how this would make a difference, but I could be wrong...

xpatar 05-26-2009 07:34 AM

In some motherboards the order of the ram modules plays a significant role.
Al long as you don't have any problems with your psu (faulty one) and you are doing fine at stock speed then it sounds to me like a premature bios or a compatibility issue.
Simply the system works fine at stock speed but when you set the rated timings problems arise. Check it with a set of rams that are approved by gigabyte to be compatible with your board. Also try Intel burn test it will give you much quicker the answers you are searching for.

Davyboy 05-26-2009 08:01 AM

I talked to Gigabyte about this issue already, they even went to the bother of setting up a test system using the same mobo, RAM, CPU and graphics card as me, and left it for 2 days running Prime, and a bunch of other benchmarks, and they found no faults...

Also, it says on the main corsair site that the RAM is guaranteed to work on the EX58 UD5...

I am still Priming away with no problems... I think I may well have found stable settings...

Though I will still order a new set of DIMM's and a new PSU...

I can get 6GB of 1333MHz RAM for 42, and I am going for the BeQuiet! Dark Power Pro 750w PSU... I will hopefully have these in the next two weeks...

I will only be using this other RAM for testing, I want to have these Dominators as my main RAM, but if I install the new RAM and my rig is perfectly stable, I will be RMAing these particular Dominators to have them tested...

RAM GUY 05-26-2009 05:22 PM

How many modules are installed and what do you have the memory Voltage and settings set to? Also what BIOS version is installed?

Davyboy 05-26-2009 07:13 PM

Hi,

I have got the latest F7 BIOS installed from Gigabyte website, and I have got 3 sticks of RAM installed in my rig (installed in the correct slots!!!)...

For RAM timings and voltages, I have tried setting them manually, and I have also tried using the XMP profile...

Like I said already, I tested each stick one at a time using memtest bootdisc, and they ran for 5 hours before I stopped them, after that, I tried running again with all sticks installed, and it went for a few hours without any errors, but I went back to check on it a short while later, and it had froze, and the bottom half of the screen was corrupted, and I had to hard reset to exit memtest...

Also, if the RAM is left running at 1066MHz it works perfectly, but the instabilties start happening when I set it to run any higher...

To begin with it would only run Prime for about 5 or 10 minutes before it would reset itself, freeze or BSOD.. then I tried loosening up the timings, and it would manage to run fine for a few hours before crashing again...

And then the next day, I could turn it on, and start Prime95, and it will run all day no problem, and this is still using the same settings as before...

However, I read one of your posts on another thread RAMGUY where you suggested to another i7 user that increasing the QPI voltage to 1.6v might increase stability, so when I installed all three modules again yesterday, I decided to try and increase the QPI voltage up a little bit to 1.355v instead of the 1.35v that the XMP profile sets it to, and it has now been running Large FFT's in Prime95 for 26 hours without any problems at all....

I am really puzzled, and I am still worried that my system isn't running properly, because I have managed to get it Prime stable before, and then the next morning it would start crashing again....

The problems seem to be intermittent, they come and go...

And I really dont know what could be causing this issues...

I talked to Gigabyte last week, and they configured a rig using the same mobo, ram, CPU, and graphics card, and they ran a bunch of tests, and found no problems at all...

Also, like I mentioned already, I can run games on my machine and play them for ages without any trouble, it is only when I start stress testing for stability that the problems start...

I really do think the RAM has something to do with it, because @ 1066MHz it is perfect, no problems whatsoever... but any faster and the issues start happening...

It really is driving me crazy.. I will be ordering my new PSU and some other RAM at some point next week, so I will be able to test the mobo and CPU with different RAM to see if I still experience the same issues... if I find no problems with the new RAM, then I will know for sure if the Dominators are faulty...


::edit::

Forgot to say, I have the RAM voltage set to 1.65v... Well, it is 1.64 in the BIOS, and I have also tested with 1.66v, as there isn't an option for 1.65v in the BIOS... however, when I check the voltages using software it shows up 1.65v on the DIMMs when it is at 1.64 or 1.66v in the BIOS...

Also, timings I have tried were the rated 8, 8, 8, 24, and I also tried 9, 9, 9, 26, and 9, 10, 10, 27... but this current Prime run I am doing atm, I just entered the BIOS and loaded optimized defaults, then I enabled XMP profile, and then I raised the QPI/VTT voltage up to 1.355v, and it has ran perfectly for 26 hours, I will leave it on over night again just to see if it is still going tommorow...

Do you think this issue is RAM related, or do you suspect it is another piece of hardware that is causing these problems?

RAM GUY 05-26-2009 07:47 PM

Try setting the QPI Voltage to 1.6 Volts and see if that solves the problem.

Davyboy 05-27-2009 06:32 AM

LoL.. did you actually read my comments RAMGUY???

I already told you that I read a comment by you on another i7 users thread where you recommended upping the QPI voltage to 1.6v, however, Intels say that the QPI voltage shouldn't exceed 1.35v, so I didn't want to chance running the voltage that high, so I raised it one increment in the BIOS from 1.35v to 1.355v, and then started testing for stability with Prime95, and it has been on the exact same run for 37 hours now, and still going strong...

I think by raising the QPI voltage up a notch seems to have done the trick, though I am still unsure whether this is my rig properly stabilized as I have managed to get it stable previously and then the next day I go back to my PC to test it, and it started crashing all the time again!!!

What piece of hardware would cause intermittent crashes like this??

The CPU can be ruled out of the equation as it has already been tested on someone elses rig and it was working just fine...

I personally think that it is the RAM or PSU that is causing these problems, however I will just wait and see how this new RAM works in my rig, if it works fine after installing different RAM these Corsair modules will be getting sent back...

Does Corsair have anywhere in the UK that handles RMA's, or will I just have to deal with the e-tailer I purchased them from?

I would prefer to go direct with Corsair...

xpatar 05-27-2009 09:04 AM

Bumping up a little bit the qpi/dram won't hurt your cpu/ram. Personally I enter ram parameters manually except the qpi/dram which I leave it @ auto, entering os reading through turbo v ,the system applies automatically 1.475v for qpi/dram and my system is rock solid using memtest for 3 passes, intel burn test for 20 passes and prime 95 for 8 hours, the system runs flawlessly without having the need of searching which value I should use in order for my rig to be stable.

Davyboy 05-27-2009 10:16 AM

Which software do you use for monitoring QPI voltages xatar?

Also, my machine has been running large FFT's now for about 40 hours without any issues with the 6GB of RAM installed and running @ 1600MHz using the XMP profile, and my QPI voltage set to 1.355v...

As I said already about Intel recommending no-more than 1.35v, this had made me scared to raise the voltages any higher, but then I saw that DDR3 modules running at 2000MHz required the QPI/VTT voltage to be set to 1.6v,,,

I guess Intel just say no-more than 1.35v to cover themselves...


All times are GMT -4. The time now is 02:19 AM.

Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2019, vBulletin Solutions, Inc.