timby_65

Distinguished
Nov 18, 2009
22
0
18,510
0
I recently built a new system with a GA-P55-UD3 motherboard and am getting random reboots (around 20 so far?). I've tried a number of things and have had no luck resolving this issue. I am in urgent need of help, as this is a work computer and I need to get back to work!

I've been trying to find a pattern, but nothing concrete. Here's what I have so far:
- Has only rebooted while in windows.
- Once rebooting starts, it happens more and more frequently (to the point where I don't even get to the login screen). Leaving the computer off for a while seems to reduce the frequency of reboots.
- It has happened twice while using Virtual PC.

Here are my current specs:
GA-P55-UD3R rev. 1.0
Intel i7-860
4x2GB OCZ3G1600LV4GK
Corsair VX550W
Radeon HD 4350
Western Digital 1TB
Windows 7 64-bit

Here's what I've tried to do so far:
- I originally bought the machine with 2x2GB Corsair CMX4GX3M2A1600C9.
- Installed Windows 7 64-bit. It rebooted a few times. During one boot up, it found errors on disk and fixed them. I could not get to the login screen after this.
- Updated BIOS to F3 and loaded optimized defaults. Didn't help.
- Reinstalled Windows 7 64-bit and all drivers (no Catalyst, though). Ran it on and off during the weekend at home. Ran Prime95 for 15 minutes with no issues. Watch an HD video with no issues.
- Brought it in to work on Monday. It worked until about noon, when it rebooted while in Google Chrome. Rebooted at 4:35 while not being used (I left at 4:30).
- Tuesday I was using it for a couple of hours. It rebooted while in Virutal PC. About 5 minutes later it rebooted while browsing in Google Chrome. Tried fail-safe, did not help. Tried adjusting RAM voltage to 1.64 (1.65 was not listed?). Didn't help. Could not get to login screen at this point. Left it for 10 minutes and tried again, but still couldn't get to the login screen. Left it for an hour and used it for 5 minutes and it rebooted. Left it for a little while longer and it didn't reboot for the rest of the day.
- Exchanged RAM for supported OCZ model. Installed 4 GB. Loaded Optimized defaults. Change voltage to 1.64. Ran a video in a loop for an hour, no problems. Ran the Windows 7 memory diagnostics in highest setting (2 passes). Ran from about 8pm until about 2:30am. It didn't find any problems. Booted up into windows autoamtically and ran until about 7am until I shut it off.
- It worked today from about 9am until 2:00pm or so with lots of usage. Rebooted while in Virtual PC.
- I found there was a new video card driver released yesterday. I installed it.
- I've checked the temperature a number of times, but it is always under 40c for both CPU and MB (something like 39c and 34c). It doesn't seem to fluctuate much.

I noticed after installing the supported OCR RAM, it listed in BIOS as 7-7-7-16 at 1066MHz instead of 8-8-8-24 at 1600MHz. I just ran Speccy and it's showing 7-7-7-19 at 533MHz(???). I'm not sure how to set this to 8-8-8-24 at 1600MHz.

Any help would be greatly appreciated.
 

Fatbird

Distinguished
Oct 11, 2009
69
0
18,640
4
System reboots itself, means that system suddenly resets like pressing the reset button?
Try remove all the power cables to the board and re-plug it.
Remove the memory module and re-sit it in series, you bought two OCZ 4G kits.
Look carefully to the memory S/N, they are paired, put the same S/N in same channel (means in different DIMM color).
Check the voltage info on the OCZ memory, adjust the BIOS according it, raise a little bit MCH core voltage and drop a little bit MCH reference voltage.
 

bilbat

Splendid
Is neither "MCH core", nor "MCH reference" on a P55-UD3R, unless it was added to one of the three dozen or so recent BIOS revs... What used to be called an MCH (memory controller hub - northbridge) on FSB architecture Intels is now an IOH (mistakenly labeled MCH/ICH on most GBs) on the new platforms (1156/1366s)...

You don't even want the RAM at 1600 - there's no real-world advantage to it (http://www.tomshardware.com/reviews/memory-scaling-i7,2325-5.html - aimed at 1366s, but the concept's the same...), it just makes things hotter and harder to get stable, and it's officially unsupported by Intel; you want the RAM at the lowest latency possible at 1333, which is the highest speed supported on the 1156 platform; theoretically, your RAM should be OK, assuming what you're reading is accurate: 8-8-8-24@1600 should run 7-7-7-20@1333, or 6-6-6-16@1066 (but you may have bad sub-timings); the reason that you're seeing 533 in some utilities is that it's DDR, which, recall, stands for double data rate! Transfers information on both the rising and falling clock edges; i.e., 1066 nominal RAM is actually clocked at 533 MHz...

First thing to check - source of lots of weird problems, is socket pins - takes a flashlight and a magnifying glass - Intel 'procedural' here:
http://www.intel.com/support/processors/sb/CS-030850.htm

Next is to run MemTest86+ here:
http://www.memtest.org/download/4.00/memtest86+-4.00.iso.zip
first, on your whole setup, preferably overnight; and then, assuming errors, to run it on one stick at a time, at least a whole pass, to 'qualify' the memory; as the major 'tool' I've previously used to come up with 'calculated' memory settings (especially sub-timings) is 'defunct', so far, on P55s, I'm back to timing by timing hand research and calculation, so it's time consuming - last one I did worked, but took three days of spare time - and am currently working on another; hoping I can get it down to a little more science, and a little less educated guesswork, and find a way to 'churn them out'; meantime, take heart, and read a bit here - people are getting these things to run, and in a lot of cases, to run spectacularly! :
http://www.xtremesystems.org/forums/showthread.php?t=234529


 

timby_65

Distinguished
Nov 18, 2009
22
0
18,510
0
Thanks to both for your replies.
System reboots itself, means that system suddenly resets like pressing the reset button?
Yes.

Try remove all the power cables to the board and re-plug it.
Remove the memory module and re-sit it in series, you bought two OCZ 4G kits.
Done.

Look carefully to the memory S/N, they are paired, put the same S/N in same channel (means in different DIMM color).
It is correct.

Check the voltage info on the OCZ memory, adjust the BIOS according it, raise a little bit MCH core voltage and drop a little bit MCH reference voltage.
I currently have everything set back to auto. If you're interested, here's what the bios has set it to:
■CPU Vcore 1.20000V
■CPI/Vtt Voltage 1.100V
■PCH Core 1.050V
■DRAM Voltage 1.500V
■DRAM Termination 0.750V
■Ch-A Data VRef 0.750V
■Ch-B Data VRef 0.750V
■Ch-A Address VRef 0.750V
■Ch-B Address VRef 0.750V

I originally had the DRAM Voltage set to 1.640V (1.650V was not an available option), but it didn't seem to make a difference. Should I change this or any of the other options?

you want the RAM at the lowest latency possible at 1333, which is the highest speed supported on the 1156 platform
I guess I should have done my research on this one. I was looking at the NewEgg comments for the i7-860 proc and it seemed most were buying 1600. I guess we should have gotten 1333?

theoretically, your RAM should be OK, assuming what you're reading is accurate: 8-8-8-24@1600 should run 7-7-7-20@1333, or 6-6-6-16@1066 (but you may have bad sub-timings)
The bios is actually reading 7-7-7-16@1066. Is there a way I can change this to 7-7-7-20@1333?

the reason that you're seeing 533 in some utilities is that it's DDR, which, recall, stands for double data rate! Transfers information on both the rising and falling clock edges; i.e., 1066 nominal RAM is actually clocked at 533 MHz...
I didn't even know what DDR stood for. :whistle:

First thing to check - source of lots of weird problems, is socket pins - takes a flashlight and a magnifying glass - Intel 'procedural' here:
I skimmed through this before installing. However, I did not look too closely (with a magnifying glass). I'm a little concerned about removing the CPU and putting it back on. I guess if I'm careful, I have nothing to worry about? Also, I'll need to purchase some TIM before doing this, right?

Next is to run MemTest86+ here:
I guess the MS one isn't good enough? ;) I'll run it tonight and see what it says.

as the major 'tool' I've previously used to come up with 'calculated' memory settings (especially sub-timings) is 'defunct', so far, on P55s, I'm back to timing by timing hand research and calculation, so it's time consuming - last one I did worked, but took three days of spare time - and am currently working on another; hoping I can get it down to a little more science, and a little less educated guesswork, and find a way to 'churn them out
Thanks for all your hard work. I'm sure everyone appreciates it. If you can think of something simple I can try, please let me know.

meantime, take heart, and read a bit here - people are getting these things to run, and in a lot of cases, to run spectacularly!
If this was a home PC, I wouldn't be worried at all. However, it's my work PC and I'm a little stressed right now. I just spoke to my boss and he's not too happy about this. I've built a few machines before with no issues and it was my recommendation to do this rather than buying a pre-built one from Lenovo or HP. It sounds like I'm going to have to pay for this one out of my pocket (and use it at home or something) and get a pre-built one if I don't get it fixed soon (like tomorrow). :(

It doesn't look like my boss wants me to check the pins myself. Maybe I'll bring it in to the tech support dept of the store where I bought the stuff and have them take a look.

My computer just restarted again about an hour ago, so it doesn't look like the updated Radeon drivers helped. However, the frequency of this issue has seemed to reduce since switching to the OCZ memory (once yesterday and once today...so far).

I have an older nvidia based card lying around here...I'm going to try that and see if that makes a difference.

Any other recommendations?

Thanks again for all your help. I really need it. :??:
 
One other thought, since you need to get this fixed quickly. When you built the PC, you did not use any paper washers between the standoff/mobo or standoff screw/mobo (or insulating standoffs) in this build, correct? Gigabyte boards are very sensitive to this and will have random reboots due to insufficient grounding if you insulate the standoffs in any way.
 
Yes. I have a Gigabyte board (with the Phenom II in my sig). I used the paper washers at first, just because they were there. I would get random restarts when plugging in USBs, when I walked near it, randomly, etc. Once I took them off, not one restart since. They supply the washers for a different kind of board. Gigabyte boards need the case ground through the standoffs, otherwise they will be unstable or ground through the wrong things (USBs) causing restarts.
 

timby_65

Distinguished
Nov 18, 2009
22
0
18,510
0
I have removed the paper washers. Thanks for the suggestion.

I have also replaced the ATI card with an nvidia card. No reboots yet, but not enough time has passed yet.

I will be running MemTest86+ shortly.
 
Yeah, I know what you mean. I know several times when I had the issue I thought I had fixed it, only for it to come back randomly a couple days later. I wasn't finally satisfied i had fixed it until it didn't reboot for a week or two.
 

ekoostik

Distinguished
Sep 9, 2009
1,327
0
19,460
88

If this turns out to be the fix that's a great call. Only someone who's been down that road before could've pulled that out of the air. What ships with paper washers? Is it the case?
 
It was the case. Its not their fault though, just mine for not reading more into it. Apparently, some older boards (or maybe new ones too, not sure) needed insulation on some of the standoffs. That is not true at least for Gigabyte boards though, as they are specifically designed to have the standoffs conduct (and need it, at least in the case of my board). If you look closely at the boards, there are even solder beads to ensure good conduction.
 

timby_65

Distinguished
Nov 18, 2009
22
0
18,510
0
The paper washers did come with the case, not the motherboard. I also noticed the solder beads around the wholes while I was taking them off yesterday. I really hope this was the cause of the problem. In any case, I think this is a great tip.

I ran MemTest86+ for hours last night. 4+ passes with 0 errors.

I currently have my DRAM Voltage set to auto, which is coming up as 1.500V. The memory should be 1.650V. Should I change this setting to 1.640V? Should I change anything else? Or should it be stable with the current settings?

BTW, my boss told me that one more restart and that's the end of the line with this machine. I really hope this works!
 

papatte

Distinguished
Nov 15, 2009
10
0
18,510
0


Hi timby_65,

It seems that we have quite the same configuration and similar problems. I've also experienced frequent reboots on my machine and I've still found no workaround. The detailed description you made at the beginning of the thread was very interesting.
My config is: GA-P55-UD3R F3n + Core i7 860 + 8 GB GSkill DDR3-1600 + HD-5770

I have the feeling that the frequent reboots I have are caused by my Sapphire HD5770, but that's quite difficult to be 100% sure. However, if your config remains stable with a Nvidia card, it would prove that the video card is definitely involved.

My symptoms:
* The first installation of WIndows (XP 32bits for the moment) went without problem ... until I install Sapphire's Radeon drivers. Then, after a while, ugly red squares appeared all over my screen and then a blue screen of death appeared. After that, the PC wasn't stable for more than 2 minutes and kept rebooting. I tried to install 8.66 drivers, 9.10 drivers, 9.11 beta drivers with the same result. The PC behaved exactly the same way with Ubuntu when I activated ATI Proprietary drivers.
* Then, I noticed by chance that, that if my HD5770 card was hot enough, the PC didn't crash anymore!! I've tested this feature for 2 days and as soon as the GPU Temp does not go under 38°C (98°F), the PC keeps running without any problem (even for hours). I've tested it on Ubuntu and Windows. On Windows, to prevent the system from crashing, I've tried to run FurMark3D continuously (but the CG gets hot) or to change the CG parameters thanks to ATI Tray Tool (reducing fan speed works for example). On Ubuntu, I still haven't found a perfect solution so far since HD5770 is still not supported by the latest ATI drivers -> Therefore, it's impossible to monoitor the temperature or to change the fan speed for example... However, running fgl+glxgears prevents Linux from crashing. I think that a long-term solution would consist in modifying HD5770's bios.

I suggest that you have a close look at the temperature of your Radeon graphics card (with a tool like hwmonitor). Maybe, this can help. You can also have a look to the thread http://www.tomshardware.co.uk/forum/page-55024_12_50.html

Papatte
 
That's interesting papatte. Good job investigating. It may mean you card is defective though, as I have never seen a card that thought 38C was too cold.

Timby, passing memtest is a good start. To be more certain, a program like prime95 would be good as it stresses CPU and memory.

As for the ram, I would leave it at Auto. The reason it (appears at least) runs fine is because of this:

When ram manufacturers set the default speed, they always go back to the 1066 (533x2) JEDEC specification, as all certified mobos should be able to boot that without issue. Thus you see 533 (1066) and 1.5V and CAS7, a JEDEC spec. There is nothing wrong with that setting, and since you are focusing on stability right now, I would not adjust it. You could play with it later once you are not under so much pressure, but the benefit of setting it to its rated higher speed is fairly small (in real life apps) and the risk of inducing a reboot (by wrong settings or poor ram) I think is too high for your situation.
 

bilbat

Splendid
Timby_65;

Seeing as how you are in 'desperate straights', here's what I'd try: Once again, take out two sticks, disable XMP, and do a "Load Optimized" with the remainder; save, exit, reboot; enter BIOS and set:

CPU Enhanced Halt (C1E) disabled...
C3/C6/C7 State Support disabled...
CPU EIST Function disabled...
Performance Enhance to standard
QPI/Vtt Voltage to 1.250V
PCH Core 1.200V
DRAM Voltage 1.650V

For your RAM timings:
Under standard, raise CAS, tRCD, tRP by one count; raise tRAS 10%, rounded to the next integer;
under advanced, raise all 10%, again, rounded up to the next integer, except tWTP, which you want to round up to the next odd integer...

'turnarounds' should be OK at the 'Load Optimized' settings

Save exit, reboot; power down, add the other two sticks (as a fall back position, you can always run with two DIMMs 'till you can get the other two going - not likely to 'see' a lot of difference, depending on your workload...)
 

timby_65

Distinguished
Nov 18, 2009
22
0
18,510
0
Thanks to all for your replies, I appreciate the help.

Unfortunately, the machine started rebooting with great frequency on Friday. The reboots were a little different, though, as I was getting blue screens before it shutdown, where before it would just act like the reset button was pressed. The error started with MEMORY_, but I forget what the second word was.

I bought a new HP machine an hour later and have had no issues with it. The problem machine is now my home PC (along with the cost).

I went to the store I purchased the parts from and they were nice enough to offer to troubleshoot it for me. Yesterday I called and asked for an update. They said that they ran MemTest86+ again and it reported errors (although they're not sure if they caused them by changing some settings first). They said they adjusted the timings and ran MemTest86+ again, and it ran without errors. However, the machine rebooted later that day. They then exchanged the OCZ RAM for GSkill and haven't had a reboot since. It was almost a full day at that point.

I'm going to let them keep it a day or two longer, just to make sure the issue has been resolved. I'll post an update once I get the machine back.
 

RyanEricW

Distinguished
Dec 1, 2009
2
0
18,510
0
I too have a problem with random rebooting using windows 7. The problem is, I have no dump files, and never get a BSOD even when I hit F8 before bootup and disable automatic reboot upon system failure.

I have an AMD setup.
Mobo: GA-MA790FX-DS5
CPU: Phenom x4 9850 2.5ghz?
RAM: 4 x OCZ 1066 2gb
PSU: Corsair 750w
VID: Powercolor 4850 2gb
HDDS:
1 x 1TB WDD Black Caviar (Storage)
1 x 160GB Seagate (WinV x64, WinXP x86)
1 x 160GB Seagate (Ubuntu / random unpartitioned free space)
1 x 300GB Seagate (Misc / Win7 x64)

Weird thing is, I never have a problem running vista 64, or windows xp. I only have had this random restart problem when using Windows 7, and it seems that we have similar hardware "1tb hdd".
Perhaps the hdd drivers?
I've had nothing but problems with win7 where vista has treated me better since I've been using it starting 2 years ago.
 

ekoostik

Distinguished
Sep 9, 2009
1,327
0
19,460
88
Next time it crashes, check in Computer Management, under System Tools - Event Viewer - Custom Views - Administartive Events. Look for warning or error messages written at the time of the crash that could point to what is causing the problem.

Others have found updating or rolling back drivers on certain hardware have resolved the issue. http://www.sevenforums.com/crashes-debugging/6541-constant-bsod.html

You can also check in Computer Management under System Tools - Device Manager. Expand all the listed devices and make sure none have a warning or error icon displayed on them.
 

timby_65

Distinguished
Nov 18, 2009
22
0
18,510
0
It appears that my issue is resolved, as the machine has not rebooted since 2009-11-24 (about a week ago) when the OCZ RAM was exchanged for GSkill.

To summarize, I had random reboots with the following specs:
GA-P55-UD3R rev. 1.0
Intel i7-860
4x2GB OCZ3G1600LV4GK OR 2x2GB Corsair CMX4GX3M2A1600C9
Corsair VX550W
Radeon HD 4350 OR NVIDIA 8500GT
Western Digital 1TB
Windows 7 64-bit

Replacing the RAM with 2x2GB GSkill Ripjaws F3-12800CL8D-4GBRM seems to resolve the issue.

I will be replacing the NVIDIA 8500GT with the Radeon HD 4350 card on the weekend and will update if I have any further problems.

Thanks again to all who have tried to help.
 

RyanEricW

Distinguished
Dec 1, 2009
2
0
18,510
0
Glad to see your problem has been resolved!
I can't see my OCZ RAM being faulty.. as it has been 100% stable with vista 64 and xp even playing the most demanding games of today. (GTA 4)
 

timby_65

Distinguished
Nov 18, 2009
22
0
18,510
0
I edited my reply and entered the RAM model that has resolved the problem (2x2GB GSkill Ripjaws F3-12800CL8D-4GBRM).



I actually stumbled upon your thread before posting mine and noticed this. I was really hopeful when I installed the nvidia card, but it didn't help.

I did want to mention that I had a similar issue with the video card in my Xbox 360. I had the RROD issue. Before this happened, the Xbox was locking up very frequently. I couldn't even use it for more than a few minutes before it would lock up on me, even if I left it for a couple of days. I eventually got the RROD. I intentionally overheated it to reset the RROD status and was able use it for hours without any lockups. However, a couple of days later I tried to use it again and it locked up after 20 seconds. I did some research and the theory was that when the board overheated, it warped enough to temporarily correct the problem. I'm wondering if your board has the same type of problem? If you can get your hands on another video card at least temporarily, I think you should try it out.



Originally, I also did not have any dump files and I didn't get the BSOD when the computer restarted. I only started getting the dump files and BSOD on the last day before I brought it in to the service department.



I originally checked the event viewer logs, but there were no errors or warnings (other than the one that said the computer shut down unexpectedly). I had no warnings or errors for any devices in the device manager.



If I'm going to ask for help on a forum, the least I can do is mention what the resolution was. :)

Thanks again to all for your assistance.
 

ASK THE COMMUNITY