GA-55m-ud2 stability

[NigelR]

A Media PC built from the components below has proved to be extremely unstable. These components (except PSU which was left in the case) are all new.  They replace an Athlon 64-3200 based MB that I put in the case 4 years ago, that system has been rock solid stable.

Gigabyte GA-P55M-UD2 iP55 Socket 1156 8 channel audio mATX Motherboard
 Intel Core i5 750 2.66GHz Socket LGA1156 8MB L3 Cache Retail Boxed Processor
Corsair 4GB (2x2GB) DDR3 1600MHz/PC3-12800 XMS3 i5 Memory Kit CL9(9-9-9-24) 1.65V,  CMX4GX3M2A1600C9
ASUS GeForce GT 220 1GB DDR2 PCI-E Graphics Card
LiteOn iHOS104 Blu-Ray Reader DVD-Rom Serial ATA
Samsung F3 1TB SATA disk
Seasonic S12 430w PSU
Windows 7 32-bit

I have tried the system with the F3 bios it came with and the latest F4 bios from the Gigabyte web site. The system has never been overclocked. Currently the system has the Intel ‘Turbo’ function disabled (enabled by default) and the turbo function on the memory (enabled by default) disabled, to try and achieve some stability.  The system will randomly lockup during use, not necessarily under heavy CPU load. Currently with the F4 bios if I enable the XMP function that the memory has it will not boot in to windows.

Memtest works correctly for two passes from a bootable CD
Prime95 fails if 4 threads are started (windows 7 multi thread problem?), but appears to work on a single thread.
A simulator I use has been left running for a couple of hours, only to start using the machine to surf the web and have it lockup.  I have tried the latest Nvidia drivers from Nvidia.com and the ones that came on the ASUS install disk.
Cpu-z appears to show that the CPU(2.67) and memory(1333) are running at the correct speeds.


Any suggestions on what to try next?

[R_N_B]

I would suggest two things.  First check and see that each of the power connectors are firmly in plugged in, and that they are all connected. What I have seen with the seasonic PSU is that sometimes the little metal inserts shift out of the plastic plug holders. That results in continuity issues. Please try each and every connector and each and every wire going into the connector.


The second thing is change your PSU and see if that sorts the problem out.


Your system should work just fine as listed, The only thing that you are using that is not new (4 years old) is the PSU. Sadly even the most expensive PSU's degrade over time. 4 Years is about the limit for a budget PSU.

Let us know if you were able to borrow or swap out a PSU please.

[NigelR]

Thankyou for your responce,

I can find no connectors that might cause a problem.

I fiited the PSU back with the AMD64-3200 , nvidia 6800 graphics  and it works fine.

I have ordered a 650w (corsair) PSU, I hope this is enough! I thought these new prcocessors were more efficient!

Nigel

[R_N_B]

Nigel could you do me a favor mate. Take a very high res photo of the underside of your CPU. Or if you have good eyes have a look at the pads under the CPU.  I wonder if all the pins are making contact on your CPU - you will notice that some ( or all) of the pads will have little scratches or indentations on them. If there are some pads that do not have those marks then that might be the cause of your problem. Not saying it is, just offering possible cause. It is easier to see using a high res photo.

I say this because the PSU seems to be working just fine and you are correct the 750 should use less wattage than your old CPU.

[NigelR]

Hello,

Well I changed out the power supply to the Corsair 650w and the results are the same.

Prime 95 crashes (4 threads) almost immediately with hardware errors detected.

After a week of evenings battling with this thing, I think its time to RMA the motherboard.

Nigel

[R_N_B]

Have a look at the underside of your CPU for me please mate. See earlier replay.

[NigelR]

I don't have a suitable camera to take a Hi Res close-up of the processor.


Bios settings below have been my most successful so far....

CPU clock fixed 20X (2.66)
Turbo Boost Tech disabled
Block control enabled at 133
XMP profile (results in 1.65v Dram voltage)
System Memory multiplier set at 10 (1330)
Memory Performance Enhance STD

Results;

Using OCCT V3.1.0
CPU OCCT does 1hr test ok
CPU LINPACK 1Hr test ok

Prime95 runs for about 1.5hrs before failing.

 I'm obviously getting closer!

Nigel

[NigelR]

Seems my success was short lived, the chances of completing the above tests appear to be random, I cannot get OCCT to complete now, but did have one instance of Prime95 running for nearly 2 hours.

[cpharris]

Assuming you have flashed the latest F4 BIOS, go into the BIOS and change the following settings (ignore the XMP settings):

First "Load Optimised defaults" then try:

MB Intelligent Tweaker -> Advanced Memory Settings -> System Memory Multiplier [10.0]
MB Intelligent Tweaker -> Advanced Memory Settings -> DRAM Timing Selectable [Quick]
MB Intelligent Tweaker -> Advanced Memory Settings -> Channel A Timing Settings [Enter]:

Change the following if these are not the default values. Most likely they are, and you can leave them at AUTO:

CAS Latency Time [9]
tRCD [9]
tRP [9]
tRAS [24]

You can also change the following values if the above are still unstable:

tRFC [88]
Command Rate(CMD) [2]

MB Intelligent Tweaker -> Advanced Voltage Settings -> DRAM Voltage [1.640V]


Seeing as you are running at 1333MHz, not 1600MHz, you should in theory be able to run at a faster timings. Not sure why it's failing at stock speeds. In fact, enabling the XMP should work, as the CPU will run at 160x16. Perhaps you need to set the x16 multiplier before enabling the XMP?

Let us know how it goes. Chris.

[NigelR]

Hi Chris,

Thanks for your responce.

Ok F4 Bios and

First "Load Optimised defaults" then try:

MB Intelligent Tweaker -> Advanced Memory Settings -> System Memory Multiplier [10.0]
MB Intelligent Tweaker -> Advanced Memory Settings -> DRAM Timing Selectable [Quick]
MB Intelligent Tweaker -> Advanced Memory Settings -> Channel A Timing Settings [Enter]:

Change the following if these are not the default values. Most likely they are, and you can leave them at AUTO:

CAS Latency Time [9]
tRCD [9]
tRP [9]
tRAS [24]

Will not boot fully into windows.

add...


-tRFC [88]
-Command Rate(CMD) [2]

Boots correctly, runs Prime95 for asbout 5 minutes before it fails .

-Change to 1.64v dram voltage

Runs Prime 95 for nearly 2hrs before it fails

The best I have managed so far is to run XMP and clock back the memory to 1066mhz, this managed to run Prime95 for 9hrs with no failures.

I am at the point of buying new memory, can anybody recommend some that works and can be purchased in the UK?

Thanks

Nigel

[cpharris]

Hey Nigel,

If you read my other posts here (and all over the Internet!) I've been having problems with the P55M-UD2 and OCZ Platinum 1333 Memory. Seems the video problems are down to drivers, but was getting random freezing which was corrupting files and meant I couldn't really use my computer at all...

At rated speed (7-7-7-20, 1.65V, 1333MHz) it was running Prime95 and OCCT for hours but still randomly freezing. Memtest ran for 12 hours with zero errors. Gigabyte tested their own system with same motherboard and OCZ memory and everything worked fine for them, everything on auto (7-7-7-19, 1.5V, 1333). Memory is on the compatibility list...

I have recently purchased new memory: G.Skill Ripjaw 4GB (2x2GB) DDR3 PC3-12800C7 1600MHz Dual Channel Kit (F3-12800CL7D-4GBRH) and installed it this evening. Loaded optimised defaults then set 7-7-7-24, 2T, everything else on Auto. Motherboard sets speed to 1333 and voltage to 1.5V (measures 1.57V in HWMonitor). Ran for a few hours and no freezing at all! Needs some time but looking good so far!

The ram is obviously specced for those timings at 1600, 1.6V. And I haven't tried running at 1600 yet. I basically overspecced the ram so it would be super stable at 1333. Hopefully 1600 (using XMP) will run fine.

I don't know if this motherboard is very fussy, but seems a lot of people with "good" memory are having compatibility issues. Even those with memory off the compatibility list! Now I have to convince OCZ to RMA my old ram. Perhaps it was from a sub-standard batch that shows up in P55 boards or something??

I lost a month trying to fix it through the BIOS, before getting the new ram.

Chris

[cpharris]

The other thing I was going to try if the system still freezes is to rebuild the computer, examining the CPU socket for any bent pins. A lot of people suggest this as a source for instability, but I've never actually read a thread where they actually did have bent pins!! Worth a look at least.

Chris

[NigelR]

 

ITS FIXED!

I have changed out the Corsair CMX4GX3M2A1600C9 memory to G.skill F£-10666CL8D-4GBRM, and it works! This is the first time I have been able to 'Load Optimized Defaults' and have the machine work correctly with the memory at 1333MHz. It has run Prime95 for 7 hours on 4 threads with no errors and passed the OCCT tests.

It is interesting to note that Memtest found no errors with the old memory at any speed, even when run for 10 hours, so this program cannot be completely relied upon to show all memory problems.

Thanks for your help Chris (CPHarris).

I note that there are numerous other threads here that have the Corsair 1600 c9 memory as a common factor!

Nigel

[cpharris]

Good news on the memory!

After days with my computer running fine, it froze a few times in a row  . So I stripped everything out and found one of the motherboard pins had burnt off, and stuck to the CPU!!  Some pictures here: http://img189.imageshack.us/gal.php?g=cpu1z.jpg

Currently the Motherboard and CPU have been sent off for RMA, but it's been about 6 weeks since I bought them and the shop isn't very happy about it. I think I'm going to have to put up a fight to get them replaced!

[Beekeeper]

Currently the Motherboard and CPU have been sent off for RMA, but it's been about 6 weeks since I bought them and the shop isn't very happy about it.

what does it mean the shop is not happy?    they earned money on you so it's their duty to service you now when you have problem