Random reboots when transferring data

[parody]

Firstly, Hello to everyone!

I have a problem, and am reaching the end of my knowledge / patience!

First of all Im by no means a pro system builder but have some experience, having knocked up 30 or so in my time, and have overcome various problems along the way. This one though has me kerfuddled.

This is a new build for a friend. He wanted a relatively kick ass system because he "doesnt like seeing egg timers", but wont actually use it for anything more that syncing his palm, internet, email - the normal stuff.

With low noise and high speed a priority here is the spec:

Gigabyte GA-EP45T-DS3R
Intel Quad Q9550 CPU
4Gb OCZ DDR3 Platinum 10666
600w OCZ PSU StealthXStream
Nvidia 7600GT with Zalman VF900 silent fan cooler
Zalman CNPS7500-CU CPU Cooler (supplied Zalman thermal paste)
74Gb WD Velociraptor 10k rpm HD
2 x 250Gb Seagate 7200rpm in RAID0 (using motherboard controller)
Sony DVDRW, card reader, rear esata ports
1 x 120mm fan top of case exit air, 2x80mm fan front by HD pulling air, 1x80mm fan rear pulling air. All fans inc CPU controlled by front mounted fan controller.

OK so here it is...

1) Built system, installed Vista x64. No problems encountered. System was very fast and very quiet, idle temps in the low 40's which was acceptable given the relative silence.

2) Friend uses PC and experience frequent random hanging, normally after screen saver has been on. After reboot vista reported error with Gigabyte DynamincEnergySaver software - apparently a known incompatibility, so uninstalled it.

3) This didnt resolve the issue. Took the PC back and ran some tests. Was able to reliably recreate hanging by running OCCT memory test. After some investigation it appers the mobo had set the default ram voltage to 1.5v. I upped it to 1.8v (as per ram sticker) and set timings recommended in OCCT. OCCT lasted longer, but still hung. Upped voltage to 1.96c (manufacturer within warrenty maximum). OCCT then ran stable for hours. Also tried prime95 for 4hrs, memtest86+ for a whole day, and the dreaded intel burn in for 20 passes successfully. Temps under full load of 20 intel passes were 78 degrees. Quite hot, but completed OK, and a situation which would never be encountered in his day to day use. As I understand the Q9550 are good for 100 before thermal throttling kicks in. And this was all with fans on minimum. On max they were low 60s from memory.

4) Thinking the system was now stable, and rebuilt with XP 32bit as he didnt like vista and it wasnt compatible with his palm, I gave it back to him. (tests done after XP install)

5) Next day, he starts to copy all his many gigabytes of data over to the PC from external drives, and now he is experiencing random reboots - seemingly when he leaves it copying data. After one reboot XP asked him to download and run the microsoft memory tester. He is doing that this weekend but after my tests I doubt it is memory?

Next steps... have asked him to run all fans on max while copying data to see if its an overheating mobo chipset, and d/l and run seagate and WD HD test tools.

Any more insight from you uber clever hardware type dudes?! Am i missing something painfully obvious? This was a bit of a bleeding edge build for me - Im more at home with AMD dual core and DDR2 ram - it all seemed to just work for them! Never had problems like this before.

All replied greatly appreciated and thanked in advance.

 

[blackwidow_rsa]

sounds like a memory issue. make sure the ram you have is compatible with the mobo (check manufacturers site), if they are either test on another system or RMA it and get new ones

 

[Proximon]

I haven't been real happy with OCZ for reasons just like this.

Max Tcase is 75C I think for that processor...

http://www.tomshardware.com/forum/221745-29-core-quad-temperature-guide

If you decide to switch out coolers I suggest you consult this list:
http://www.frostytech.com/top5heatsinks.cfm

Other possibilities would be power fluctuations at the wall, or underpowered USB ports.

 

[parody]

Hi Proximon thanks for the

Im not too worried about the CPU temp - 78 was the max core using realtemp under extreme runs if the Intel Burn In. And he doesnt play games, or anyhting intensive what so ever. Given it was stable at those temps - I dont think this is the issue when just transferring files. Would you say that is a logical assumption?

Are you referring to OCZ PSU or ram?

The external HD's are 3.5in so are mains powered.

Thanks again.

 

[parody]

Hi blackwidow_rsa

You would still say ram despite successful memory testing with memtest and OCCT etc?

 

[Proximon]

While you are right about the temp probably being not a problem, it's far from where it should be. I would say it needs fixing, but it's not the source of your issues.

I was referring to the RAM. The OCZ SXS 600W should be adequate. It's what I would term a good low budget choice.

I see far too many issues here regarding OCZ RAM that doesn't perform at spec. I may be a bit oversensitive to it though, as I've been bitten myself.

It's no exaggeration however that half the issues we see here with at least semi-working builds turn out to be RAM issues. Download and run CPU-Z and check the SPD for each stick. Make sure it's what it's supposed to be, and that your motherboard settings correspond with one of the sets on the SPD tab.

 

[parody]

ram timing is set to the CPUZ / OCCT recomendation - they were the same. Shame I dont have any DDR3 motherboards or other DDR3 sticks to try it with. How easy is it to return ram when you dont ultimately have proof its faulty? I got it from ebuyer.

Also, as for CPU temp, surely the Zalman is doing a better job than the stock HSF would have done? Dont want to start a CPU temp debate... Ive seen some of those threads!!

 

[Proximon]

If you go to the OCZ forums and work with them a bit, they'll give you an RMA number.