Strange random shut-downs on new build

stefk005

stefk005

stefk005

Associate
Joined
12 Sep 2008
Posts
397
Location
London
This possibly should be under the motherboard thread, but since it could be something else, thought I'd better post here...

Just built a system with a Gigabyte Z87X-UD4H and i7-4770k. Everything installed fine and was going well till I attempted some overclocking. The 4.3ghz mobo preset booted up fine, but was running too hot, so I tried a lower OC. That wasn't liked, and ended up in a re-boot cycle. Eventually got that sorted by flashing to the latest F8 bios. However since then, the computer has randomly switched off 3 times (and then started itself up) whilst doing fairly normal things. On the first 2 occasions, I was installing something from disc, which had me thinking it might be related to the optical drive. However the 3rd time it switched off whilst copying files from a usb drive and browsing the web. There's no indication in the windows events log; the mobo debug code goes to "AO" (all normal, I think), and everything else appears ok.

Really stumped here, and somewhat annoyed as there's nothing obvious.

Specs are as follows:
Corsair AX860 psu
Gigabyte Z87X-UD4H
Intel i7-4770K cooled by Phanteks enormous CPU heatsink (temps are all fine)
Teamgroup Vulcan 2x4GB 2400Mhz RAM (using XMP profile)
Samsung 840 250Gb boot drive + WD green 2Tb storage drive
MSI 290 Gaming
LG blu-ray drive
Creative Soundblaster Z soundcard

Any suggestions appreciated! Thanks!
 
ThaReaperGuy said:
Reset the CMOS (take the battery out of the motherboard for 10 mins).

Will remove all the overclock settings back to factory.
Click to expand...

Sorry, should have made that clearer: I'd done that already to sort the reboot cycle. Since flashing to a newer BIOS I've kept the CPU at stock. Problem is since then.
 
I reset the memory to BIOS defaults - 1333Mhz, etc - and it's run most of yesterday and today without a hitch. I've just set the memory manually to it's proper spec, and running a load of HCI memtest to see if it will accept that. Maybe the XMP profile was the problem?
 
Well, set it to the advertised 2400mhz, 10-11-11-31, 1.65v and HCI memtest brought it crashing down - PC had restarted when I went to check on it and was at the option to start windows normally or safely. I've seen a few people claiming to be running Teamgroup vulcan memory at this speed on this board with no probs. Could the memory be faulty?
 
stefk005 said:
Well, set it to the advertised 2400mhz, 10-11-11-31, 1.65v and HCI memtest brought it crashing down - PC had restarted when I went to check on it and was at the option to start windows normally or safely. I've seen a few people claiming to be running Teamgroup vulcan memory at this speed on this board with no probs. Could the memory be faulty?
Click to expand...

Yes. It should work on its proper settings.

Could also run a sfc /scannow in Command Prompt as Admin, to make sure there are no errors in the Windows installation, as bad RAM can cause corrupt files (even though they may not be vital).
 
Ok, this is getting strange. I have just been testing the memory on a different system with a Gigabyte Z77X-UP4. I've set it to its full spec - 2400mhz, etc - and its currently got through at least 250-500% coverage in HCI memtest, which had the memory in the other system crashing. Why is it working in the Z77X-UP$ board and not in the Z87X-UD4H?! Anyone else having issues with TeamGroup memory and the Z87X-UD4H? Any suggestions as to what to try?
 
I set it manually in the bios to the 1.65v specified on both boards. Do you mean there might be some voltage droop with the z87x?
 
No, will give it a shot. Do you think it's worth increasing the voltage slightly on the Z87X, in case it's not quite supplying the right amount?
 
for it to work at 2400mhz dram must be 1.65v

you might also need a touch more memory controller voltage vtt,not sure what its called on haswell tho
 
Ok, sorted now. First I tried adding 0.03 extra volts on the RAM (so 1.68V) - that didn't work. Then added 0.021v to the system agent and 0.02v to the analog and digital i/o volts (which is the haswell memory controller), and that seems to have stabilised it. At least no crashes on 300% coverage with memtest, which was impossible before. Slightly disappointed that this board needed so much tweaking to get something so simple going, especially when the Z77 board managed it without any fuss!

Do we have a GBT rep here who I can pass this on to?
 
You must log in or register to reply here.
Back
Top Bottom