D0 I920 and Gigabyte EX58 UD5 issues.

[RadioheadR6]

Certainly sounds to me like a problem with the DIMM slot, or the soldering for that pesky slot. I would suggest sending the lot off to Gigabyte for testing, and making a point to say that you wiggled the memory in the slot and found that it failed / worked normally, and also stress the amount of time you have spent carrying out testing...

I'm not impressed with the service you recieved from the supplier... Not good!... Imagine if it were someone with less knowledge than yourself - At least you can challenge the tech people at that place!...

Will continue to follow this and hope that you get it sorted soon...

 

[Plec]

I'm still really confused as to why the DIMM channel worked fine for the tech who tested the mobo, yet when I get the mobo back and test it out, it was EXACTLY the same as it was before I RMA'd it

When the guy at the e-tailor tested it he either had: more compatible sticks, memory with slightly thicker contacts or was just plain lucky.

It’s got to be one of the above or a mixture of all 3. As you have tried different memory and changed all other components bar the CPU.

I had to wiggle the module gently in order to make it work again... that is screwed up if you ask me... .

It is screwed up - plus it’s only a matter of time before it buckles again as the slot is clearly temperamental. You’ve got to decide: whether you can live with it, knowing you can wiggle the slot to revive it, until the slot completely dies. Or, whether, you ‘chance your luck’ and return the MB, along with memory and CPU, and hope that Gigabyte do a thorough test.

Unfortunately, the reality is, if they install all your components and it passes their tests at stock and then double-check the socket with their own memory and achieve the same results – it’s unlikely they’ll pursue it further.

So with the above in mind:

About the overclocking, the whole reason I bought this rig was so that I could overclock it like most other people are doing with their i7 rigs!!

Yeah, i sympathise Dave and i can appreciate your frustration – but you need to establish how often this problem is occurring and you need to rule out an unstable clock as a possible cause. A complete mind-**** for you i know... :/

I would recommend that you run the system at stock and stress test it daily over the next week and see how often it buckles – if at all. You need to work out 'the average/mean' as DATA will be your best defence in getting a replacement MB – and it will also enable you to rma with confidence if the problem occurs frequently.

It may be that you don’t want the grief – and i wouldn't blame you if that is the case. You have a long guarantee so hopefully the slot will 'give out' completely soon and save you any ambiguity regarding the return.

Once again good luck with it – at least you’ve finally identified the problem – scant consolation but at least you don’t have to spend hrs tweaking and doubting your settings anymore.

Good luck with it DavyBoy – i hope the slot completely buckles for you soon!

 

[DavyBoy]

Well, I have managed to get my rig stable with it clocked to 3GHz and my memory at its rated 1600MHz, I left Prime95 Blend running overnight and it was still going when I checked in the morning, I then let it run through 20 passes of IBT at maximum stress setting, 1 hour of OCCT, and also I let 3d Mark Vantage run through 5 cycles, and it has been absolutely perfect...

I have had to play with all of the main voltages in the BIOS to get it stable though, and it took me almost all of yesterday to find this, but as I said it has been rock solid since before tea time yesterday... Though I am going to keep stress testing over the course of the next week or so to see if it buckles on me again, because I got it stable previously before I sent the mobo and CPU off for testing, and then the next day I would turn it on with the exact same BIOS settings and it would start crashing all over the place...

So I am going to leave it at 3GHz for a week or two and see if it crashes at all...

I read on another thread here today that there is a second revision of the EX58 mobos from Gigabyte being released in the next month or so, they look pretty much identical to the current lineup, except they have got 24 phase power design, and 4 x SATAIII 6GB/s, if my mobo does decide to die on me after these new mobos get released, do you think they would send me out one of these new mobos to replace my existing one???

Also, I wanted to see if anyone could answer a question that I have?

I have got my system built up outside of the case right now, and I have noticed that my idle temps are a lot higher than what they were when it was built inside the case, but my load temps seemed to have dropped.. does anyone have an explanation for this??

My idle temps used to be around 34c, and loads were around 69c, but now my load temps are sitting at 41c, and loads are around 63c...

At first I thought it was due to the hot weather, but then wouldn't my load temps also be affected by the weather??

I would have thought that temps would have been worse when the system was built inside a case??

:edit:

I forgot to say, is it possible that my CPU has a problem with the 20x multiplier, I read about this somewhere else online, but I didn't fully understand what I was reading, anyway, I have set my Bclk to 160, and my CPU multi down to 19x to get me 3.04GHz with 1.12v Vcore, and it has been perfect ever since, although I did have to mess around with the IOH Core, QPI/Vtt, CPU PLL, QPI PLL and DRAM voltages to get it stable...

 

[Plec]

I have got my system built up outside of the case right now, and I have noticed that my idle temps are a lot higher than what they were when it was built inside the case, but my load temps seemed to have dropped.. does anyone have an explanation for this??

It's been so hot recently that the ambient temp is probably hotter outside of the case because it’s not getting the benefit of the case fans pulling cool air in and blowing the hot air out. Outside of the case it’s hot and there's no breeze – so at the moment it’s probably better off inside a well ventilated case for temp purposes. (unusual i know - but it's boiling here!)

The load temps could just mean that the HS has made slightly better contact with the CPU with your recent install.

 

[Plec]

Soz, missed your edit (and didn't edit mine :/):

I forgot to say, is it possible that my CPU has a problem with the 20x multiplier, I read about this somewhere else online, but I didn't fully understand what I was reading, anyway, I have set my Bclk to 160, and my CPU multi down to 19x to get me 3.04GHz with 1.12v Vcore, and it has been perfect ever since, although I did have to mess around with the IOH Core, QPI/Vtt, CPU PLL, QPI PLL and DRAM voltages to get it stable...

I haven’t heard of the above but it doesn't surprise me - with my E5200 I’ve found that it’s happier using a whole number rather than using a fraction. e.g. I use 12 rather than the 12.5 multiplier. (I haven’t really questioned it – I just know after a lot of tweaking it seems to be the stabler option.)

There could be numerous factors why your confiuration prefers 19 to that of 20 (it seems to develop new characteristics on a whim – it’s possessed!) – but now that you’ve found the balance I would certainly stick with it.

You must be the leading authority on configuring that particular MB now!

Good luck with the testing.

 

[DavyBoy]

Hmmm, I spoke to soon...

My rig is still playing up.. I decided to try running IBT again for 20 passes, but it reset itself after only 2 passes...

I'm now back to stock settings again, and am running IBT, it seems to be fine so far, but I will leave it at stock settings for a week or so and test it every single day to see if it buckles or not...

This rig is a total PITA!!!

 

[Plec]

I know it's galling for you - but testing it at stock for DATA gathering for rma purposes makes sense. Otherwise you'll doubt the actual cause when it crashes.

Was the system recognising all 3 sticks when it buckled? Even if it was the slot could have just lost integrity for a moment causing the the system to crash.

At least you getting a log for rma purposes...

 

[DavyBoy]

Yeah, all 3 sticks are now being recognized, I check the POST screen everytime I reboot (which is A LOT!!!) and it shows the full 6GB..

It is now at stock settings, and I just checked it to see whether it was still running IBT, and it had completely froze...

I think I am going to give the latest beta BIOS a go just to see if that makes any difference....

 

[Plec]

I think I am going to give the latest beta BIOS a go just to see if that makes any difference....

I admire your dogged determination Dave, but i personally wouldn't bother - when taking in to account all your other troubleshooting and results.

If the slot is flaky a different BIOS isn't going to fix it and you're just going to start adding stress to an already stressful situation.

I would kick back and carry on stress testing and logging how many times it buckles. Let the computer take the load - all you have to do is reset the test and make a note of how and when it crashes. (I wouldn’t add a BETA BIOS to the mix – it will be something else for you to doubt - and add yet more stress if it fails.)

Your call of course and I understand totally why you’re trying so hard to resolve this – but you appear to be in a situation where you just have to keep on stress testing and letting time pass. The errors will hopefully occur frequently enough to give you confidence with the rma to gigabyte.

This, hopefully, will be the last phase of an incredibly difficult process for you.

 

[DavyBoy]

You are right Plec, I never bothered using the beta BIOS...

It is crashing all the time with IBT under stock settings, but I just tested Prime95 Blend and it ran for 1 hour before I stopped it manually..

It managed to pass IBT several times yesterday, and also today, 20 passes each run... and most of the time I tested with IBT I was overclocked to 3GHz, but now everything is at stock settings, and the crashes are happening everytime I try running max stress on IBT, usually after 2 passes...

I think I am just going to ship the mobo off to Gigabyte on Monday, because there is definitely something wrong with this board... I just wish that something would die permanently so they have to give me another mobo, instead of these intermittent crashes/freezes that I experience when stressing the system out...

I might see if I can get a new mobo this week coming, just so that I am able to test all my hardware out on it while my UD5 gets shipped back, though I am unsure if I will have the cash this week to splash out on yet more hardware... I will just have to wait and see...

I am going to try installing my OS onto another HDD and see if that makes any difference.. I think I will give XP 32bit a run...

I will post my results back in here to let you know how I am getting on Plec..

 

[DavyBoy]

Well, the DIMM channel has died yet again... but I have found that if I remove the CMOS battery, close the clear CMOS jumper and leave it like that for 10 minutes, and then put the battery back in and fire it up all the RAM gets detected, but after about 10 minutes the system BSOD's and when I restart and look at the POST screen only 2 sticks are picked up again!!

So I have decided to just remove the RAM from the first white slot altogether and just run with 4GB in dual channel, and now everything appears to be just fine...

I have another UD5 arriving tommorow, I found it in the MM, and got a really good deal on it as well, so hopefully this mobo will sort out all these issues I am having...

Though I am actually starting to think that it is the processors IMC that is damaged, because it seems to be fine in dual channel mode, but refuses to work in triple channel.. if the problems still keep happening on the new mobo, then I will have to RMA the processor back to you guys yet again!!!

But hopefully it wont come to that, I am praying to the gods that the new board makes it all better!!!

I will be buying another processor very soon to complete my Fiancees build, this time I will get a retail chip instead of OEM so that I get a longer warranty...

 

[Plec]

Well, the DIMM channel has died yet again... but I have found that if I remove the CMOS battery, close the clear CMOS jumper and leave it like that for 10 minutes, and then put the battery back in and fire it up all the RAM gets detected, but after about 10 minutes the system BSOD's and when I restart and look at the POST screen only 2 sticks are picked up again!!

I had a similar problem to that on my last build - similar in that if i cleared the CMOS the memory would run fine until i rebooted and then it would corrupt requiring a CMOS reset.

The memory passed memtest after memtest until after a few weeks one stick started to consistently fail.

Run memtest all night?

EDIT: Actually, ignore the above as it doesn't marry up with it being consistently the same DIMM slot.

Will re-read your above post later tonight when i'm not doing 3 things at he same time (your problems require a lot of attention.)

 

[DavyBoy]

Sorry to be a PITA all the time Plec... you are probably sick and tired of this thread by now!!!

 

[Plec]

Sorry to be a PITA all the time Plec... you are probably sick and tired of this thread by now!!!

I feel really sorry for you but oddly enough I’ve found the whole thread interesting – the 'freakish problem' itself and the fact that I’m beginning to think that you’re a long lost ancestor of Job (a biblical character from the Old Testament who God delighted in sh*tting on from a great height as he had a side bet running with Satan about his faith…) I would start checking your family tree...

I’m guessing that you must greet every new component with suspicion rather than excitement... :/

The new MB ‘should’ prove once and for all what the problem is – either MB or CPU. So by the end of today you’ll hopefully be in a position that you can make some informed decisions about what to do next and finally plan/build your clocked spec and enjoy it. (if it doesn't reslove the issue then you need to call in a priest and get that rig exorcised as it's clearly posssed!)

What do you plan on doing - rma’ing the duff MB and using the replacement or selling it on and buying a completely different board and putting the other UD5 in your fiancée’s rig?

Good luck DavyBoy!

 

[DavyBoy]

Hi Plec,

When the mobo comes, I will set it up and hope that the issues get resolved... if it does sort out the problems, I will ship off the current UD5 and 1 of my sets of RAM to Gigabyte and explain to them EVERYTHING that I have tried, and all of the issues I am experiencing, and just hope that they can replicate the problems at their end...

If however, the new mobo doesn't make any difference, then I am going to assume that it is my CPU that is faulty, so I will have to return that to OcUK again, but this time make sure they test it thoroughly, over the course of a few days if possible...

You know, after all the years I have been building, repairing and overclocking PC's I NEVER thought that I could experience anything like this, but how wrong was I?!?!

This system has made me feel like a total n00b... it really has!!!

 

[Plec]

This system has made me feel like a total n00b... it really has!!!

We all would have done exactly the same thing.

It's been clear that you've known what the next process has been all through this thread but - like any of us - you wanted to talk the process through and reason/deliberate your decision process. You haven't come across as n00b in any of your replies - and i stand by my statement that you're, now, probably one of the most informed ppl regarding that particular board and its settings - apart from, maybe, poor old rjkoneill who has to build about 100000 units a day using the UD5.

I've started similar threads before and will no doubt do it again - it helps clarify things and can help prompt other lines of thought. Plus, it saves you from ‘barking at the moon’ with rage as you can at least vent your frustration to other sympathetic members. The forums are as much a cathartic process as they are for troubleshooting advice when you have a problem like yours.

It's been a b*tch of a problem and it's only through lack of components that you haven't diagnosed and fixed your suspicions earlier (that and a decent local pc shop).

Hopefully, this is the finale - if it’s not, call Mulder and Scully or that priest.

Better still paint a: black then a blue and then a red ring with a single yellow spot in the middle onto the side of the case and take it to your nearest artillery range…

 

[DavyBoy]

well, the new board didn't come today... even though the seller paid for next day delivery...

However, I have spent the entire day testing out my rig with only two sticks of RAM installed, and it has been perfect!!!

It can do 20 passes of IBT no problem at all, but if I have a stick fitted into the dodgy slot it crashes after 1 pass of IBT...

I have also ran a couple of hours of Blend test in Prime95 without any issues several times today as well...

I just hope that it is the mobo that is dodgy and not my processors IMC... guess i will find out tommorow when the new mobo arrives...

It is about time i got a result with this rig, after all the money I have pumped into it.. though at least now I only need to get the processor to complete my 2nd i7 build!!!

And with all the trouble I have had with this system, I feel that I have gotten to know a little bit about X58/i7 platform... so it isn't all bad at the end of the day!!!

Instead of getting an OEM processor I will buy a retail one this time, as I can use the hsf for my fiancees rig because she wont be doing any overclocking on hers so a stock cooler will be fine...

 

[hlennie]

sorry to hear about ur trouble sounds like a real nightmare

i have the same memory and have had the same sort of problems but only with the memory your MB is FCUKED just snap it in half and write it off as a loss im sure ur new motherboard will sort ur problems out

i know how how u feel i had a nightmare job with my EVGA 680i, the findings below apply to my corsair mem hope u dont have any of the problems

1. they dont like the 1600 speed when overclocked

2. some times the third module does not show up in boot

3. would run games mostly with no problems the odd bsod

4. got it to prime stable by setting the qpi voltage to 1.4V - 1.6V (default XMP profile 1.35V crap)

5. they passed mem test with no problem at all

on further checking i found the SPD settings (poweruptech SPD programme) where wrong on the third module and where showing an others corsair make, so i just rma'ed them

now got me some OCZ Blade Series 6GB (3x2GB) DDR3 PC3-16000 Triple Channel and the system is just now flying

 

[Plec]

well, the new board didn't come today... even though the seller paid for next day delivery...

Any luck with the new UD5, DavyBoy?

 

[PaulGreyhead]

DavyBoy, you have had the most epic of journeys with that problem. I think the majority of people would have given up a long time ago.
Plec, you are indeed a saint. The way you've stayed with DavyBoy, calmed and encouraged him is superb. (Are you ready for me yet? )

DavyBoy, I have the same board and a very similar problem. Mine is stable, but has problems posting/booting. Mine does appear to have a fault in a DIMM socket as it won't fully read the memory fitted to it. The bios doesn't see the memory but Windows sees it, but says it can't use it!
I'll be waiting anxiously to see how your new board performs.
Best of luck with it.