Anyone help with OC issue

paranoidandroid

paranoidandroid

paranoidandroid

Associate
Joined
29 Nov 2006
Posts
28
I recently set about overclocking my i7920 D0. Once I had done and tested a few times I ran it through Intel Burn Test okay, no problems there, and highest temp was average 65. On idle, mid to late 40's.

The odd thing that I have noticed though is on boot up, sometimes it recognises my full 6gb, most the time it only picks up 4gb. I have tried tweaking some of he settings but it still has the same problem I have tested all of the memory sticks individually and in different slots. I have run memtest and all is okay, so I am thinking it must be something in my settings. Any guidance on this would really be appreciated. Settings and spec as follows:-

CPU Clock 20
Turbo Standard
QPI 36
Uncore 13
BCLK 190
Mem Multi 6

V Core 1.28125
QPI 1.315
IOH Auto
VDRAM 1.66

RAM Timing 9-9-9-24-2

Motherboard Gigabyte EX58 UD3R
CPU Intel i7920 D0 Stepping
RAM CORSAIR 3X2GB DDR3 1333C9
Video card Evga GeForce8800gts

Any suggestions


Thanks

Paul
 
I'm guessing you've tried the obvious - reseating the memory and looking at the DIMM slots and memory for obvious physical damage or post production residue?

If you have sometimes a bent pin in the CPU socket can give up intermittent errors like this - and can usually be easily fixed. Unfortunately the only way to confirm this is by removing the CPU and having a look... :/ (link - to a similar problem)

Another cause can be a faulty memory controller on the CPU but that's a big maybe. ( link - again to a similar problem. It's a long read but gets concluded on the last page if you want to skip the troubleshooting process. However, the bent pin is the most likely if it's nothing more obvious)


I would look at the cpu slot as i've read a couple of threads with this particular problem on the UD3/5 boards and this has turned out to be the problem - that is if you have ruled out the obvious of poorly fitting memory/damaged memory/slots etc...
 
Last edited:
Hi

Thanks for getting back to me, I have checked all the seating on the memory and cpu and they are all okay. Always boots with the correct RAM when tried in different slots/modules.

If I leave on optimised settings and just alter VDRAM it sees 6gb every time. Reading the other posts I am wondering if my power unit is sufficient, I have a Thermaltake 680w and I am running 4 hard drives, could it be anything to do with that I wonder
 
dr4g0nf1y said:
It's just a side effect of overclocking. you still have the 6 and will use 6 but it's displaying it wrong.
Click to expand...

I have never, ever seen this written anywhere before. Overclocking doesn't accidentally stop you seeing 2gb of ram.

It sounds very odd, but what cpu cooler are you using? The issues Ive seen with this board related to the cpu cooler causing the issues, which was rather hard to diagnose. I'll see if I can point the relavent chap at this thread
 
dr4g0nf1y said:
It's just a side effect of overclocking. you still have the 6 and will use 6 but it's displaying it wrong.
Click to expand...

LoL!!

Where did you get this information from?!?!

OP, I went through hell with a very similar issue to this, I RMA'd my mobo, and I RMA'd my processor twice, and each time they were found to be not faulty!

I had suspected it was the IMC on my first chip that was flaky, and after talking with Gigabyte and Intel support, they to also agreed with me after all the testing I had done myself.

I went and bought a new mobo, PSU, and another set of RAM to try and find the cause of this fault, and nothing made a blind bit of a difference.

Anyway, the kind people at OcUK actually sent me another processor even though they tested my first one and never found there to be any faults.

But I have been talking with a guy from XS forums about this issue, as he experienced the same thing as well, he also bought a new mobo, RAM, and processor, and again, nothing made any difference.

Someone suggested to him that it maybe a bad HSF mount that was causing the problem, so he removed the HSF and reseated it, making sure that it wasn't to tight this time, and ever since then he's been up and running for 2 months now and hasn't had the 4GB of RAM issue ever since.

I actually suspect that it was my old Zalman 1366 mounting bracket for my 9700NT cooler that was causing the problem on my system.

When I got the new chip from OcUK, I fitted my TRUE and when I fired up my system everything was perfect, all 6GB of RAM has been recognized ever since!

I have read quite a lot about this problem, and there can be a number of different things that cause this to happen.

Having RAM timings/sub-timings set to tight can cause this issue. Bent pins in the CPU socket, dodgy IMC, I have also read that some processors will do this when you have the incorrect distance between QPI/Vtt and VDimm in the BIOS, see this thread here read memory tip 6 I think it is.

Out of curiousity which HSF are you using?

I suspect that it is probably heatsinks that don't use a backplate and screw through from the underside of the mobo into the heatsinks bracket that can cause this issue.

It sounds like a really stupid thing that would be the last thing you would expect to cause an issue like this, but I am certain it probably has something to do with the mounting bracket warping the board or something like that.

::edit::

Also, this isn't unique to Gigabyte X58 boards neither, I have heard people complaining about only having 4GB of RAM instead of 6GB with eVGA, and Asus X58 mobos.
 
Last edited:
DavyBoy said:
Someone suggested to him that it maybe a bad HSF mount that was causing the problem, so he removed the HSF and reseated it, making sure that it wasn't to tight this time, and ever since then he's been up and running for 2 months now and hasn't had the 4GB of RAM issue ever since.

I actually suspect that it was my old Zalman 1366 mounting bracket for my 9700NT cooler that was causing the problem on my system.

When I got the new chip from OcUK, I fitted my TRUE and when I fired up my system everything was perfect, all 6GB of RAM has been recognized ever since!
Also, this isn't unique to Gigabyte X58 boards neither, I have heard people complaining about only having 4GB of RAM instead of 6GB with eVGA, and Asus X58 mobos.
Click to expand...

Ayeup DavyBoy

I told you you were fast becoming the leading exponent regarding this problem.

Interesting read (and seconds JonJ's findings) – the heatsink problem replicates a bent pin so i wonder if an over tightened HS could cause the CPU to become displaced slightly causing it to lose contact with a pin?..

This is the trouble with the modern CPU fitting – the slot and fitting is too delicate and prone to damage and, perhaps, mis-alignment if a heatsink is fitted incorrectly/tight. (time and research will tell i guess)

EDIT: I’ve posted this in your thread – link - in case we take this thread too far off topic. ;)
 
paranoidandroid said:
I have an Asus Triton 81, there isn't a back plate to install it clips on as per the Intel Fan supplied with the chip. I'm wondering if it is down to my settings the more I look at it
Click to expand...
DavyBoy said:
Someone suggested to him that it maybe a bad HSF mount that was causing the problem, so he removed the HSF and reseated it, making sure that it wasn't to tight this time, and ever since then
Click to expand...
Highlighting DavyBoy’s comment above (and JonJ’s post) and taking into account that you have a mighty HS attached via the pushpin method – i would still consider this as a strong contender (dispite the absence of a backplate.).

The fact that you have a huge HS attached using the pushpin method could be causing contact issues as it’s very difficult to be sure that you have an even and sustained/proportionate amount of pressure applied to the CPU. As a result you may have a disproportionate amount of pressure on one side of the CPU – and in turn poor pin/CPU contact – maybe... (worth ruling out.)

For the sake of £10 i would buy a bolt-thru-kit and try and rule it out as a possible cause. You would then be able to experiment with different fitting tensions and ensure that you have an even amount of pressure on all sides of the HS fitting – which is impossible to do with the pushpin method.

Something like this - link - but i would double check that it's compatible with your HS (it should be).
 
Last edited:
Are you running a 64-bit OS or a 32-bit OS?

A 32-bit OS will only ever show 4Gb RAM, and it will knock off a bit for the OS and a bit for your graphics card too, so you could have 6Gb, but you only see 2-3Gb depending on your OS.

It should always show 6Gb on boot-up though. If that's variable, and you've changed slots, then RMA the RAM.
 
Last edited:
WJA96 said:
Are you running a 64-bit OS or a 32-bit OS?

A 32-bit OS will only ever show 4Gb RAM, and it will knock off a bit for the OS and a bit for your graphics card too, so you could have 6Gb, but you only see 2-3Gb depending on your OS.

It should always show 6Gb on boot-up though. If that's variable, and you've changed slots, then RMA the RAM.
Click to expand...

Faulty RAM is another thing that I completely forgot to mention, thanks for that WJA96!!

Though this problem isn't always caused by faulty RAM, as I said, I spent about 2 months trying to figure out what on earth was causing this issue on my system, buying new a new mobo, RAM, and PSU to try and rule out my current hardware, and none of that worked.

When it first started happening, there would be maybe one day out of the week where it would POST and see the full 6GB, but as soon as I loaded up the processor I would get an instant BSOD and then it would go back to only seeing 4GB again.

I tested each one of my modules individually (from both kits of RAM) using memtest boot disc, and left them to run for 20+ passes each, and not one stick gave me a single error.

@OP, have you tested each stick individually with Memtest boot disc?

If not that should really be your first port of call.

But I suspect that it has something to do with that HSF tbh, if you have the stock cooler then you could install that and then test your system to see if it recognizes all 6GB.

And like WJA96 said, if you are using a 32bit OS it will only ever see a maximum of 4GB, usually more like 3.25 to 3.5GB actually, but I am assuming you already knew this though, and are using a 64bit OS?

If you have the stock cooler, I suggest that you install it, then reset the CMOS, enter the BIOS, load optimized defaults, save and exit, and then see if all 6GB is being recognized..

Let us know how you get on with it mate.

Good luck ;)
 
That was the same for me mate, however, when I updated CPUZ to 1.52 it would only see the 4GB, also if you browse to the Performance tab in task manager, you will see that it says in there that only 4GB is usable.

Have you tested each module individually with Memtest?

If not, try that, and if they all pass, then you should try my suggestion using the stock HSF.

Honestly mate, I spent the best part of 2 months trying to figure out what the cause of this issue was.
 
I don't think the heatsink is the answer.

paranoidandroid said:
If I leave on optimised settings and just alter VDRAM it sees 6gb every time.
Click to expand...

So the 6GB is always recognised when it's not overclocked. This suggests that it's an overclocking/setup issue rather than a hardware fitment issue. If it was the HSF then surely the problem would occur consistently or at least randomly?

The first thing I'd try would be to increase the QPI voltage slightly.
 
Damn, I should have read the entire thread and not just his first post!!

Mattus is right, it won't be the HSF that is causing this issue then, not if all 6GB is being detected when using optimized defaults.

Try raising QPI/Vtt a little bit more to see if that solves the problem.
 
Well problem solved, I started looking at the RAM again and started swapping sticks etc. Guess what, one of the sticks would not boot the computer, tried it in each slot with the same result. The only thing I can think is that this particular offending item was a bit flaky from day one hence the 6gb sometimes and 4gb others, when I tested first time round it must have been having a good day, it must have given up totally over the last few days.

Ah well RMA, just hoping I can do one stick and not all 3
 
You will have to rma all 3 as it's part of set. This may be an inconvenience but it's better to get them replaced for another supposed pre-tested tri-set.

It's not unusual to get faulty memory - it seems to be part if my annual ritual when building a new rig :/
 
Last edited:
Glad you figured out what the problem was mate.

Like Plec said, you will have to RMA all 3 as a set, which can be a PITA if you don't have any other RAM you can use while your others are being RMA'd.

There is a cheap set of 3GB G Skill 1600MHz RAM in the Clearance Memory section of the shop website.

http://www.overclockers.co.uk/showproduct.php?prodid=MY-032-GS&groupid=701&catid=8&subcat=400

If you had the spare cash you could buy those and use them while your other modules are being RMA'd, and then sell them on once you get your RMA replacements.. just an idea.
 
You must log in or register to reply here.
Back
Top Bottom