Q9450 + Asus Striker II Extreme + Raid0 OC

[Clunk]

Am I correct in assuming, that no matter what will happen further on in while testing, the memory WILL NOT be the reason for failed overclock if left at the current settings which seem to run fine under memtest 86+?

I wouldn't go that far. In fact, while you get the CPU overclocked, you might want to run the memory at 1333MHz or so, just so there's no extra strain on the NB.

Unlinked is what you want if you want to leave the memory at a certain speed.

What do you mean by calculated GTL refs? I thought your board was pretty much automated in that department, what are you calculating them from?

Next step is to stress test with prime95 and make sure its stable at stock...you can leave the memory as it was when you were memtesting.

Give it a couple of hours of prime blend test.

[Shanara Red]

Ok. Posting back. I managed to screw up Master Boot Records on one disk and the RAID0 array. Just goes to show, how OC'ing and testing various Voltages and components and using old Win 98 boot floppy disks by mistake can REALLY set you back one day of OC'ing.

After fixing all the issues with a fresh HDD format and windows installation, I should also like to let you guys know that on fresh Windows XP , Prime 95 was failing all over the board (Blend tests), until I installed the separate drivers for the Graphics' card. Now it's been running just fine for about 3 hours. Temps are low (as expected 40 - 45 degrees C at Load - Prime 95 Blend) so that is encouraging as the processor is running at stock. I have the memory linked and synced (1:2) and is running at 1333 MHz.

Unless any of you has any other advice, I will leave PRIME 'till the evening (6-8 hours) and will the follow Clunk's begginers' OC guide to the letter. The memory is checked out and should be rock solid at 1600 7-7-7-7-20 @ 1.86v.

I will begin by raising the FSB value from 1333 to 1400 (bus: 333 to 350) and see what happens. Leave EVERYTHING on auto and take it from there. I have my doubts, but Clunk's advice seems more than reliable. Although I must admit, I am a bit sceptical, as my earlier attempts were not succesful. I will not give up just yet, and will let you know how it all went as soon as I have something else to report.

P.S.- One of my cores is consistently 5 degrees hotter (average) than the other three. Nothing serious, but if I manage to OC to say 3.2, the temps might be getting closer to 65 degrees, and a 5 degree difference can hurt at that point. What is more likely ... that the hotter core is heat-spotted or that I have placed the heatsink on in a sloppy way? I will try and replace the heatsink and use more thermal paste. (This time around, I dabbled a VERY thin layer of paste onto the cpu and heatsink with a sponge-like material ... no leaking out at the edges) .

P.P.S.- I read all over the place, that GTL_REF's should be as close to 65-67% ov VTT as possible. Once you get these numbers increasing them by a single step is supposed to give slightly more stability, especially with all the items for Spread Spectrum Contol turned off.

[affiliate13]

Keep going, i am interested to see how you get on with the 790i.

[Shanara Red]

One other thing. While running memtest 86+ v2.01 default tests 1-7 (I'll call it 'spectrum' test for easier refernece) I ran into some difficulties. The memory test would always fail at the first pass of test #4. I went into the memtest manual and f.a.q. found that these problems occur often. The way to fix it, would be to turn OFF Legacy USB support in BIOS while running 'spectrum' tests. This means, that some USB devices WILL NOT function. If you are like me and have USB keyboard it in fact means, that you will only be able to stop memtest with a rig reboot.

When USB Legacy support was turned off, the all the memory sticks passed 1:30 hours of 'spectrum' testing without problems.

[Clunk]

Originally Posted by Shanara Red

Ok. Posting back. I managed to screw up Master Boot Records on one disk and the RAID0 array. Just goes to show, how OC'ing and testing various Voltages and components and using old Win 98 boot floppy disks by mistake can REALLY set you back one day of OC'ing.

After fixing all the issues with a fresh HDD format and windows installation, I should also like to let you guys know that on fresh Windows XP , Prime 95 was failing all over the board (Blend tests), until I installed the separate drivers for the Graphics' card. Now it's been running just fine for about 3 hours. Temps are low (as expected 40 - 45 degrees C at Load - Prime 95 Blend) so that is encouraging as the processor is running at stock. I have the memory linked and synced (1:2) and is running at 1333 MHz.

Unless any of you has any other advice, I will leave PRIME 'till the evening (6-8 hours) and will the follow Clunk's begginers' OC guide to the letter. The memory is checked out and should be rock solid at 1600 7-7-7-7-20 @ 1.86v.

I will begin by raising the FSB value from 1333 to 1400 (bus: 333 to 350) and see what happens. Leave EVERYTHING on auto and take it from there. I have my doubts, but Clunk's advice seems more than reliable. Although I must admit, I am a bit sceptical, as my earlier attempts were not succesful. I will not give up just yet, and will let you know how it all went as soon as I have something else to report.

P.S.- One of my cores is consistently 5 degrees hotter (average) than the other three. Nothing serious, but if I manage to OC to say 3.2, the temps might be getting closer to 65 degrees, and a 5 degree difference can hurt at that point. What is more likely ... that the hotter core is heat-spotted or that I have placed the heatsink on in a sloppy way? I will try and replace the heatsink and use more thermal paste. (This time around, I dabbled a VERY thin layer of paste onto the cpu and heatsink with a sponge-like material ... no leaking out at the edges) .

P.P.S.- I read all over the place, that GTL_REF's should be as close to 65-67% ov VTT as possible. Once you get these numbers increasing them by a single step is supposed to give slightly more stability, especially with all the items for Spread Spectrum Contol turned off.

I keep telling you, none of the guides on here will work for your board, it's a completely different beast. The only thing that you can take from the guides is the theory. All the settings are different and will require different amounts of tweaking.


There are known problems with disk corruption on the 790i boards, have a google of that, there may be a workaround or a BIOS fox for it now.

Not sure about the GTL ref on your board, it's different for pretty much each individual Intel based board, so yours will be different again.

Don't worry about the difference in core temperatures, it's normal.

[Kelainefes]

If you really wanna reset your cooler (wich is not necessary atm), i suggest you get yourself a few items that will make it easy:
IPA, or Iso-Propyl-Alcohol, the best is the farmaceutical grade wich will be 99% when first bottled, but any variety over 70% will be ok;
a piece of sinthetic skin, like the ones that are meant to clean windows and windshields, to use together with IPA to remove the old paste and clean both contact surfaces;
a single razor blade, like this one from wilkinson, to spread the paste evenly:

optionally, surgical gloves, to keep your hands clean and to avoid leaving fingerprints on the CPU's IHS or the HSF's base.

[Shanara Red]

Ok ... guys ... 790i ultra DEFINETLY has data corruption issues. This is the third time, I'm loading windows today ...

[Clunk]

Send it back, it's unacceptable for any motherboard, never mind a hideously expensive one!

You'll be better off with an Intel based board

[Kelainefes]

Yeah, it seems that in some boards the problem goes away with a BIOS update, some never have it...
I guess you got a unlucky one.
Unless you need SLI, I'd tell whoever sold you the motherboard about the issue, and ask for a different, X48 based board.
A DFI board would be a good choice, and it would be easy for us to help you, but there are also other brands such as Gigabyte that have good models out.

[Shanara Red]

Ok. I am new at this business ... but I am far from stupid. And this mobo is taking me for a ride. Please confirm or deny what I will write below to confirm my suspicions.

-First and foremost. The ugly memory issue. The memory I have is Mushkin Ascent Extreme 1600. It is rated (by the manufacturer) to be 1600 7-7-7-20-T1 at 1.85v-1.90v. I have tested BOTH sticks on all memtest86+ tests (spectrum and test #5) at 1600 7-7-7-20-T1. They both checked out absolutely fine (after the legacy support tweak; which is a memtest86+ issue).

1)When I leave everything on [Auto] (CPU speeds, voltages ... EVERYTHING) on the mb it reads the memory as 1066 and feeds it 1.9v. The memory runs fine on windows, and Prime is stable (3 hrs test length). Good.

2) Still leaving EVERYTHING on [Auto], I set the memory to be linked and synced. The BIOS boosts it up to 1333. All voltages ... EVERYTHING is auto. The BIOS feeds it 1.9v of power. Prime again runs fine for 3 hrs +.

BUT... the memory is supposed to run at 1600! Since BIOS does not it recognize as such, I have to set the speed manually. 1600 7-7-7-20 it is. AND TESTED. When I set it to 1600 (and leave EVERYTHING else except the timings on auto) it boots, but prime Blend fails after 30 seconds ... sometimes with an ugly blue-fatal-error-screen.

Ok I figure. maybe the MB is voltage sensitive and does not know how t set the voltage properly. No problem! The memory is memtest86+ tested at 1600 7-7-7-20 @ 1.86v. I input the voltage and timings and speed manually. ... prime fails after 10 seconds ... sometimes with an ugly blue screen.

2) I am not even CLOSE to CPU overclocking, if the damn memory doesn't run at what it's supposed to! And on a MB, which natively supports speeds up to 2000 MHz!!!

... ok ... am I an idiot, or is this board taking me for a ride?

Yes ... I have looked up some other places where they are having 790i discussions. Some people get very nice results WITH HARDLY ANY HASSLE! Here is a rough tutorial as well! LOL. I realize I went about the initial OC'ing from the wrong starting point. But starting from square 1, the troubles I'm encountering (not to mention windows reinstalls!!!) seem to me to be of such a basic nature, that setting the correct GTL_REF's and NB voltage is about as far away from me as Mt. Everst (I live in the EU). To be clearer. It appears, as though the wall I have hit now is in fact very far from your-cpu-does-not-overclock-at-4GHz-'cos-you-got-a-bad-batch debate in fact it appears to be a very fundamental error in at the most basic BIOS-MB-NB circuit management levels. Please confirm or deny this assumption.

... uffff .... Yes ... The intel - nvidia chipset thing. The sole reason I went for the Stiker II Exteme, was because of SLi. The main reason I went for SLi is CUDA support. The reason I go for CUDA is that it not only offers top-performing gaming but also functions as another CPU core. This cpu core is invaluable when rendering images and fly-through movies in Autocad and Cinema 4D (both supported by CUDA-GPU 2xx core) ... which is 50% of what I will be using the machine for. Upgrade options abound with dual or even triple GPU's in SLi. So altough the X48 is looking MIGHTY appealing at this point in time, I think I would still like to explore the 790i ultra avenue. Some people seem to have no hassles with it! Please comment on this situation and offer you advice. I am also limited to ASUS MB's because of retailer stocks.

If you guys confirm my suspicions, I will take the MB back. Then (ofcourse taking any advice you have into serious cosideration!) I will decide whether to try it again on a different MB or re-try a different 790i.

And thank you both for offering very constructive advice at the intermediate steps on this 'adventure', as at times, it was the only thing that kept me going

edit
P.S.- Is it possible that any software based errors are causing these errors? Windows or some bad drivers perhaps? Everything I loaded on there is basic from the supplied cd's ...