Rampage III Formula BSOD

Hello. It's been a while since I've been [H]ard. I'm glad to be back in the game!

I had been getting the itch to build a new rig. My Core 2 6600 was long in the tooth from a speed perspective, but everything was rock solid for 24x7 operation. So about a two weeks ago I ordered a new setup for a Core i7 950. Here are the details:

Core i7 950
Asus Rampage III Formula
Xigmatek Dark Knight S1283V
Corsair Dominator 1600 C8 3x2GB (2 kits for total of 12GB)
Crucial C300 64GB (on Marvell 6GB controller with 6GB cables, using MS AHCI driver)
2x1TB Samsung F3 Spinpoint (Raid 1) (On the Intel 3GB controller)
X-Fi Titanium Fatality (I'll just spell it like normal people do.)
EVGA 8800GTX (My old card. Haven't decided between a 6870 or an OC GTX470)
Corsair HX850
Coolermaster HAF X
HP DVD-RW
Windows 7 64-bit Home Premium


When I built the rig, I connected only the Crucial drive to its port for the WIndows install. Once Windows was stable, I connected the Raid array. I originally had all 6 sticks of RAM in the machine. From a cold boot with all 6 sticks of RAM, the system only sees 8GB, and only at 1066mhz...Unless you go into the Bios and configure XMP to run. Then, it sees all 12GB @ 1600mhz. I didn't think anything of this because XMP seemed to pick it up fine. The voltage for the DIMM is a concern to me. The XMP profile set it to a shade over 1.65 (I think it is about 1.6564.) And the "real time" voltage monitor shows the RAM is using right at that limit - IN THE BIOS, with no stress put on it. But, I went ahead and started using Windows like I normally would.

Everything seemed fine. I installed Steam to the array because I didn't know how much space I'd be using with the games. Through the course of downloading stuff, I've blue screened several times. I know sometimes the Marvell controllers can be a pain, but this felt more like it was in the Mobo/RAM/PSU category.

So I started troubleshooting Tuesday evening. I took out one of the sets of RAM, so that only an original "matched" (I only mean that it was the exact sticks that came in one of the packages) was in the machine. Immediately it would only see 4 of 6GB, until the XMP was set. Again, it saw 6GB @ 1600mhz like it should. The voltage was about the same as with all 12GB.

Seemed to work......Then again downloading my stuff from Steam, it bluescreened. So I thought I'd roll back the speed manually. I ran at 1066 mhz and it seemed fine. Did a 64bit Prime 95 for a while with no problems. I thought I would try manually setting the profile basically using the exact settings that the XMP would use. It wouldn't see all the menory no matter what I tried. It seems that XMP is the only way I can use all 12GB.

Nevertheless, I poked around the Bios again. I turned off speedstep/turbo. That seemed to make it run stable at 1600mhz with XMP and all 12GB of RAM.

But my question is this....if I'm not in any way overclocking at the moment, trying to run "stock" settings (at least for the RAM), shouldn't I be able to use Speedstep at least? I mean, that's a core function of Intel platforms - To scale back when processing power isn't needed.

I am leaning towards the Rampage III Formula being unstable. It should be able to handle running at 1600 with 1.65V no problem. But I'm open to suggestions. I'm getting pretty frustrated with this build, and I'd like to know what I can change (even if it is an RMA). I've torn down the hardware and rebuilt everything about 5 times now (don't ask).

Anyone had a similar issue?

I have had zero problems aside from the lan dissapearing and then coming back for whatever reason. Have you run memtest to factor out the sticks possibly being bad?

@Afterburntdw,

Try increasing Vtt/QPI to 1.35v (or a little higher). Running 12GB of RAM is going to heavily stress your IMC. Sometimes XMP profiles don't worry correctly or at all, so it's best to resort to manually setting everything.

DarkStryke said:

I have had zero problems aside from the lan dissapearing and then coming back for whatever reason.

Click to expand...

Try downloading the latest Intel LAN driver. Link

Most manufacturers don't publish the latest driver updates, and Asus has theirs listed back from March.

Thanks for the replies. To answer each question - I have not run memtest yet. I did not because I assumed that because the XMP profile sees all 12GB, they probably all work. But that's an assumption, and I could be very wrong. I've actually been considering simply running with 6GB instead. I will run the memory test on all sticks.

For the voltage, I left it auto...I will try increasing that and see what happens. Should I increase the headroom for the DIMM voltage as well? It's using right at the limit.

Also, is there any way that this can be a PSU related issue? I know the HX850 is a highly regarded PSU, but it's possible that it is bad.

I got exact the setup as you, but with 6 gigs of Patriot Mem. I have 2 Samsung F3, but setup in Raid 0 and my intel 80 gig SSD as the OS drive. I have had 0 issues running it. Check what voltage that memory is suppose to run at and change the voltage to it and see if that helps. Mine says 1.65 but it's stable at 1.62 so I left it at 1.62 with a 4 ghz on the CPU OC.

I have a Rampage Formula (X48) and I learned the hard way to NOT enable LoadLineCalibration aka VDroop on my board. Make sure that is disabled on your board.

Also as someone else said, adjust Vtt to 1.4V and I think that will help.

I don't think pushing VDimm past 1.65V will help.

Run memtest as someone else stated for about an hour or so on all your sticks.

Might consider upping VCore a little bit higher, say .1V

I will run memtest when I get home.

Question: If the XMP puts the DIMM voltage already a hair past 1.65, is that going to cause long term problems? I've read time and again that you don't want to go past 1.65v.

Afterburntdw said:

I will run memtest when I get home.

Question: If the XMP puts the DIMM voltage already a hair past 1.65, is that going to cause long term problems? I've read time and again that you don't want to go past 1.65v.

Click to expand...

Nope. Heck, you can run 1.8v and be fine.

For LGA 1366 boards, the rule is to keep Vtt and Vdimm within 0.5v of each other. So if you're running 1.67v Vdimm and 1.4v Vtt, you're okay.

You may need more than 1.65v though. With all of the memory slots filled, the voltage drop increases so you have to add a bit more to compensate.

Ok. Thanks. As a side note, it sure helps having a great community here.

I think the QPI was running on "auto", and it was using about 1.11-1.12v. Does that seem low? Maybe that is why it was unstable - More than .5v away from each other?

(I'm just going off of MY memory here....could be faulty )

One other question. Should Speedstep & Turbo be disabled, or should they still work once the voltage is raised? (I will probably overclock later, but I want the system to be stable first. If I did OC, I'd disable speedstep for sure.)

Afterburntdw said:

Ok. Thanks. As a side note, it sure helps having a great community here.

I think the QPI was running on "auto", and it was using about 1.11-1.12v. Does that seem low? Maybe that is why it was unstable - More than .5v away from each other?

(I'm just going off of MY memory here....could be faulty )

Click to expand...

I think the default is around 1.15v-1.20v.

Afterburntdw said:

One other question. Should Speedstep & Turbo be disabled, or should they still work once the voltage is raised? (I will probably overclock later, but I want the system to be stable first. If I did OC, I'd disable speedstep for sure.)

Click to expand...

Depending on the board, you can disable Turbo but force the 21x multiplier.

EIST can be enabled either way but it's advised to disable it when you're overclocking as it helps eliminate some variables. At stock, it's fine to have it enabled.

Hey there, i'm so glad i found this post. I was just about to post with an almost identical problem... with an almost identical build!

I just got my R3F, 950, and 12gb of the corsair dominator, and the same SSD as you as well.

Would you mind posting your bios settings for me? I'm having all kinds of trouble. Its been consistently seeing all 12gb, but ive been getting BSOD's as well with BAD_CONFIG etc type errors.

One thing I did which was dumb was use the drivers from the CD. I'm pretty certain they were leading to some of my problems.

Edit:
I enabled XMP as well and seem to be having some better luck. I think I should be still increasing something else too, and i'll just come right out and ask it because its been so long for me..

Can someone tell me which lines in the bios config are the "vtt" and "vdimm" ?? sorry for the dumb question!

Thanks in advance!

No problem. Well, I'm currently not able to reboot and check all of the BIOS settings. But aside from changing the Marvell controller to operate on AHCI, I set the speed manually to 1600 mhz. That's it I believe. What this doesn't do is set the timing to what they should be. I think I'm running at 9-9-9-24. I havent had a chance to fine tune it but this seems to work if you don't want voltage tweaks. I ran 64 bit prime 95 for an hour and a half with no Bsod.

Hope that helps a little bit.

A new problem has manifested itself. During boot up, the marvel initialization portion takes slightly longer than normal and then only my 1TB sata III drive is seen, not the SSD on sata III.

i haven't put in any drivers yet, did you get them off the asus website?

If i clear the CMOS, it lets it see it (this happened once before and i'm assuming the same will occur again), but i dont really want to do that...

Kreator said:

A new problem has manifested itself. During boot up, the marvel initialization portion takes slightly longer than normal and then only my 1TB sata III drive is seen, not the SSD on sata III.

i haven't put in any drivers yet, did you get them off the asus website?

If i clear the CMOS, it lets it see it (this happened once before and i'm assuming the same will occur again), but i dont really want to do that...

Click to expand...

Hmm. I'm not sure about that one. From what I have read, you may be better off putting the mechanical drive on the Intel controller, despite the faster SATA III interface. That's bc there is little gain for a mechanical drive past SATA II. I think you'll find better stability with this setup as well.

As for the drivers...I haven't loaded the Marvell driver - seems to be great speed with the MS AHCI driver.

Still haven't blue screened, and I've run stress tests and downloaded through steam (which definitely caused it before.)

Please post what you end up doing. This thread has really helped me!

Here's a bit of an update from me. I've found I have much better stability if I boot with the option to allow unsigned drivers. I cant recall the exact phrase in the boot menu for this, but under advanced booting its pretty much the last option in the list.

I'm running under the XMP profile right now and seem to have great performance from that. The only problems im having are with the drives. In my system event viewer, I'm seeing things from source mv91xx. I haven't searched on that yet, but my hunch is that its a Marvel something that isn't happy. They arn't actually errors, Warnings really.. except when the SSD is the one that has an issue and then after about 5 min of freezing in windows i BSOD with a message that says it stopped some essential process from running or something similar to that.

The errors themselves are:

Request to physical disk 1 is timed out.
Command [0x28] on physical disk 1 failed. Sense key=0xb, ASC=0x0, ASCQ=0x0.
Physical disk 1 is plugged out.

This is from source disk:
The device, \Device\Harddisk1\DR1, is not ready for access yet.

This is from source ntfs:
The system failed to flush data to the transaction log. Corruption may occur.

And last but not least, this is from source Application Popup:
Application popup: Windows - Delayed Write Failed : Exception Processing Message 0xc0000222 Parameters 0x000007FEFCFE715C 0x000007FEFCFE715C 0x000007FEFCFE715C 0x000007FEFCFE715C


During this time i've been downloading Starcraft 2 via the blizz downloader to the 1TB mechanical drive. I have a Marvel driver installed that i got from the asus site...

I did switch from IDE to AHCI in the bios. What MS AHCI driver are you using, just one that Win7 picked up on its own and used or did you do something special?

I think my next step is going to be to uninstall the Marvel driver and see what happens. My concern is that, as i said in an earlier post i think, I had some problems during boot when it loads the marvel driver and identifies drives where it would sometimes not at all see the SSD.

Hmm. Sounds like your setup is a bit different than mine. It may not make a huge difference (or it could make all the difference in the world), but I built the computer from scratch using AHCI on the C300. I also didn't have any other drives (besides optical) connected during the build process. For drivers, I am using the MS driver for the Marvell interface that is installed during Windows 7 installation. I've heard that the official Marvell drivers aren't as fast, and it's entirely possible that that is part of the instability in your setup. You might want to try changing the driver to the MS spec'd one.

I know my setup is running MUCH better now that I rebuilt it again. Just a suggestion (I know it's a pain.)

Did you have any luck with it since your last post?

Seems I spoke too soon. It blue screened when i attempted to launch Left 4 Dead. It's a different blue screen this time - Referring to dxgkrnl.sys, which I think is the DirectX kernel module. I'm running the latest nVidia drivers with a brand new GTX 470. Any ideas?

Sorry for the delay in posting... I worked on it quite a while yesterday. A guy from OC.net forums volunteered to try and help me out.

I was having a pretty bad problem with heat on my i7 950, can you tell me what idle temps you're getting and under load? I didn't have the HS properly seated, or used too much compound, or something the first time... i was getting idle temps around 70c. WAY high.

I took everything apart yesterday and redid it all. temps are down from 70c idle, to around 50... which is still high i believe. In the bios, i see temps around 40c but yet when i enter windows and run Real Temp, i see 50c..

Anyway, I did end up reinstalling yesterday. One problem I've encountered is during installation my install just completely stops. No error, didnt BSOD itself after about 20 min or so... but i left the installer alone in the same spot for 20 min and it never budged. I had this happen the first time i tried to install last time as well. A second install attempt worked fine though.

I still get issues with random freezing. Mouse still moves, so its not a complete lock... I can't right click and open any menus, click start button, or anything like that.... When i had the Marvell driver installed and this happened, i would come back to an event viewer filled with disk time outs. Now, i'm not seeing anything except for the typical kernel was shutdown unexpectedly error.

When i just reinstalled, I had both the SATA III and SATA II controllers set to AHCI. My question right now is, I still see the Marvel controller listed in my device manager in the "other devices" with a yellow ? mark, indicating it has no driver, etc. I've tried figuring out how to install a MS driver for this, and i believe this is what i need to do, but i cant get it done.

Today my plans are to check on the temps again to see if I've got a faulty CPU, then after that, see what i can make of the current install of Win7. I'm going to try to get an MS driver installed for the marvell controller, which i dont think has been done automatically, and see if my timeouts on the drive continue.

If they do, my next step was to try the SSD in a sata II port and see how I fare. At least then I can rule out the SSD itself as being faulty, and can blame it entirely on the shitty marvell controller

Wow, seems like you had an interesting day!

Well, those temps are sky high. I've tried a couple of different types of motherboards, so my CPU/fan has been reseated several times, and the results are the same in my case. However, just for reference, I'm seeing about 31 degrees at idle, and I think I never went past 70 under load from Prime 95 (64 bit).

I may need to talk to the guy you were referring to....Would you feel comfortable PM'ing me his information? If not, please send me a PM and I'll give you my information so you could pass it along instead.

As an update to my post last night, I raised the voltage on the QPI slightly so that it's not "auto" but just under 1.4. It's too early to tell if that worked or not.

I'm considering all things....Maybe my PSU mounted facing down doesn't have enough space to vent the heat between the carpet and the case (it's a HAF X case with the wheels installed). Perhaps I have a faulty PSU. Perhaps the PSU is fine, but the connector I'm using is faulty. It could be that the Marvell controller sucks and I need to use the Intel port instead. Perhaps my CPU is bad. Or the RAM. Or it's the motherboard. Geez, it could never end....

I'm starting to wonder if I even want a Core i7....or at least the parts I chose. I need some help!