Official MSI P7N Diamond (780i) Discussion/Review/Overclock/Guide/BIOS Thread

[theYipster]

It's weird that some would have this issue and others don't. X-Built, what is your 3DMark score with SLI correctly enabled and your proc at 3.6ghz? If you don't know of hand, can you run the benchmark and post results?

[loonym]

Interesting evening installing 9600gt parts. What I discovered is that using bios .P05, XP64 will not load with any available nvidia driver. It stalls at welcome screen every time. My only recourse was to flash .114 in order to use new parts.

[xraymtb]

Xray,

Are you using two bridge straps? When I put my cards (2) in SLI, I originally only had 1 bridge strap installed. After I got lousy 3DMark scores, I checked the MSI manual and diagram showed 2 straps. Adding the second strap got my scores up.

An interesting point for the GTX/Ultra owners but the GT/GTS only has one SLI connector.

I do recall that I also had trouble getting the SLI mode enabled in the nVidia control panel after I had made some hardware changes. I had to set the SLI mode, reboot, and then re-install the video drivers. You might want to try that also.

When I ran the benchmarks I had the SLI indicators visible and they looked normal suggesting SLI was enabled correctly. The difference I have seen does appear to be a lot larger than others have which suggests something else was having an effect.

[gamin]

hi there.

I've tried some overclocking using theYipster's unvaluable tips and tricks, but I have to admit I'm a little scared when my computer freezes or give a bsod, so I'm really gonna try to learn the beast completely before trying again.

I have some questions regarding the board and the cpu :

- as said in my sig, I'm using E8500 C0@3.16Ghz with 4x1Gb Crucial Ballistix Tracer PC6400@4-4-4-12 2.2v (had to set it manually in the bios otherwise they were only 5-5-5-18) with bios 1.14b. when I look in the bios for my cpu temp, it says 33°, which seems to be quiet good. but when I look at the temp from windows (coretemp, everest, hwmonitor), it tells me core#1 53° and core#2 43° on idle. is it possible to have such a difference between both cores or is there an issue with the reporting softwares ? if yes, what have I done wrong ? it seems to be so hot. I'm currently using a Noctua NH-U12P with fan at 1200rpm and Noctua NT-H1 thermal paste. shall I try Arctic Silver 5 or Arctic Ceramique instead ? (got both at home, so I can try)
- once, I noticed that the bios (using 1.14b) automatically changed my cpu multiplier to 9 instead of 9.5 (so cpu @3Ghz) and ram fsb to 533 instead of 800Mhz. to be honest, I can't remember after which operation this happened. I've had to manually put right settings in bios for them to come back and I left them that way.
- for those having troubles with usb keyboard, I noticed that each time I had to clear cmos and to restart the computer, my usb keyboard wasn't responding before getting into windows. when you load default optimized settings, everything is back to normal and I can use my usb keyboard to enter the bios (have to do it with a PS2 keyboard plugged of course). can anyone tell me which is the f!!@#ing setting in the bios that control that ?

told you, still learning from the beast...

thanks to anyone who can help

[theYipster]

- When overclocking, computer freezes and blue screens are to be expected. It just means that your current settings are unstable. Overclocking is all about trial and error and no matter how good you are, chances are there will be quite a few errors.

- 45nm CPUs have well know issues with temp reporting software. It's a common bug and I'm not sure if new software resolves it, or if a BIOS update is needed. Note that you haven't done anything wrong in reading core temperatures. If you haven't already, please make sure you are using the latest versions of CoreTemp, Everest, etc. You may also want to browse the E8400/E8500 thread, where many others have discussed the same issue. I don't know if a common solution has yet been found, but that's where I would look for one.

- As for the multipliers, note that there might be lingering compatibility issues with the current BIOS in supporting half multipliers. If you see that your BIOS isn't maintaining a 9.5 multiplier (even though you haven't changed settings,) you should inform MSI.

[gamin]

as usual, thanks for the quick reply so temp should be ok (no freeze at all, stable, fast, etc...)

I forgot to mention 2 things :

- my DRAM voltage is marked RED in the bios after I changed it. normal ?
- cpuz is reporting a 5:6 ratio when I manually set everything to what it should be when normal (cpu 3.16Ghz, RAM 4-4-4-12@2.2v). am I not supposed to get 1:1 ? if so, any clue on how I could get my 1:1 back ? when I leave everything default, it says 1:1, but RAM is only 5-5-5-18 then ?

[theYipster]

You should set RAM voltage to the manufacturers specification, regardless of what the BIOS colors it.

With your system at default settings (proc at 3.16ghz, RAM at 800mhz,) you won't be running on a 1:1 ratio. Remember that a 1:1 ratio means that yoru RAM is synced to the base FSBs. Since the base FSB of your proc at stock is 333mhz, RAM needs to be set to 667mhz to run 1:1 (remember that the RAM mhz you set in the BIOS is 2 x base FSB.)

To keep a 1:1 relationship, set RAM manually in the BIOS to 1/2 your FSB. Note that this is really best practice for overclocking. When running at stock, there is no issue having your RAM run at 5:6 ratio (so that it maintains 800mhz.)

For your RAM to run 1:1 at it's stock setting of 800mhz, you will need a 1600mhz FSB. This will give you an overclock of 3.8ghz, considering your chip's multiplier of 9.5

[x-built-Stan]

Yipster,

Project M3cury 3DMark Score: 18736
Here is the link:
http://service.futuremark.com/result...eResultType=14

[x-built-Stan]

Everyone - Here's a leasson learned.

Gamin's post reminded me of a situation that everyone could learn from. This happened last year when I was overclocking my first system. As is expected, I experienced several BSODs and freezes. What I didn't realize at the time was when that happens you run the risk of slowly corrupting your hard disk. Mostly due to Windows delayed disk writes and caching. Anyway, after numerous occurances I was no longer able to boot Windows - to corrupted on the disk. The result was a complete re-install of the OS and all software - it took a couple of days - what a pain.

After that happend, I got a backup utility (Ghost). Now I always make a backup before I start an OC series - just in case. I really recommend using a backup utility before you start OCing.

FYI: If you buy one, make sure it support the size and type of HDDs that you use.

[theYipster]

Great point,

I myself keep my main system environment on a central RAID array (properly backed up of course,) but then also keep a hard-drive called "Bench Drive," which contains a bare-bones XP-32 install plus all necessary stress test, temp monitoring, and benchmark software. I conduct my overclock endeavors on this bench drive, so that my main data and system information is free from any ill-effects caused by bluescreens or corrupt writes.

[x-built-Stan]

Yipster,

I also use a barebones harddisk for OCing my builds. But a re-build is certainly more painful than a restore. Additionaly, when the failure on occured on that first build, I thought I was protected because the drive was mirroring in RAID. Well it turns out that they were mirroring and the corruption was mirrored onto the second drive - both were useless. Mirroring is really designed to protect against a physical drive failure, not corruption of a drive's data - Also a good lesson.

[BKA]

Great tips X-built and Yister about making sure you have a full back up of data before doing any overclocking. I also learned the hard way years ago. Before I do anything BIOS related I make sure I have whatever I need to do a quick re-install of the OS.

Also to Gamin, I have that problem with the half-multipliers not showing up with my Q9300, at least if you change yours you can get it back without swapping CPU's.

I have a Q9450 OEM on the way which has a 8 ratio but is still listed as "under testing" on the MSI site. I'm still going to try and overclock it though. With my luck I'll only get half-multipliers for it though!

What's up with a new BIOS release though, MSI is normally quicker than this aren't they?

[theYipster]

Yipster,

I also use a barebones harddisk for OCing my builds. But a re-build is certainly more painful than a restore. Additionaly, when the failure on occured on that first build, I thought I was protected because the drive was mirroring in RAID. Well it turns out that they were mirroring and the corruption was mirrored onto the second drive - both were useless. Mirroring is really designed to protect against a physical drive failure, not corruption of a drive's data - Also a good lesson.

Ah yes the great RAID-1 debacle As a software architect, I've come across my fair share of developers, sys-admins, etc. who are lured into the false sense of security RAID-1 provides. Moral of the story is that no matter your hard drive setup, always backup, backup, backup!

[x-built-Stan]

Yipster,

I ordered a 250mm (850 RPM, 20 dBA, 120 CFM) side fan for my rig. That should help get the NB temps down. I plan on installing it this week. I'll take NB temp measurements before and after and post them with pics.

[gamin]

ok, here I am with another problem. the X-Fi sound card... does anyone else is experiencing poor quality, scratches when moving the mouse, etc ? XP32 and Vista 64, same problem : awful sounds, sounds crackeling all the time, ...

I've read many things, and it seems to be a common problem with nforce chipsets since nforce 4 if I'm right.

for now, I've tried the latest vista drivers available from creative website, but it did nothing. maybe I should move my X-Fi to another PCI-e slot (in the yellow one actually). I'm gonna try theYipster's tip with SLI and put it in one of the PCI-e 1x with one of my 8800GT in another slot.

I've read somewhere that a guy solved the problem by putting all the settings in the bios to the right one for his hardware manually, but it was in 2006 with an older mobo... maybe another way to solve the problem... I'll keep ou in touch.

[loonym]

My best x-fi solution has been Youp_Pax drivers by Robert McLelland. The site is here: http://www.youp-pax.org/.
There's also a big thread about this at DriverHeaven forum here

edit: If you do decide to experiment with drivers, be it audio or graphics, do yourself a favor and grab driver cleaner pro from guru3d site or somewhere. A traditional uninstall still leaves unwanted files and registry entries.

[gamin]

well, with 2 NX8800GT Zilent, there's no need to move one to the white PCI-e : there's enough space for the X-Fi to go in the free PCI-e 1x slot remaining. anyway, this fixed the problem, but not completely. this is not worth as it was, but there are still some glitches, scratches, etc...

thanks for the link loonym, I'll give it a try tomorrow after my job day have to take some sleep after this 2 days testing session...

also thanks for the tip with driver cleaner pro (will go in my survey utilities folder), but I already made a drivers free XP image, so I just need to redump it

[silent-circuit]

I'm already learning to hate this board with a quad, just like every other Nvidia chipset board I've been through (EVGA 680i SE, XFX 680i LT, Abit IN9-32X MAX).

Tried Mark's suggested settings and plenty of vcore with my X3220 -- it'll get in to Vista occasionally, but I've yet to get more than a couple of minutes in with Prime95 on 4 cores. It doesn't just stop the threads, it locks hard.

Tried upping a few things, but the northbridge is already running hot from feel and I'll be the first to admit I don't know what half those settings are referring to, so I'm a bit leery of setting them sky high. Mark's default settings don't normally get in to Vista. With a bump, sometimes they do, sometimes they bluescreen before getting a desktop up, other times they black screen.

I've seen plenty of "DRIVER IRQ NOT LESS THAN OR EQUAL" and similar bluescreens, and multiple references to 'ecache.sys' if I'm remembering correctly. Nothing very helpful; just general instability I figure.

It's fine at stock, of course... but I didn't pay this much to run stock. I'm pretty pissed off.

[theYipster]

Hi Silent,

Standard questions apply here:

what bios are you using?
what are you trying to OC to?
what are you setting for vCore and vDimm?
how far have you pushed other settings?
Lastly, what is the VID of your X3220?

Just for s and giggles, try a GTL REF between 30-35 instead of 65-67 and see if that improves things any.

[silent-circuit]

Hi Silent,

Standard questions apply here:

what bios are you using?
what are you trying to OC to?
what are you setting for vCore and vDimm?
how far have you pushed other settings?
Lastly, what is the VID of your X3220?

Just for s and giggles, try a GTL REF between 30-35 instead of 65-67 and see if that improves things any.

Bios is 1.1 B4, taken from the MSI forums. I haven't tried the Performance BIOS as this one was apparently better for quads in most cases.

I'm trying to hit 3.6Ghz (400Mhz FSB x9) stable with the X3220 and move up from there.

vcore is +0.2 (probably more than it needs) and vdimm is 2.1v which is suggested by Patriot (at 5-5-5-22 2T, not pushing the sticks at all, just trying to get stable first)

I just bumped things a notch or two above your recommended. It's very hit and miss, mostly miss right now. So far this isn't doing any better than the 680i LT did, and it cost a hell of a lot more, so I was expecting /some/ improvement.

I don't believe this board is reading my VID properly; the IN9-32X MAX and the 680i LT said it was 1.18, this one seems to think it's 1.24.

Cooling is a Black Ice Pro II with a pair of Panaflo low speeds at ~8 volts on a rheobus and a Swiftech Apogee GT with AS5. I've cranked the fans and had no better results. Temps are higher on this board, presumably due to its setting the vcore higher, but they don't top ~45C load at stock and I haven't seen higher than high 50s C when I've managed to get it to do anything at 3.6Ghz.

[theYipster]

Silent,

I have never seen a G0 Quad with a VID below 1.2. In cases when they've been reported, the user usually had speed-step turned on or some other issue which can cause VID to be reported incorrectly. I don't recall the differences between your Xenon and the G0 off hand, but if they are more or less the same chip, I wouldn't be too sure if 1.18 is accurate.

For your next step, I would raise vCore to +.3 and work down from there. It may be much more than you think you need, but stable Quad settings with this BIOS do seem to rely heavily on lots of vCore. This may be in relation to GTL Ref and what it takes to get that stable. In my experience, there appears to be a correlation between vCore and GTL ref. GTL Ref with the latest BIOS does seem to be a bit more finicky than vCore, such that GTL Ref may require a certain range for stability -- thus requiring a higher complimentary vCore.

With 1.1B4, I need a vCore of +2.5v to maintain stability. This is higher than I think is ultimately required by my chip, given its ability to bench at 4.05ghz at 1.55v. MSI is aware of this issue, and I'm hoping I'll be able to reduce it either with a future BIOS update and/or by playing more with GTL Ref when I have an opportunity to sit down and do extensive benching.

[silent-circuit]

I managed to kill Vista at some point, cause it won't boot at stock now. Time to play the reinstalling game.

[theYipster]

Everyone,

For those wondering, here is a thorough and popular explanation of what GTL Ref does.

http://edgeofstability.com/articles/.../gtl/gtl1.html

I've seen a lot of people set GTL Ref on this board when using dual cores, and I wonder whether or not that's necessary (as GTL Ref is intended for quads.) For those with overclocked dual cores, was setting GTL ref necessary in order to achieve stability? What happens if you put it on auto?

[loonym]

I've seen plenty of "DRIVER IRQ NOT LESS THAN OR EQUAL" and similar bluescreens, and multiple references to 'ecache.sys' if I'm remembering correctly. Nothing very helpful; just general instability I figure.

I think you are trying to overclock an already unstable os. Those look like trademark errors of running and or installing vista w/4gb (+) mem. Did you apply the hotfix?

I'm also wondering if anyone here has or is using bios A7510NMS.116? I received it to test with my 9600gt/XP64 issue. It still has the multi issue with kentfield and I haven't done any oc except my standard conservative 3.2 so I don't know much about that aspect but I'm curious if it could solve the multi problems people are reporting using 45nm quads.

[silent-circuit]

I think you are trying to overclock an already unstable os. Those look like trademark errors of running and or installing vista w/4gb (+) mem. Did you apply the hotfix?

I'm also wondering if anyone here has or is using bios A7510NMS.116? I received it to test with my 9600gt/XP64 issue. It still has the multi issue with kentfield and I haven't done any oc except my standard conservative 3.2 so I don't know much about that aspect but I'm curious if it could solve the multi problems people are reporting using 45nm quads.

Please put that up (.116) somewhere, as it seems to be more recent than anything I currently have access to.

Also, the OS was unstable at that point, but I reinstalled, patched up to near current via SP1 (which so far as I can tell includes the hotfix, and the hotfix returns "not applicable to your system" after installing SP1) and tried to OC again only to get the exact same error within 10 minutes. Not happy with this board at all.

I'm also running in to a number of USB related bugs which hamper use even at stock. One of my keyboards (Razer Tarantula) wouldn't even be recognized as a keyboard by the BIOS until after a firmware update. Now if I plug anything in to the hub on it it still isn't. My other keyboard (Apple Pro Keyboard) won't be recognized if I plug /anything/ else in to the other USB ports at the same time as it. The only way I've managed to flash my BIOS is by connecting my USB stick to one of the ports on the hub on the Apple Pro. These issues are persistant with every BIOS I've tried (1.0, 1.1B4, 1.1P5). What the hell, MSI? This is I haven't had to deal with since USB was new! These keyboards work with literally every other BIOS I have ever tried them with, so what are you doing that's so different? (and clearly... well... wrong.)