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

[shawn1998]

I would like to post more results with this board, but I seem to have a serious cold boot issue when I try to overclock my QX9650. I suppose I just posted my results, then...

[theYipster]

Hey Shawn,

Can you post the settings in the cell section of your bios? Thanks

[shawn1998]

Sure. First number is what I am currently at, the second in () is what I tried.

DOT - Disabled
EIST - Disabled
System Clock Mode - Auto (Tried linked and manual)
CPU Ratio - 9 (9.5, 10, 10.5, 11)
PCI-E Freq - 100
Auto Disable DRAM/PCI Freq - Disable

CPU Volts - 0.025 (all the way to .1000)
Dram Volts - 1.95
NB Voltage - 1.275 (up to 1.35)

All the rest at auot for now. I messed with FSB Terminator, had it to 11.


Nothing I change with those settings make a difference in the ability to successfully cold-boot when the multiplier as set to anything but 9, or the FSB is set to anything but 1333.

[theYipster]

I wonder if anyone else has this issue. It's a real conundrum. My last suggestion is to try 2.0 or 2.1v on the RAM and 1.4 or 1.45v on the north bridge. Then try raising the FSB. My guess though is that it's likely BIOS related, and despite what MSI support tells you (and I strongly believe they're wrong, and would make a strong case against them,) I would imagine that their developers are working on a BIOS fix.

[shawn1998]

I truly do believe that it is bios related. I tried upping the fsb just 10mhz, that's it, and bam, cold boot city. It's crazy.

[theYipster]

That's unfortunate. Well hopefully you can return the board... the next question would be is whether or not this issues occurs on anyone else's with a 45nm quad.

[r1ch]

That's unfortunate. Well hopefully you can return the board... the next question would be is whether or not this issues occurs on anyone else's with a 45nm quad.

You think it could be a famous nVidia FSB hole? Try skipping further ahead.

Also, go extreme underclock on the rest. Drop the RAM down to nice and easy 5-5-5-18 timings and 1:1 ratio to comepletely rule it out. Also might not hurt to take all but one stick out.

If you get it working at higher FSB at some point, then you can one-by-one start adding the RAM and settings back to see what's causing the issue.

[theYipster]

You think it could be a famous nVidia FSB hole? Try skipping further ahead.

Also, go extreme underclock on the rest. Drop the RAM down to nice and easy 5-5-5-18 timings and 1:1 ratio to comepletely rule it out. Also might not hurt to take all but one stick out.

If you get it working at higher FSB at some point, then you can one-by-one start adding the RAM and settings back to see what's causing the issue.

Good point. These steps are certainly worth trying.

[shawn1998]

I have actually tried it with one stick, each stick by itself and different ram slots. I have two motherboards arriving this afternoon, my EVGA 780i from rma and a new Gigabyte EX38 DS4 (as a back up). I will use them to figure this out.

[lee63]

I have a QX9650 and this board, the multi shows the right value in CPUZ. I just cant seem to run my Ram at 1066, what its rated at.

[loonym]

My setup doesn't seem to be very tolerant of big nb volt increases. By experimenting with the gtl voltages I have had modest success with fsb increases. I think msi is doing a decent job at trying to address specific user issues through beta bios releases, the most recent being A7510NMS.114 or .1B4 as it's otherwise identified, dated 3/2/08. I've only started to play with this but it is stable and seems to have the vdimm, at least, default to a higher voltage. I think this is either an attempt to fix coldboot or perhaps for those populating 4 slots.

[BKA]

Hello all,

I was wondering if somebody could tell me which driver (folder) on the CD for Vista are for the esata ports? I am working on a Vista SP1 disc with drivers slipstreamed and would like to include the drivers for the eSata ports. I understand they are part of the jmicron chipset, but the IDE\JMicron folder doesnt appear to include the eSata port drivers from the looks of it. Any help would be appreciated.

EDIT: Figured it out!

[theYipster]

Hi all,

Thought I'd give a quick update on my progress. First, my apologies for the delay in updating this thread. My busy schedule from last week extended into the weekend, and as such I haven't been able to sit down with the rig until this evening.

My eventual end goal remains the same -- 3.8-4.2ghz with a Q6600 G0 (L737B277.) Currently, I'm trying for a stable 3.6ghz. I'm not stable yet, and it is proving to be a bit of a challenge, but I do remain optimistic. For now, the challenge is due to my own lack of experience with the board's voltage settings, which do work quite differently than the settings on the reference boards. As I moved up from 3.0 to 3.6ghz (9x400 FSB) I didnt't experience any FSB holes, nor did I ever have to clear the CMOS from a bad overclock. I've been able to successfully post from both warm and cold boots, and Windows loads to the desktop with as little as 1.325 vCore. The challenge right now is getting the rig Prime or OCCT stable at 3.6ghz.

Here is what I've found so far:

The board is completely stable at 3.0ghz (9 x 333mhz FSB) with RAM 1:1 at 667mhz. The only voltage set manually for this setup was RAM, which was bumped up to a factory-spec 2.0v. The chip is running on stock voltage (1.2375v) and all other voltage settings are on AUTO.

Right now, my 3.6ghz settings are 1.45v in the BIOS (which I expect may be more than eventually needed,) 1.4v for the NB (SPP,) 1.55v on the SB (MCP,) and 15% on the FSB Terminator Voltage. RAM remains at 2.0v and is now running 1:1 with the FSB at 800mhz. Timings are at a JDEC default of 5-5-5-15. What I am finding is that while I can run Super PI, any of the Everest benchmarks, and 3D Mark 06, Prime95 will fail around the 10 minute mark. This remains true even if I up the vCore a step or two. What's interesting is that in all my testing tonight, Prime 95 and OCCP have not thrown one error themselves, nor has Windows thrown a blue-screen. Instead, with each attempt at 3.6ghz, the system will eventually completely freeze (with mouse cursor and all) or simply reboot. This leads me to believe that the issue lies with one of the many motherboard voltages, likely NB or SB, but perhaps with one of the many reference or terminator voltages. Lastly, throughout my testing, the motherboard temp hasn't exceeded 55c, and the temps of the CPU cores haven't exceeded 60c. After 10 minutes of prime w/ 1.378v (on load,) temps are in the mid to high 50s. Compared to my old G0 (an L734A,) this is a bit high. However, ambient is a bit warmer than usual due to the warm weather outside today, and the AS5 hasn't yet cured.

Seeing that there is a lack of information on OCing the P7N Diamond, there is not much else to do but to continue to test different combinations. However, I would appreciate any and all insight and/or suggestions as I continue. For those familiar with the voltage settings on this or a similar board, I do have a couple of questions:

1. It's clear that as far as vCore goes, an AUTO setting in the BIOS translates to stock vCore, regardless of clock speed. This is different from the behavior of any of the reference boards, which will guess an appropriate vCore based upon your overclock when vCore is set to AUTO. My question is whether or not this is also true for the other voltages, such that regardless of FSB overclock for example, FSB voltage will remain at stock when set to AUTO.

2. On the reference boards, FSB voltage is set directly like any other voltage (i.e. you set the amount you want to pump--1.3v, 1.4v, etc.) However, on this board, FSB Termination Voltage (which is the same thing,) is set by choosing an integer (1, 2, 3, 4, 5, etc.) There is no indication in the BIOS as to what this integer setting refers to, although after reading some P6N threads on the MSI forum, I gather it means percentage. Does anyone know what a setting of 12 or 15 relates to in actual FSB termination voltage?

3. In fact, all voltages besides vCore, NB, SB, and DIMM are set using similar integers which bare no indication to what they mean or any relation to an actual voltage being pumped through the component. Does anyone know how to translate these voltage settings, and can anyone recommend proper gtl, reference, and/or termination settings for a 400-450mhz FSB?

Hopefully these are questions that I will be able to answer on my own as I strive to find a successful combination of voltages for a stable 3.6ghz and beyond. To others who have this board, please post your benchmarking results as you retrieve them. The more information we can gather about what it takes to get a stable high OC, the better off we all are, as are people considering this board as an alternative to the reference designs.

Many thanks!

EDIT: One more thing I forgot to add -- an unfortunate, if not unexpected, finding. My board is showing to have as much vDroop as a reference board w/o the pencil mod, yet there is no BIOS setting to disable it (like Loadline Calibration does in the ASUS boards.) Hopefully MSI adds it to a later BIOS, or that someone finds a similar pencil mod. Please share if you are experiencing similar vDroop. If not, it may be just an issue with my particular board.

[x-built-Stan]

Hey everyone,

I've been working on overclocking my newest build. Here are the specifics:

Project Mercury - CPU Q6600 (L742A473) @3.4 GHz
MSI P75 Diamond (780i)
2x1GB OCZ PC2-8500 Dual CH SLI-Ready EPP
2 MSI 8800GTX 768MB 610MHz
1 LiteOn 2x Blu-ray Disc Triple Writer
2 500GB Hitachi HDD
Silencer 750 Quad Power Supply
Thermaltake Armour LCS Case
D-Tek Fuzion CPU Waterblock
Windows XP

Like theZipster, I am trying to get a Q6600 up to 3.6GHz on the MSI P7N Diamond board. I started out at 3GHz (1333 FSB) which was easy to achieve at stock voltages and memory settings. I slowly worked my way up to 3.4GHz increasing the CPU, Memory, and Northbridge Voltages. The last Prime95 run made was stable for 19 hours and 9 minutes before I stopped the test. (eventually I'll do a full 24hours). I have attached images of my bios and Prime95 screens for your review.

Temperatures during the 19 hour run got to 61C on the CPU and 52C on the system sensor. No doubt driven by the increased voltage on the CPU and NB.

Additionally, I have made several attempts at 3.6GHz (1600MHz FSB x9) and all have failed. Typically the failure occurs within ten minutes with one of the Prime95 threads dying.

Please review my data and let me know what you think.

[ban916]

I'm thinking of getting one of these boards with an e8200 or an e8400 and run it on air. Anyone running any of these in the board, Also what would be the best air cooler that will fit with the northbridge cooling?? This will be going fo max air overclock. Thanks

[Overspeed]

I am running this board with the E8400. So far, I have not been able to achieve anything past 3.3Ghz regardless of voltages. It will boot at 1545FSB but is not stable. It will not boot at 1550FSB. I have my 1066 ram at 2.1v, vcore at 1.35v, NB at 1.5v. I have tried increasing/decreasing voltage values across the board without success. I will wait for an official BIOS update in all hopes, but right now I kinda wish I went with an Intel board. :o. I bought the E8400 specifically to reach 4.0Ghz as I have read so many guides online on how easy it is to get there.

[theYipster]

x-built-stan, why is your FSB Terminator setting so high? From what I've read at least for the P6N, a setting between 12-15 was usually needed for higher ocs.

[x-built-Stan]

theYipster,

I started out at 10 with my terminator voltage at 3.2GHz. As I ran my prime95 tests for longer and longer periods one of the threads would die. I started upping the terminator voltage which made the system more stable (longer successful test run). By the time I got to 3.4GHz, I was pushing 60 on the terminator with succesful test runs of more than 10 hours. My last run, as noted previously was 19 hours and 9 minutes before I stopped it. Eventually I would like to get the terminator voltage down if possible.

I also have a MSI 6PN Platinum with an X6800 in it. I have never had to run the terminator voltage up as high on that board.

Still working and tuning. I should have more results later in the week. I'll add a picture of the rig later. Any insights that you have would be welcome.

[theYipster]

The fact that you're getting Prime to error out and not system reboots suggests that the processor isn't being fed enough voltage, but that your motherboard is capable, hopefully, of supporting high Quad core OCs.

What is your processor's VID? This number alone can be a very good predictor of how much voltage your processor needs to be stable at high speeds. You can read VID from CoreTemp, and if you're using Vista x64, simply search google for a script to get it working w/o having to disable driver signing.

Also keep in mind that processors may need a bit more voltage on nForce boards than on Intel boards to reach the same speed.

[ban916]

I am running this board with the E8400. So far, I have not been able to achieve anything past 3.3Ghz regardless of voltages. It will boot at 1545FSB but is not stable. It will not boot at 1550FSB. I have my 1066 ram at 2.1v, vcore at 1.35v, NB at 1.5v. I have tried increasing/decreasing voltage values across the board without success. I will wait for an official BIOS update in all hopes, but right now I kinda wish I went with an Intel board. :o. I bought the E8400 specifically to reach 4.0Ghz as I have read so many guides online on how easy it is to get there.

Damn are you serious?? 3.3Ghz?? damn this boad should be able to push it a lot higher than that...Maybe I'll change my mind about the board...anyone else with an e8400 e8200 or e8500 on this board get to 4Ghz??

[x-built-Stan]

The VID is about 1.42v (CPU voltage increase 0.175 in the bios). I agree about VID and if I was running a dual core instead of a quad I would focus on that aspect. However, I understand that the the quads use the FSB to "communicate" between the two dual cores. That is why I have been focusing on the FSB terminator voltage. I'll try upping the CPU voltage a little and see if I can drop the terminator voltage some.

[theYipster]

That's most likely not your VID. On a retail G0 Q6600, VID ranges from about 1.2 to 1.325 when EIST (speedstep) is disabled. I've never seen a G0 Q6600 with a VID lower than 1.2 or higher than 1.325, although according to Intel's spec they could exist.

VID is the default voltage supplied to your chip when running at stock speed, assuming you leave your chip's vCore BIOS setting on auto. The VID is imprinted on your processor, and is unique to each chip. Basically, it tells your motherboard what the processor needs for voltage at stock operation. On the MSI P7N Diamond, notice that vCore is set by increments to VID, where as on other boards you simply set a straight voltage (as done with the NB and SB on the P7N.) Therefore it's even more important to know what your chip's VID is on the P7N. Otherwise you don't know how much additional voltage your supplying your chip to reach a given OC.

Knowing VID on a quad core is just as important in gauging what voltage you'll need for a given OC as on a dual core. FSB Termination Voltage is not nearly as important as vCore on a Core 2 chip.

If you don't have Core Temp, download it, open it up, and read the number next to VID. Paste that here, and we'll have a good idea as to whether or not you should be tooling with vCore instead of FSB Termination Voltage.

[x-built-Stan]

I have it. But I'm at work right now. I'll post it later tonight.

[theYipster]

Sounds good. If you know that your BIOS is feeding 1.42v to your chip, and that you've selected .175 increase for vCore in the BIOS, then you can reason that VID is or should be around 1.245. This isn't bad, but as far as 3.6ghz on an nForce goes, you'll likely need more than 1.42v set in the BIOS.

The issue is one of vOffset and vDroop. On many motherboards, the voltage supplied to your chip at idle is less than what you've set it to in the BIOS. This difference is vOffset or vDrop. When you run Prime 95, you will likely notice that the voltage supplied to your chip drops further. This is vDroop. You can monitor vOffset and vDrop using CPU-Z. See what the voltage is after your Windows desktop loads, and note the difference between that and what you've set in the BIOS. Then, with CPU-Z open, start Prime95 and watch the voltage dip more. The trick is to supply enough voltage in the BIOS so that voltage doesn't drop below what is required for stability at load. As you increase vCore in the BIOS, you can expect the vDroop and vOffset deltas to increase slightly.

One thing I discovered about this board is that vOffset and vDroop are about as noticeable as they are on the reference 680i/780i boards, which is to say that they're quite noticeable. Unfortunately, I don't know of a way to reduce vDroop on this board. ASUS provides a BIOS setting called Load Line Calibration that reduces vDroop on their boards. On the reference boards, shading a particular resistor with a #2 lead pencil eliminates vDroop and vOffset almost completely.

[Overspeed]

I've spent another couple hours trying to reach a stable OC, so far w/o any luck. The highest I have been able to OC this E8400 stable is 3.21Ghz. I'm really stumped, and I don't think its a voltage problem. I have tried bringing to voltages up to 1.45 on the core and maxed out the NB. SB increases so far have no effect. RAM is OC series rated at 1066. I've seen the E8400 go to 3.6Ghz with no voltage increases and stock heatsink on Intel boards. 4.0+ was easily acheived with minor core increases only.

For the sake of trying, I tried the D.O.T. feature and set to 10%. I put the CPU under heavy load and monitored CPU-Z. At no point did the core increase past 3.0 and then all of a sudden complete crash. This was supposed to be MSI's new flagship. Shrug.