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

BababooeyHTJ, I am having mixed results with your settings. At first, PC posted ok and loaded Vista. Shortly after, BSOD complaining about win32k.sys. Leaving the CPU GTL & FSB TERM alone (0.866 & 1.313), I tried varied setting on the CPU, NB & SB. Having the CPU at +.025000, NB at 1.375 & SB at 1.6 I didn't get any more BSODs in Windows, and prime ran 1 hour without error. Was looking good until I ran a SoftSandra benchmark on the Memory & Cache. 17% in got BSOD again. Rebooted using same settings and ran 3DMark06, BSOD at Firefly test #2. I've pretty much tried all variations on the CPU & NB without long term success. Keep getting BSODs on 3DMark06. I have tried upping the CPU and NB a fair amount, and that seems to make the system less stable. Odd, that my VID is very low at 1.0375. If I don't try to run Sandra tests or 3D games, windows seems stable enough and I was able to produce the following benchies:

http://alpha.customfitonline.com/cpuid.gif

http://alpha.customfitonline.com/everest.gif

http://alpha.customfitonline.com/hm.gif

http://alpha.customfitonline.com/coretemp.gif

What was not noted above is that during heavy load, my CPU cores never exceed 58C so I can't imagine temps on the CPU is an issue. Despite my greatest efforts I can't seem to get stable for 3D applications or heavy CPU Cache tests. Going above your 3.85Ghz using any settings seems to fail instantly and will not post. Most BSODs are related to win32k.sys & others are just blank or not complaining about anything particular.

That really sounds like a memory issue. Have you run Memtest 86+ ? The only difference I see in our rigs is that you have 4 sticks of ram vs my 2. Your ram's stock voltage is 2.0. It might not like 2.1. I would definatly run memtest.

Well after a few more hours of testing, I think I must conclude that this setup is not going to be an adequate overclocker. Any NB setting over 1.375 causes high instability. I have not been able to get through a 3DMark06 run without error. Perhaps I just have a bad batch of E8400 or the NB is cooking, I can't be certain. As mentioned before, I can run windows stable enough, even prime won't crash after a long run, but Sandra benchies and 3Dmark06 die everytime. Here are the settings I used for basic windows stability:

FSB: 1700
Ram: 850
Dram Timing: Auto
CPU +.025000
Ram: 2.05
NB: 1.375 - 1.4
SB: Auto
CPU GTL: 0.866
FSB Term: 1.313
Rest Auto

I don't imagine the P7N BIOS is going to get much more refined than the 2.0, at least not for dual cores. I'll put this to rest for now, unless anyone has suggestions on what might be causing the failures. Thx.

OS

Quote:

---

Originally Posted by BababooeyHTJ

That really sounds like a memory issue. Have you run Memtest 86+ ? The only difference I see in our rigs is that you have 4 sticks of ram vs my 2. Your ram's stock voltage is 2.0. It might not like 2.1. I would definatly run memtest.

---

My sig is old. I'm no longer running the 4x1GB Axeram, now running 2x2GB Mushkin PC-8000 Ascent, brand new, 1 day old. I will nevertheless, run a series of memtest. How long am I suppose to let it run for accuracy?

I ran it overnight. It could also be a corrupted install from bad ram timings. If memtest passes I would format or at least format a partition and install windows and some programs there. Try you dram v @ 2.1 and nb v @ 1.4. From what I hear you should let it run for at least 7 passes. I just ran it over night.

Quote:

---

Originally Posted by BababooeyHTJ

I ran it overnight. It could also be a corrupted install from bad ram timings. If memtest passes I would format or at least format a partition and install windows and some programs there. Try you dram v @ 2.1 and nb v @ 1.4. From what I hear you should let it run for at least 7 passes. I just ran it over night.

---

If I had a corrupted installation, wouldn't that mean windows would continue to fail after turning off OC and running at stock speeds? So far, when I set back to stock, all programs and benchies work fine without error or BSOD.

Well, memtest has gone 5 rounds so far with no errors. <shrug>???

Quote:

---

Originally Posted by Overspeed

Here are the settings I used for basic windows stability:
CPU +.025000

---

Is this .25 or .025? What does this make load vcore?

Ok, so I let memtest run over night, 8 passes on both ram chips, 4 passs with only 1 ram chip installed, no errors detected. loonym, during my initial trials, when I had windows semi stable, cpuid reported my load vcore in windows was 1.384, that was with a +.25 increase. When not overclocking, my core is at 1.128. Now that memtest has run successfully, I tested running all benchies NOT overclocked ... all Sanda benchies ran with no error, 3DMark06 ran 10 times with no error, so I can't believe I have a corrupt windows install. During my OC trials, I gave all voltages adequate increases ... something just doesn't like it when pushed far. The NB HATES any voltage higher then 1.375. This morning I tested putting a spot fan on the NB, it didn't change anything. I'm not sure what else try try?!?! My CPU temps are very reasonable during the OC ... never passing 58C under heavy load. I am running 3 sata HDDs and I gave the SB a fair increase at one point as well.

If you read some of my early posts in this thread I had the same issue with the nb not liking any voltage over perhaps 1.35-1.375v. It wasn't until I pulled the chipset pipe, cleaned everything thoroughly, and reinstalled with ceramique that I started being able to really push my board. It's unfortunate there's no way to accurately read nb temps.

Ive come to the conclusion with the help of xytrius that my board has faulty CMOS chip im just unable to shut down anymore guess its time for rma.

Really sucks that a lot of boards don't used socketed bios anymore.

My computer's been running very stable at 3GHz so far, but every time I boot up my computer after leaving it for a while, it either won't POST (so I have to clear CMOS) or Windows fails to boot. I've been slowly upping voltages every time this happens, but I'm wondering if this is normal behavior. I don't remember the exact settings offhand (might be off by a couple thousandths), but I'm pretty sure I'm running 1.25v NB, 0.831v CPU GTL, and 1.223v FSB Term.

Edit: Still having trouble booting up after the computer's off for a while. It either doesn't POST or Windows won't boot. I'm up to 1.25v NB, 0.84v CPU GTL, and 1.225v FSB Term. Don't want to go any higher with FSB term because my cpu is running at 1.225v. Not really sure what the deal is with this problem; settings work great at first and later stop working altogether, like the hardware is degrading every time I run it.

Edit 2: And the random reboots are back. I guess I'll try messing with NB and CPU GTL voltages some more, but am really getting discouraged about this boardset. My computer can pass memory tests and survive long gaming sessions and stability tests, but it just keeps crashing under normal use and is unpredictable when booting up. The similar problems other people have and overall lack of success stories (sans loonym :P) is making me wonder if I should jump ship for another board. I've yet to experience stability with any setting on this board, stock or otherwise. :/

Just figured I should mention this...last night I pulled the Circu-pipe cooler (per Loonym's suggestion), cleaned it real well, and reseated it with some nice thermal compound (knockoff AS). Got everything put back together and got to clocking; needless to say it made a world of difference. My CPU easily is stable at 3.6ghz (400fsb); I upped my vCore a modest .125v for about 1.25v after vDroop, FSB term is at 1.313v, CPU GTL at 0.875v, NB at 1.375v and SB at 1.55. My RAM is at 800@4-4-4-12-1T and 2.2v (stock). So, I would recommend doing the same to everyone who is having stability issues. So far I have been rock steady at 3.6ghz. Haven't "prime'd" it yet, but it was good for about 2hrs of GRiD last night :D Can't wait to push it further here later today!

Has anyone here ran the P7N DIAMOND with the arctic cooling freezer 7 pro, im just wondering if it misses the cicu pipe OK or if theres any attatchment issues. I just got one, waiting for it to arrive, so hope its fine.

bluspikez, did you have that real hard purple stuff on yours?

crucial33, that ac freezer will fit fine. That's not a real strong cooler so keep that in mind when you oc.

Quote:

---

Originally Posted by loonym

bluspikez, did you have that real hard purple stuff on yours?

---

Yeah, I did. Was figuring I would have to heat it up to get it off, but used some 3M adhesive remover instead (took some rubbing, but it came off :)). One of the chips (don't remember which, probably the expander) was barely in the thermal stuff that was on the heatsink. I left the pad that was on the MOSFETs, since it was actually decent stuff (not the purple gunk...was a light blue pad) and I figured that they didn't need the other paste that I had.

That being said, I think that reseating the sink is the key to success...just booted and am currently at 4.05ghz (9x450) w/1.36v vCore. Oh yeah, I never put a spot fan on it either (may do it eventually though). Gonna get to doing some benches and stability testing here in just a moment :D

bluspikez,

The nForce 200 chip barely makes contact. Same happened on mine too.

I had a pretty sketchy contact pattern on the nb as well as the nf 200 pci-e expander. Anyone doing this will also benefit from just tweaking the whole pipe unit into the proper shape and flatness too before putting it back.

Quote:

---

Originally Posted by AaronYuri

bluspikez,

The nForce 200 chip barely makes contact. Same happened on mine too.

---

Right...its pretty sad to know that most, if not all, boards come from the factory like that. That being said, I think that the heatsink does a good job of cooling the board once you reseat it with some good TIM. I've got mine at 450fsb now for 4.05ghz (~1.4v-1.41v vCore after vDroop) and it has been running prime for almost 2hrs now with no crashes, along with some UT3 and Crysis for about a half an hour each before that. :D I am happy with the newfound stability that came with reseating the circupipe. That being said, I wish I could run some 3dmark06 runs, but for some reason it doesn't like my drivers for my video cards and it isn't utilizing SLi properly...guess I'll just wait till tomorrow when nVidia releases the new ones to worry about that.

EDIT: The 4ghz run of prime just crashed on me...my temps were starting to get a bit high for my liking (close to 60C for each core) at that voltage, so I decided to back off and go back to 3.6ghz. I know I could get it stable at 4ghz, but I just don't want to pump the voltage I'd need for it (probably around 1.45v) to my processor right now. Running prime now to see how stable my 3.6ghz settings are....

bluspikez, loonym, AaronYuri

Might any of you offer suggestions on where to start in regards to pulling the pipe off. This is not something I've ever played with before. I assume I will need to pull all cards and cords/wires off the MB and pull it from the tower? I figure that part should be easy enough. Then theres the matter of new compound. I have read your suggestions as far as properly cleaning the surfaces. I will need to buy new professional compound ... and would value advice on make/type as well as proper application techniques. To take it a step further, if I already have the MB out of the tower, I will put a 40mm on the NB underneath the curved/arc'd pipe tied down with zap straps as we have seen a few users do who have posted pictures in this thread. I have a pretty good CPU cooler on here but it wasn't installed by me, so I'll pull it off as well and use professional compound or pads based on your advice. I'm looking forward to getting better results with this effort. I would hate to go into this not knowing what i'm doing and rip a few chips off the board or bend/damage the pipe while pulling it off :o

Overspeed,

Remove the board from the case. Push the pins through, then pinch them with your fingers on the back. It will release the push-pin, remove it. Do that 8 times over. The Circu-pipe should just fall off. If it doesn't, grab the NB base of the heatsink (not the loop), and twist - it'll come right off.

Rub the crappy TIM off with a tissue or something similar, then get every bit off with what you usually would. Check the Circu-pipe for any bent heatpipes. The SB and NF200 heapipes were bent up for me. Straighten them out if there's any kinks.

Get some AS Ceramique or some other high performance TIM (not AS5 because it has a curing time), put a blob on each chip (NB, SB, NF200. Leave the PWM if you wish), spread it with a razor or credit card. Place the Circu-pipe back on with the push-pins, or preferably, get some nuts, bolts and washers, then put them 'sinks on real tight.

Quote:

---

Originally Posted by AaronYuri

Get some AS Ceramique or some other high performance TIM (not AS5 because it has a curing time), put a blob on each chip (NB, SB, NF200. Leave the PWM if you wish), spread it with a razor or credit card. Place the Circu-pipe back on with the push-pins, or preferably, get some nuts, bolts and washers, then put them 'sinks on real tight.

---

How thick should the ceramique be before resteating the pipe? You mention using a razor blade, which in theory would scrape all the ceramique off (if the blade was pressed tight to the chip face). Should the pipe be forcefully bent in any way to better cover some of the chips that the pipe barely covered from factory?

I just use the tip of the tube to spread it, but remember all you need is a thin layer. If you use to much it'll just ooze out from the pressure of the sinks. You don't need much force to move the pipe into shape and flatness, more like gentle tweaking. You'll know when it's right. :up:

edit: I found that purple crap that's on there to be a real pain. It's hard like epoxy. I used alcohol and a plastic scraper.

Yeah...I don't use anything special to spread it out either, just a fingertip. Like Loonym said, the TIM on the sinks is a real pain; I was going to use isopropyl alcohol but found some other stuff in the garage that seemed to work better. Oh yeah, you may need a pair of pliers to squeeze the pushpins...I couldn't have done mine without it. As a side note, prime has been running for just about 8hrs now @ 3.6ghz and is still going strong. :D