What is your cooling setup? The P7N definitely requires decent cooling to keep it stable especially for the northbridge.
What is your cooling setup? The P7N definitely requires decent cooling to keep it stable especially for the northbridge.
I don't have any fancy cooling for the motherboard, standard fan set-up in the Cosmos S, and getting a NH-U12P for the CPU. I'd stick a small fan on the Northbridge, but both my current CPU cooler and my GPU prevent me from mounting one.Quote:
Originally Posted by stickboy00
A decent cpu cooler makes a huge difference.... the Xigmatek I have made a world of difference with the cpu temps. You also really need northbridge cooling to get decent overclocks with the P7N (at least on mine anyway). Without the 120mm side case fan my rig was very unstable when trying to overclock past 1450 FSB. With the side fan, it easily does 1600 and its rock solid. I also have to give the NB lots of juice. At 1600 i need at least 1.425 to be orthos stable. I give it 1.45 and as long as its not 100 degrees outside, temps are under control.
Try increasing FSB term to its max amount and giving SB some more volts. And loosen up your memory timings so you don't have to worry about that being an issue and then just raise them once you get it stable.
Of course, if you've done all this already, then you just may have a dud of a cpu and/or board. It happens.
http://img190.imageshack.us/img190/5...t5cachemem.pngQuote:
Originally Posted by ijyt
I have CM cosmos S too, no added fan on NB /all stock cooling on board/.. Cooler on CPU was some copper Zalman horizontal to board .. NB on high temps /70-75 in game/ but stable.. I installed all 3 top fans and 3000rpm fan with reg on back/usually 2000-2400/.. 2x fan on disc cages from the front..
I think that an HR-05 would make a huge difference on this board, tbh. You should atleast replace the TIM and slap a fan on the northbridge sink if you start pushing the Northbridge.Quote:
Originally Posted by ijyt
Also, I have read through a good part of this thread but does anyone know of anyone overclocking a quad with 8gb of ram with any sucess?
If you want to reach 3.0ghz with Q9450/P7nD, that's possible with no extra voltages. If you are going to go beyond 3.5Ghz, stay away from this combination. Overclocking capabilities are a joke compared to other solutions. NB temps are painful, and high FSB's (Q9450 will need) require high vNB to run proper. Very limited vFSB on 1.313. Hope it helps.Quote:
Originally Posted by ijyt
P.S.: I have had Q9450/P7nD and burned two boards.
Interesting quad-cores with 8 GB of RAM should come up now. I've got mine almost stable. Perhaps people have some ideas I could try. This is my current set up:
Q9450 at 3.5 GhZ (1750 FSB)
8 GB 1066 RAM at 875.
CPU Voltage: .125
RAM Voltage: 2.1
NB Voltage: 1.325
NB GTL Ref: .865
FSB Term: 1.25
I've tried turning up the CPU voltage, the NB voltage and both simultaneously to no avail. As well as messing with the FSB term both up and down. From what I've seen, changing anything from these settings results in more frequent crashes. As is, I can usually make it through ~1 hour of 4-core primality testing before an error. I'd like to get this entirely stable (or, better yet, entirely stable at 3.6) does anyone have any ideas for what I should try from here?
Temps certainly aren't a problem. I have a good CPU cooler, each core stays below 60c under load. Two fans on the NB a 40mm blowing air towards the CPU cooler and a 50mm blowing air "down" towards the motherboard. Case has good airflow. "MCP" (I've read that this is both southbridge and northbridge and thus don't know what it actually means) does report rather high temps under load, sometimes hitting 71c. But that still seems reasonable.
I also don't think it's a RAM issue as I have loosened the timings a bit. (Worth noting that it won't do 1066 at stock, though.)
My Q6600 on this board is very picky about what settings I use, it took me a month of testing to get just the right combination. Most people won't have that much trouble as I feel I got unlucky with my chip, but I can share some of what I've learned about the board's nuances.
I find that FSB Term directly relates to the FSB; you will need a certain amount to do a certain FSB. The NB voltage, on the other hand, seems to have an ideal value whereby a notch higher or lower can cause instability for a given FSB Term. For example, max FSB Term allows me to hit 3.15, but only one value for NB voltage is stable at those settings. If I lower my FSB, it makes sense to lower FSB Term as well, but then I find the NB voltage reacts differently. I have tried both 4GB and 8GB of RAM with my sets, and both work at the exact same settings, but I do require a higher base NB voltage with 8GB.
Ergo, you want to take a pseudo-traditional route of finding the proper FSB Term and NB voltage combination for the given FSB by removing any potential interference from other values (set them to obviously stable settings). The way I managed this was to max out FSB Term and then push NB voltage up until it went from getting more stable to getting less stable (or no more stable), then at the best value I lowered FSB while maintaining the two. Your case would be a bit different, assuming you can hit the desired FSB, in which case I would do this in reverse - lower FSB Term until more instability is seen, pull back and then work the NB voltage in the same way.
Second, the GTL voltage was critical for me as only one value worked well with the other settings. If you now have the good NB voltage and FSB Term, you can then slowly raise GTL at your desired CPU speed until the system is stable.
Well this is weird. After installing the Q9450, one of my RAM sticks dies. Thinking nothing of it, I go to i37 LAN, where my CPU apparently idles at 60C and freezes up the system while gaming or stress testing.
I run MemTest86 to check if it might be the other RAM stick, no issues.
I buy two new fans at the LAN to try to get more airflow, but it doesn't help much, so I go ahead and buy the NH-U12P and mount it at the LAN. I add an extra Noctua P12 fan to it, and use MX-2. What do I get? 50C idle temps with the Q9450. I pass this off as warm ambient temps, but the system still crashes.
I'm back home, my room's ambient temp is aroun 18-20C. I boot up the system, and the bios says the CPU is idling at 28C. However, CoreTemp, RealTemp and Hardware Monitor all report 50C idle temps. The system also freezes during Test 3 of Prime95's Small FFT test. The CPU heatsink is cool, and the Northbridge isn't so warm I can't hold my fingers on it. Any ideas?
I dropped my RAM (http://www.corsair.com/_datasheets/T...-6400C4DHX.pdf) timings to 5-5-5-18, and downclocked it from 800MHz to 667MHz. This stopped the crashing during Prime95.
However, there's still the issue of temperatures.
CoreTemp, RealTemp, Hardware Monitor all report idle temps of ~51C and load temps of ~65C.
HOWEVER, MSI CoreCenter, which I presume takes readings from the BIOS, shows idle temps of ~28C and load temps of ~40C. What?
Core Temp and Real Temp read sensors within the cores inside the chip, while most motherboard CPU sensors (including MSI's) read the temperature outside the chip. Typically, there is a difference of 5-10c on a well cooled chip, and that difference is largely attributable to the IHS that covers the base CPU silicon.
You may want to check the mount between your processor and CPU cooler. Are you sure it is seated properly? Is the correct amount of TIM used? You may also want to inspect the chip's IHS. The flatter the IHS surface, the more effective your cooling system will be at pulling heat from the CPU. If your IHS has convex or otherwise odd shape, it may not be properly transferring heat between the cores and your cooler, resulting in the temp differences.
Just checked the cooler, it was seated correctly. But I reseated it anyway to check the paste, it was spread evenly and not too thin/thickly. The CPU seems flat enough. I'm going to blame this on faulty sensors, so I'll RMA the CPU anyway, just to be safe. If the new one they send harbours the same temperature readings, I'll be at a loss.
Also, system freezing during Prime95 stopped when I downclocked my RAM to 5-5-5-18 timings and 667MHz frequency, so I'll be RMA'ing both the dead and apparently-not-dead-but-somehow-faulty sticks.
Any chance of the MX-2 paste having an issue? :p:
I would play around with vNB and the nb GTL since that could be due to MCH stability and not your ram. Also set your trfc to 55 and the vDRAM manually.Quote:
Originally Posted by ijyt
Wut.Quote:
Originally Posted by BababooeyHTJ
This board is kind of famous for not being stable at "auto" settings. Most of the people having those problems are on Quads like yourself. Quads are more stressful on the northbridge than duals. I would play with the northbridge voltage and the nbGTL to see if you can get things stable before sending your ram in for an RMA. At least run Memtest one stick at a time.Quote:
Originally Posted by ijyt
I already ran MemTest why would I not RMA both sticks if I know they're dead?Quote:
Originally Posted by BababooeyHTJ
I think he was just saying that the RAM may not be dead and to test them one stick at a time. And also that Memtest doesn't really tell you much if the board is unstable to begin with.
Thanks for the suggestions, XavierMaxx. I've since tried them, but haven't had much success.
What I have seems to be the most stable NB / FSB term I can get. Changing the NB and CPU GTLs doesn't seem to help from here. I'm starting to think it may be one of the voltages that I haven't touched that's causing the issue.
I've ruled out Southbridge, since that one's easy enough to test. So I'm thinking I should start with either: NB Ref V, DIMM Ref V / Mem Term V (almost certain it's not the RAM, so I doubt I'll bother with these,) or PCI Expander PLL V.
Maybe PCI expander (I would hope that I wouldn't have to change reference voltages,) but my BIOS (1.23 performance, I've had serious cold boot issues with 1.3 and 1.4) doesn't always display numbers as the first character for that voltage. Which is somewhat concerning.
So, anyone have any tips on what I ought to try from here?
what is the deal with this board?? i cannot overclock well at all with my Q6600 max is 3.6 and not stable for benching latest beta bios and on my m aximus 2 formula i cvan bench stable at 4.1GHZ
I've tried multiple Bios revisions, but I can never get anything stable. Right now, I'm running v1 bios with the following settings:
FSB: 1333
Dram: 889
CPU Ratio: 9
CPU Volt: .0500
Dram Volt: 2.20
NB Volt: 1.35
SB Volt: 1.550
NB Mem Ref Volt: 1.095
Dimm Volt: 1.425
Mem Term: 1.050
CPU Ref: Auto
NB GTL: 0.842
FSB Term 1.230
With the newer bios, these figures are converted to meaningful numbers. However, for the life of me, I don't know how to properly translate what I have to updated figures, and then play with it from there to achieve stability. Can anyone help? I've had this thing over a year and have pretty much given up on using a good, updated bios because I can't get the thing to run stable, even with default values and not OC'ing at all...
Overclocking quads on intel chips is extremely easy, ANYONE can do it, at least normal levels, but nForce chips 780i and lower are much more difficult, and won't overclock as high as intel ones.Quote:
Originally Posted by overclocking101
I'm having odd issues with this board. I'm using a Q6600 which would appear to be a good clocker (3.6 with only +0.1v) and 2 x 2gb Reaper X 800mhz 4-4-3 sticks. Problem is I cant get windows 7 Enterprise x64 to install even at stock speeds? This is a second hand setup that was fully working with Vista 64. I get a reboot at some point during "expanding files" during setup.
I can however pass memtest @ 3.6/1600/800 4-4-3 using +0.1v cpu, 2.10vddr, 1.425vnb and 1.312vfsb. I can lower vnb, vfsb and vddr a fair way and still pass memtest - I've tried all sorts of different voltage combinations.
I've also tried ram @ 5-5-5-18-2T, eist on or off, HPET on or off, changed secondary ram timings to slackest possible, moved around hdd and dvdrw sata leads, changed PME wakup from bios to OS and back again, PCI latency timer @ 32 and 64, with and without soundcard (sb audigy). Only thing left to try is to disabled the network cards i suppose?
If I have the FSB at 400 it fails quicker than with the FSB at 266. It made it all the way to 97% at 266, only lasts to 35-45% at 400mhz.
Any other ideas I can try? Wits end!
I even went to replace it with a P5B Premium which seems to have expired whilst in the box on my shelf as it now wont post, just spins fans at max speed.
I'll check this thread later tonight, gonna swap ram and attempt another install like that but if nobody has any more ideas i'll try with Server 2008 x64. 7 would be ideal though as I need to run VMs and if necessary game on it (ie if my main gaming rig is in bits).
Odd, I would try a different DVD drive or try updating the Firmware. Also what bios are you on?:shrug:Quote:
Originally Posted by PiLsY
Not had chance to look at it last night, will try again over the weekend. Interesting point about the dvd-rw - its a brand new sony so should be good but... Also burned another disc just incase its that.
Bios version is a variety of 1.3 - not sure exactly which. Theres no way im tryng a bios flash with it failing to install windows though, if its instability causing the install fail I can kiss goodbye to the board if I try a bios flash on it.
Had to stick GTL voltages to maximum for both nb and cpu, then it installed win7 but failed prime95 almost instantly.
Managed to get it stable enough for a bios flash to 1.40 whgich has made a big difference - enough for it to be stable at default clocks and auto voltage settings anyway. No attempt at overclocking yet.
Real weirdness - its going into standby regardless of load. It kept going into S1 shutdown while running prime95 on all 4 cores. Had to change the win7 power options.
Also notice that with EIST on in win7 im getting it clocking back to 7.5x and 8.5x on occaision. Didnt think half multis were supported before wolfdale?