Instability: Upgraded CPU (2180 to 8400), is Windows Reinstall Needed?

I had a perfectly stable 2180 (OC'd to 3.2 GHz), and upgraded to a new 8400. The 8400 is OC'd to 3.8 GHz, but I am having some stability problems, and haven't pushed the OC any further. 4.05 GHz would be nice eventually though. I am now trying to narrow down potential problems.

Does the OS affect system stability after switching processors? Would a fresh install help with stability?

either the system spontaneously turns off (then reboots) or
the system freezes for a second or two, then turns off and reboots (e.g., happens during BF2 videogame, where screen freezes, sound effects distort/freeze)
more frequently BF2 simply crashes in the middle of the game (i.e., spontaneously turns off), but without otherwise affecting the rest of the computer (applications). I think this issue may be more BF2-related as it has happened (but very rarely) prior to the 8400, but it now has happened about half a dozen times in the last month or so.

I let memtest run a couple of hours, as did Prime95. No trouble.

Heat doesn't seem to be a problem. (One core is about 5 degrees warmer than the other. They average around 37 and 42 degrees respectively, but don't go much over low 50s during stress testing). Temps are consistent, via RealTemp, CoreTemp, and SpeedFan, adjusted for conservative TXMaxx of 100 degrees).

Temp1 is pretty high though. I thought that was my Northbridge temp, but don't know for sure. With the e2180 it hung between 34 and 38 degrees, and never much higher. Now it idles around 40, and goes up to the low 50s during stress testing. Temp2 is about the same as before, idling at 32-33 degrees, and not changing regardless of stress testing. See from my stats below, that the system is well cooled.

9 x 423 MHz = 3806 MHz
VID 1.15 (via RealTemp)
VCore at 1.32, 1.31 after VDroop.
VTT at 1.37.
NB at 1.5.
SBI/O 1.7.
VMem 2.2 (2.16 with droop).
2x2 GB RAM, with timings on Auto for now.
Bios v.1.8

I have played with about every parameter I know, except for reinstalling the OS. From a previous experience with a single core FX55 to a dual core X2 4200, the reinstall was the ticket. Could that be the case here, or am I missing somthing else? Let me know if you need more info.

Thanks for your help!
31 answers Last reply
More about instability upgraded 2180 8400 windows reinstall needed
  1. It sounds like you've pretty much gone over your problem with a fine tooth comb... is there a chance a BF2 update/patch might have introduced a problem you weren't previously having? You mentioned a problem going from a single core to a dual core with a previous system, but in this instance you went from a dual to a dual... and both are technically Core 2 Duo processors (don't care what the name is... that's what they are) If I were making that upgrade I'd have expected a drop-in replacement without any sort of software problems popping up... everything should have run faster.

    Just gonna throw this out there... have you verified you are running the most up-to-date BIOS?

    On a semi-related note... do a lot of people still play BF2? I might ask for my copy back from a friend. Sniping was crazy fun on that game... way different than sniping on TF2.
  2. Hey Rodney:

    Good to see you back.

    Reinstalled BF2. Ironic, because it probably took longer than to reinstall XP.

    BIOS is relatively up to date. There is a 1.9, but 1.6 supposedly incorporated Wolfedale and 1.8 is solidly 45nm friendly.

    Yes, lots of people are playing BF2--it is still the most fun for online gaming that I have found. A bunch of us play on a server called "Steller's" because they are all mature, good guys with zero tolerance for cheating. Be sure to say "hi" to scooter_mcgilicuddy if you are on in the evenings, EST. You should also check out the AIX mod, because it has some truly amazing improvements to the game, but I digress...

  3. It sounds like you've covered the bases pretty well. One thing you didn't mention was whether you reset the CMOS before installing the new CPU. If you didn't, I would give that a try.
  4. I may just pick up the deluxe version at Wal-Mart for $19.95. I've only got a laptop, but it's got game (C2D, 4 GB, 8800 GTS) so I know that'd play the hell out of BF2. Ugh... I need to quit spending so much money!!!
  5. No where in your original post do you mention running everything at stock. If you haven't done that, in my opinion you have overlooked the most basic troubleshooting step.
  6. Thanks shortstuff. I had to hit the CMOS a couple times when I was getting the 8400 to boot at all, so I guess that is covered.

    BF2 is technically a 3 year old game, so as long as your laptop has a discrete VGA, you should be fine. Yes--be sure to get the Deluxe edition, because most of the servers use the expansion packs. But rest assured that your $19.95 will provide more than its share of happiness!
  7. Hi Techgeek:

    Yes, everything ran at stock for a couple days. I think it was just as unstable, so that is kind of how I justified starting to tweak all the settings. Before I knew it, I was at 3.8.
  8. For that FSB the cpuv looks a tad low. I have my E7200 at 400mhz and i was having spontaneous crashes at 1.31v. I raised it to 1.33v and the crashes stopped.

    Being you are running at 423Mhz, you might the to run a little higher cpuv. I think 1.34v is safe.
  9. Thanks 50. That is the kind of commentary I was hoping for.

    If it goes up to 1.33, or even 1.34, you think that is still OK? I don't fear OCing generally, but I do fear bumping up against thermal limits, or voltage limits. The consensus I hear is keep below 1.35. Do you agree?
  10. husky mctarflash said:
    Thanks shortstuff. I had to hit the CMOS a couple times when I was getting the 8400 to boot at all, so I guess that is covered.

    Was this meant as a reply to me? I guess I should have been more clear, did you back everything back to stock and retest?
  11. Sorry caught your reply after posting. Well if it's unstable at stock, and it only appears in BF2, seems pretty weird. I don't play that particular game, so I don't really have anymore to add.
  12. Hi Techgeek:

    BF2 is the most common scenario for crashes, but then again it is the common high-stress activity for my machine. It did crash the other day during routine use (Microsoft Office, with some Firefox thrown in).

    Come on, techgeek, one cannot claim to be "born to frag" without playing Bf2!!! ;)
  13. Up that vcore.
  14. Your vtt looks a little high for a 45nm cpu. On mine I run 1.1v vtt as high as 475 fsb (Haven't tried any higher since it will take 1.4+ vcore and I don't care to run that high) with 1.3v on NB 1.5v on SB but only have 2gig of ram so you may need more NB voltage. I don't know if lowering the vtt will help your problem but it will help the cpu live longer at least.

    Are you running your mem 1:1? if not it may be your problem, maybe a little more voltage or loser timings on your mem may be what you need.
  15. I agree - looks like it's time to start raising the vcore until you get stable. You might not have got the best overclocking chip. You mentioned you ran prime95 and memtest for "a couple of hours". I would run each for 8-12 hours. I've had overclocks I thought were stable because they passed prime95 for about 4 hours, but then fail at around 8 hours. It looks like you've still got a little room on your temps. I wouldn't go over 1.35v either though. That does seem to be the max recommended for the 45nm chips.
  16. Thanks roadrunner, ancient1.

    I bumped the vcore to what I understand is the max for a wolfdale: 1.35. Admittedly, I am a bit scared by that. I ran Prime95, and paid attention to voltages. Once the stress testing kicked in, the voltage drooped from 1.35 theoretical (was 1.34 actual), down to 1.31. So 1.31 v at 425 mhz FSB--thats pretty low, isn't it?

    Mem is at 1:1, and on auto. Auto gives it like 5-6-6-20 or something like that.

    I was trying different parameters with VTT to check on stability, but it had typically been on auto. I will move it back to auto. I will double check NB and SB...

    Maybe it is just that VDroop. 1.31 does seem low for >3.8 GHz from what I have read.

    I will likely reinstall XP, but from what you all think, that isn't likely the issue is it?

  17. To me I think that 1.31 is plenty high for just 3.8. From what I have seen most will hit aprox 4.0 with that kind of vcore, it is usually above 4.0 most need more voltage and it seems that from that point it takes a lot more voltage to get small increases.

    Mine will run 3.87 with 1.23 (with all energy saving and vdroop enabled) which I am running now 24/7

    But I can run 4.05 with 1.29 4.1 with 1.31 4.15 with 1.35 and 4.2 needs 1.38 but I have to eliminate the idle/load power savings to get 24/7 stable with those.

    Here are a couple shots on how I run 24/7 with the energy saving features on.



    One other thing I just thought of. A while back on a different system I had a problem similar in that I would get 1~2sec freezes and very random crashes. It turned out to be a software problem with 1 or 2 programs I loaded at boot-up (it was either Asus probe or DVD Anywhere or a combination of the 2) and once I stopped them from loading my instability went away.
  18. husky mctarflash said:
    Hi Techgeek:

    Yes, everything ran at stock for a couple days. I think it was just as unstable, so that is kind of how I justified starting to tweak all the settings. Before I knew it, I was at 3.8.

    If it's unstable at STOCK speeds, stop right there - don't tweak the voltage cuz it should work perfectly on the default settings. Remove the overclock and start your troubleshooting there - having an overclock can introduce other problems (i.e. you may have only one problem but OCing causes 2 more on top of tackling one at a time is the best practice).

    Try running 3dmark06 and see if it crashes at stock speeds. You didn't mention what kind of video card you have. Make sure all your fans (including the fan on your video card) is CLEAN - if not get a can of compressed air and clean them out. 3dmark06 should stress your system a LOT more than if it doesn't crash on 3dmark and it does on BF2 there's something weird going on there that a format/reinstall may fix.

    Try running the latest BIOS.

  19. Wow--ancient1--you are getting some great OCs with low voltages. It is good to know that it is possible to get those kind of OCs with such voltages. Maybe I have a bad chip though--it may be cosmic payback for getting such an easy OC on my 2180. ;)

    I tend to think I may have some software issues as suspected by unatommer and ancient1. I want to say that I only had a single crash via Prime95, and that was very early in my expiramentation. Every other crash has come during usage that arguably was not nearly as "stressful" as Prime95/memtest.

    I will first try a reinstall of Windows, and if that doesn't work, a fresh install....
  20. Before you reinstall windows I would go into msconfig and disable all programs you don't need and see if one of those are causing your problems.

    That's a lot easier than a reinstall and doing all the updates again.
  21. Thanks again Ancient1.

    I reinstalled Windows. What a PitA. Since my "disc" version is SP1, it won't let me directly reinstall it, so I had to peel back the versions first. (dare I say I am tempted to upgrade to Vista!) I don't necessarily notice a difference in operation. But BF2 did crash again last night.

    I ran Prime95 overnight for ~8 hrs., with no issues.

    I went into MSconfig and disabled non-essential programs, but since I tend to pay attention to that, there were only just a couple more that I could have disabled.

    It sounds like it is more of a conflict/software/driver issue that just happens to be manifesting itself now.

    I am going to snag a new HD in the next couple of weeks, and I will do a complete fresh install (and this time, create a separate partition for OS so I can reinstall with minimal pain).

    In the scope of things, 3.825 isn't too bad of an OC. The vdroop seems to be influencing things--again my voltage sat at 1.31 when running Prime95, even though it was set to 1.35 in BIOS. But then again, vdroop really shouldn't be an issue here: I am a bit disappointed that it takes so much voltage in light of others being able to hit that OC or better with low 1.20s volts.

    I've generally been pretty happy with my MSI Mobo, but it really isn't an "enthusiast" board, and it simply may not be the most efficient/effective OC mobo. (damn, I hate to want to upgrade my mobo with nehalem around the corner!) Maybe the e8500 would have been a better choice, since I would not have needed as high of an FSB due to the higher multiplier. Oh well.

    Thanks again for all of your help!
  22. Thats a bummer that it still crashed. Reinstalling with SP1 really does suck...perhaps try making a slipstreamed disc for next time?
    Either that or make an image of your OS drive after you get XP installed with only your drivers and all patches before you install anything else.

    Best of luck!
  23. If you run Prime95 8hrs+ it is highly unlikely it is a hardware issue. Look in to driver/software/device conflicts.

    Side note: Slipstreaming would help a lot. Here is an easier way to do a full un-attended slipstreaming/install:
    I use that for XP/2003.
  24. Good to see you again Shadow. Thanks guys for the advice on slipstreaming. I have heard about it, but never bothered to investigate.

    So... Final progress report: tried 4.0 GHz, but it failed to fully post. Tried 3.95, and it failed a minute into Prime95. Backed down to 3.925, and it made it through Prime for >3 hours. Stayed at about 52 degrees, up to 56 on a few spikes--I saw no meaningful increase in temps from about 3.6 GHz onward. Good enough for me, for now.

    When I get a new HD, I will reinstall everything (incorporate slipstreaming), and see if there is any OC headroom left. Clearly I am at the point of diminishing returns for OC vs. real world performance (perhaps WAYY past it in fact), so any future attempt to break 4.0 GHz is out of love of the game.

    Thanks again guys!
  25. Thanks for the help everyone. I thought I (by "I" I mean "you") had it solved, as per my last post. But the system crashed the second I entered a 3D game.

    I started a new post on it, because it is kind of off topic now. If you all have any ideas, I would love to hear them!
  26. 2 things.

    Put vcore @ 1.4 and dont ask anymore questions until you try it. If your scared your done. Be happy.

    You have a MSI P35 "not the best overclocker around"

    A top quality p35 tops out at around 450 if your lucky, yours might not like 400. Until you try 1.4 vcore, you will never know.
  27. Hey Roadrunner:

    Thanks for the reply.

    Nope. Gonna settle at 1.35, which is supposedly the max voltage.

    I think your comment on the MSI is at the root of it all: not every board will OC as far as you want it to. I guess I should be happy it does over 420.

    May have to just get an e8500, with its higher multiplier--cheaper than investing in a dead end MoBo (in light of Nehalem). Then again, I can leave it on 3.8 GHz and forget about it.
  28. You can run 1.4 for sure. I do it all the time, as well as many others.

    If you want to OC more get a P5Q mobo. They overclock like crazy. 500 is easy to hit on any P5Q. You need 1.4 on your current board to even think about it going higher. And you are pushing the wall of that board for sure.

    A new P5Q is cheaper then a e8500, and either way you go your screwed for you Nehalem upgrade.
  29. I just sold a customer a budget build with a e1200 1.6 oc to 2.4, the dam board I used had no voltage control and the chip I got had a high VID AKA all she would do was 2.0. I ended up replacing the e1200 with a e2180 and it just made it to 2.4 lol.

    Most e1200s will oc to 2.4 on the boards i use, but not all parts are the same. You win some you loose some.
  30. Hey Roadrunner:

    Thanks again.

    My southbridge voltages are maxxed. Any advice for what Northbridge should be? What is the limit for NB?

  31. There seems to be some misconceptions about CPU voltages.

    The ABSOLUTE MAX for 45nm is : 1.45v

    The ABSOLUTE MAX for 65nm i: 1.5v

    "ABSOLUTE MAX" is the point where physical damage to the CPU can occur very quickly in some cases in a few hours.

    +1 for roadrunner for getting a better board. Look in to his board recomendation or a P45-DS3L or a ASUS
Ask a new question

Read More