Jump to content


  • Content Count

  • Joined

  • Last visited

Posts posted by Clay

  1. @Will you mean you've never corupted a bios to the point where the mechine still booted. But basically needs a reflash to run right. I did worse that not long ago. I had reflashed and was trying to clear the cmos in the dark in a cramped space.Couldn't see and could barely get my hand in there. I managed to turn the bios speaker off and after 15mins I turned it back on. Then I loaded optimised saved restarted and what a mess. I pretty much knew what I'd done. As soon as I saw the first bsod.


    Then the otherday I lost the keyboard. Then I started getting all kinds of other weirdness. IRQ_LESS THAN_EQUIL It didn't take me too long to get the picture in that ordeal. Reflashed its okay now.The oddest part about this ievent s it was running really really nice. No doubt better than ever 307x10 passed everything upto and including S&M. No warnings or anything. I might have a weak cmos battery. Probably a real good idea for me to replace it.

  2. I goofed onna' coupla' timings Missed em'.. Trying again 310X10 but about 2 1/2 hrs into prime I realized I'd missed a coupla' other timings. It might skate by though. It's givings me something to do. And when it fails it kinda' gives me a better idea of what doesn't werk..This time I forgot to add some extra vdimm. But its still running. And I'm running kinda' diffrent settings now. It might not work but then?

  3. I'm sorry Hotshot I'm off topic again.


    Hehe.. Thats what my 148 was using for my superpi entries. Cpuz sez 1.662 but by the settings it was 1.695. I could get @3200 32m stable and even a 3dm01 outta' it. but couldn't even get close to a prime or even a quick occt. (I think it wanted even more vid and certainly some smoother timings.) Cold won't help it either. (Cold bug) Though this processor has passed 32m's @3267 (1.575vid) I can't even get close to my times that processor threw up. Oh well its a good spare just in case. Plus I have the other spare. LCB9E-0643 can't remember the last 4 letters. I found that stepping @XS same steooing-date but diffrent last letters. And it maybe better than this 1. I really won't know til' I test it. I'm having too much fun with this 1 right now. Right now it's priming away @310X10 [email protected] Using Kitfits timings for 310 its been going for 6hrs. I'm really pleased... I might have gave it more vid than it needed. Failed after 2hrs before this run. I had a timing out and thought it was vid.(Common mistake for me anyway) Forgot to put it back when I realized. Maybe we should move this conversation over to the 3ghz thread.


    Edit;Dang it.. 6hrs33mins failed prime. Thought I had it that time too. Oh well back to the drawing board..... I know what I did wrong.

  4. Well they are usually obsolete before they die anyway. LOL..... Does that mean we should go outta' our way to kill em'? If so I've been playin' this all wrong. Oh... wait maybe not. Eehh er' umm.. Humm... Can't decide Oh NO! I guess I shoulda' jus' kept my big fat trap shut..LOL


    On seconed thought Hotshot I'm sorry forget everything I said. Next time ya' see those processors on sale. Grab ya' a hand full!! An full speed ahead! That's how its done! Right? "Will"? :nod:


    Seriously though having spares keeps the stress down. Hey this opteron looks like might scale to 1.7vid with this cooling.. Now wait we'd better think 1.7 might be dangerious. Ok give it 1.695 that oughta' hold er' down... (Thats still a good processor.LOL..Seems to like lotsa' voltage."Not recomended for home use")

  5. @syngensmyth: He's probably still dailin' it in. Trimming the voltages and stuff.

    I hope he's okay..... He hasn't posted inna' while either.


    Once again priming @310x10 1.375*113 Kit's timings exactly.(I think) Looks good except I might have been abit heavy on the vid. 4 1/2 hrs to go. looks good. Not chuggin' now it's sorta' rollin' along. lol...

  6. Its a known issue. When powering mems off the 5v rail people have had trouble. It a good idea in this case to run some extra voltage to the cpu tp protect the memory controller from damage from the extra voltage. Here's a short post from someone whos highly respected here and XS. Advice from when I was new to nf4 overclocking I was running OCZ PC 3500 with winbond UTT BH5 ic's .



    vdimm needed depends on the modules, cpu mem controller as well as the motherboard


    for DFI NF4 Ultra-D/SLI-D/SLI-DR you have to be aware of certain issues

    1. to get >3.2v vdimm you need to set the 5v rail jumper to get 3.3-4v vdimm

    2. using 5v rail jumper the mosfet heatsink near the jumper can get very hot >50-60C if no active cooling, with a fan over it probably around 35-50C under memtest86+ load or 30-40C under windows memory intensive load

    3. with 5v rail jumper, the more vdimm you set closer to 4v vdimm, the cooler that mosfet gets, so running 3.5v vdimm would be cooler for mosfet on board than 2.8v while on 5v rail jumper. So if you need to run <3.2v use default 3.3v rail jumper, if you need >3.2v vdimm use 5v rail jumper - ideally i set my vdimm at 3.4v as it's cooler than 3.2v even if my ram only needs 3.2v --- this is what i set my BH-5/UTT BH-5 to 3.4v even if it doesn't need it to keep mosfet cooler :)

    4. low vcore + high vdimm can apparently kill the memory controller on cpu so if you cpu can do 3000mhz at 1.4v but you have vdimm at 3.4v you could be at risk, so what i do is set vcore at 1.45v -1.48v even if cpu can run lower vcore


    for OCZ PC3500 Gold UTT BH-5 read this thread for ideas http://www.xtremesystems.org/forums/showthread.php?t=90899


    Note: Issue 4#. At 3.3 you could be somewhat at risk also. (But not nearly like 3.6-7!..LOL) I would do what eva2000 sez here on this even at 3,3. I've thought about this for along time. And I think what this about is the (For lack of a better term) "Signal" or data from the dimms is comming in to the processors mem-controller at voltage somewhat greater than it was actually designed for. Most of these processors never see more than stock voltages or more than 2.6 from the dimms. So we have stock volts (1.4) cheap dimms 2.5-2.6vdimm okay thats 1.1-1.2 volts diff. Looks like a mem controller absorbs that well enough. Then we have the winbonds wanting more and more. Processor @1.4 winbonds at 3.3 thats 1.9volts diffrence. or 3.4 theres 2.0 volts. Adding abit of extra voltage to the processor seems to (for lack of a better term)"cushion" the mem-controller. Though how .05-08 volts of vid helps that much. I dunno but apparently it does. If "Neo" was still here he could explain it and no doubt would. Next time I get a chance I think I'll pop over to DFI-Club and ask him.


    Yeah OCZ RMA's are good. When I need to do that I usually talk to Sean. And my concerns are well taken care of.(Good people!) But RMA's are a pain even with the best support.


    Mem fans yeah. I killed a stick of utt-ch5(OCZ pc4000vx at a friends when I forgot to bring the fan. Hot day nf2 235x11 3.1vdimm it went pop. But the errors are the most embarassing when people are looking. ( The TCCD in the lp-b error @2.8 without fans.) for winbonds OCZ recomends active cooling @2.9+.


    HD movies humm...Any idea of what kinda' bit-rates those are throwing? Do you have a seconed pagefile system on one of your other drives? If not and want to try this.You'll need to repartition a drive. Keep the windows managed pagefile on your os drive as is. (If Raid-0 is the os drive all the better)

    On 1 of the other drives creat a 3.5-4gb primary dos partition. Right at thwe very start of the drive. I use a dos start-up disk for doing this. and FAT32 format.(drives seem to access and page abit faster) NTFS is okay and this will still help but not quite as much.Then partition the rest of the drive as you see fit. On this first partition I set the page file. I use the same sizes windows uses.For 1gb Min 1536 max 3072. I don't how much that'll help your video lag but it seems to help everything else. Boot times superpis ect. (Actully that's from "tweakhound") I've tried afew diffrent page-file configs and this works the best.

  7. For potential sub-ambient cooling I think I would insulate against condensation and go all the way. Weither that's a massive evaporative bong. tec cooling of any type. Or of course phase....


    Though honestly I can't complain about the PA120.3 8c delta between load and ambient. Not too bad I think.


    Room ambient is 29c voltage is 1.225*126% overclock 310X10 1-1 using Kitfits timings for TCCD @310. My deltas were lower before I added the maze4. Preasure drop raised my cpu temps by nearly 2c!. Oh well... Heck it seems to work ok. LOL




    Edit: As the room ambients drop so does the load temps. 28c 36c-average load. I think the lower the room temps the narrower the delta also. Or rather the higher the room temps the wider the delta. Ooops it failed prime. Back to the drawing board! LOL

  8. Yulp... Just one thing atta' time. Whatever makes the best sense. I think I'd back off the vdimm maybe just a little bit.(Unless the sticks have winbond IC's on them.) I'd defo get a fan on them.(I'm a big believer in fans on the memory sticks) Which bios did ya' flash to? Tony's 704 and 604 both default to command per' clock disabled?.


    So as best as you can tell it's a good flash? Bios flashing is allways nerve wracking for me. Not on my computers though. It's the worst when it someone elses computer and I posted up the instructions.

    Bad flash can be very sneaky. And bios corruption can be sneaky too.


    @Will I fixed my Ultra-D. (Flashed) Seem ok now. I hope your 3700 will be as easy to fix. My living room computer just died.(I hope I can fix it.) It acts 4leds of dead except that board doesn't have the leds. Fans spin up no processor heat or nuthin'. I'll probably try a hotflash off the lp-b. But I have some other things I'd probably best test first. (Don't really have time for it now)

  9. During your long cmos clear. Did you unplug the psu and drain the psu-board caps by hitting the power button? (Might help but it's a long shot)


    Unknown is right about lettin' stuff dry out if you are using distilled water only there is some more hope.


    You didn't notice any leaks or moisture anywhere did ya'? Did the water levels drop that you could tell? Do you have a DMM to test the psu with? Or a diffrent psu to try?


    I hate leaks! Thats why I allways preasure test my loops by running the pump for 24hrs (diffrent psu)before putting any power to the board or firing it up.


    I have a living room computer that's doing the same thing. Except it's not water-cooled and doesn't have 4 leds to light up. LOL. Long cmos clear for now.

    Til' I can get around to testing the psu. (I'm hoping that fixes it.)

  10. You do need a fan. For winbonds. Otherwise at best you'll get errors at worst you'll kill sticks. That might cure your video lag also. (But that could be a slow page file or a bunch of other things.(Timings-voltages.bad movie data.))


    Are you running the mems off the 5v rail? All the bios I've tried max out @3.2v on the 3.3v rail. So I figure so. Did you know that giving your mems extra voltage,can kill your processor? Oh well I mentioned it before. So that's about it,outta' me concerning that. I'll let someone else describe the gory details.

  11. Yep.. It could be mechine defects inside the IHS. or bad paste inside on 1 core. But it could the diodes reading wrong too.


    If you do pop the top.(Easy does it. Careful of the pins) Be sure beforehand you have good adjustable mounting system lined up. So you can slowly-evenly tighten your sink onto your cores.



    As far as the little chip that gets really hot And reads as motherboard in everest. Put a big fan on your mems that'll cool it off too. If you can fit it around your sink. (Plus your mems will love you for it too.) If the fan won't fit,try a small heatsink. TCCD gets really hot even @2.7 on the lp-b it runs okay though. @2.8 it gives errors without a fan. So I allways put fans on the mems no matter the voltage.. (Any computer I own.) The nf4 has never run without a fan on the mems so I couldn't really tell ya's the deal with it.

  12. I'm sorry I've been gone. Tries to Boot from cd? Does it eventually say insert OS disk or operating system not found or something simular? Yep's the guys are right check your boot orders first. Be sure all of that is good to go.


    Also we can't overlook the possibilty of a bad flash. (I don't think so..But maybe) Even done right under the best of conditions. These things can go horribly wrong. If fact when overclocking they can go wrong by themselves no bios flash is nessessary. Once you are completely sure you have the boot order ironed out. (And the hard-drives are plugged in) We might have to consider the possibilty. I lost the keyboard the other day.Semi corrupted bios.(Bad overclocking on my part.Or rather tryin' to get it to do things it really doesn't want too. Weak cmos battery might have had something to do with it too.)

  13. The reason most of don't recomending bios flashing is because sometimes even done right things go wrong. (The stock bios in my lp-b lasted 9mins)


    I flashed to tony's 6/04 (Last bios on page 2 of nf4 ut ultra bios page. On tmods cd.)the other day. and it also defaults to 2t timings.. (Ooops somebody allready mentioned it)I went back to the 6-23-3 bios shortly after though. I've allways wondered why my ultra-d shipped with a plcc puller.(Does someone there know me?LOL)


    Edit: If you need proper bios flash,post-flash instructions I'll post them up for ya'.


    Okay here ya' go. Once again I would ask that if anyone sees anything wrong or if I've missed something or forgotton something to please post up.


    1st. Don't flash overclocked.

    2nd."Important" Load optimised defaults.

    3rd. Boot to cd.

    4th. select lan-party bios.

    5th select nf4 lan-party bios.as the Ultra-d and the sli-d and the sli-dr bios are the same.Careful I've nearly flashed to an nf3 ultra-d bios before.

    (Avoid the expert and venus bios)

    6#. Select bios and flash. (Heres where we all pray the power stays on.)

    7#. Important: After pressing f1 to reset. Don't let the computer reboot. I allway try to stop it at or just before the system repost. Power switch on the back of the power-supply works good.

    8#. Cmos clear. Impotant A#.Unplug the psu and drain the psu and board caps By hitting the power button. B#. Remove the cmos battery. C#. Set jumper to the clear position. D#. "Wait" Most peeps only give this about an mim or so. Though I like to let it sit for awhile longer though. (Usually 15mins or so.)

    9#. Important A# plug it back in. B#. Put the Battery back in. C#. Set the jumper to the run position.

    D#."Important part" Don't let the the computer reboot. Enter cmos and load "Optimised" again. "Save and exit".

    10#.Re-enter cmos and start overclocking. I try to do this changing only 1-2 settings at a time as I believe this helps prevent bios corruption.


    I'll post this in your thread too. Just to be sure I got everything right.(Put it up for everyone to scrutinize)

    Edit: In step 8#. It might be better to pull the cmos battery before draining the board-psu. But probably not a real big deal.

  14. About the only contructive critism I could possibly offer. Is maybe you might opt for a bigger rad. MCR320's are going cheap and I wish I'd ordered 1 instead of the 220.(On the lp-B) Still the 220's are quite small and they present fewer troubles when mounting them internally.


    Edit: Fans The superquite yates loones. 47cfm (28dba) Didn't seem to quite cut the mustard on my 220. At least as far as I'm concerned. 2#70cfm (33dba) would have been about the all around best.(I can't hear them.I was out of those though) But the 2#88cfm's(40dba) seem work really nice with a only small amount of noise. I can barely hear them. (Beats the heck outta' two 92mm Vantec Tornados running 24/7 fer' sure! At least in the noise department)


    Edit2: 15' of tygon? It's allways nice to have some extra. And really you can't go wrong. I try to keeps my loops fairly short. 4 1/2'-5' or less. If possible. Less is better as long as there are no sharp bends or any crimping..


    It all really looks good to me,from here. Best of luck to you on that. Watch out for leaks!...

  15. @Unknown: Watercooling Yay!! The 6002 along with it's cousins (5000's) are vaunted as the very best blocks for tec cooling. If you can get 1 cheap go ahead.(It may prove useful somewhere) Apogees? The Gt I have on the LP-B seems to work ok. It's a fairly nice "low preasure drop" block. Proforms about as well as anything else.(+- a c or 2.) I would have prefered a storm on this cpu.(Single core no IHS) But whatever. It seems to work pretty good. Modern mid-range to high end water blocks. You won't see big diff. (If I hit the lotto I'm gettin' a G-7! LOL)


    Edit: Have ya got er' up and cooling? Hows it doing?

  16. Ok Better cooling should make a diff...... If you haven't allready. Be sure you have a fan on those mem sticks. (Anything over 2.8-9(2.7for tccd))

    Just curious What is your vid set at now? Total mhz? Did you volt up abit to help protect the processors mem controller?


    Movie on your hard drive? Lags? Humm... that could be alot of things.

  17. Welcome....

    LTD 2x? Looks way way too low. I get 300+ @3X. For starts I think I'd try the 3x-4x.


    For me 4x werks upto around 280. Still everyone computer is diffrent. 3x to about 307. Past that I have yet to test for stability. But reports are 2.5 is needed around 310+.


    The idea is to keep that buss operating in a certain range. Too much=Bad. Not enough=Bad.


    PS When getting help I think you need to make a hardware sig.(Rules.. I hate to point that out though)

  18. Nice werk!! 3016 Yay...For the ocdb. I'm not really too sure. But I think you still have to enter your timings in the template. I kinda' think the bios screenshots are considered extras anyway. (You don't have to do them)


    Temps: I've really can't say I've seen any real problems @55c or less. But every processor is diffrent. And everyone has diffrent ideas. It seems there is a thresh-hold Between heat and mhz's where more volts won't help. Untill you can get it cooler. Then the voltage seems to go just abit further. And then even more voltage, is then usable.(More mhz's)

  19. The stepping and date are on the lid of the processor. The bios date and version is something you shold be able to find when the mechine posts. But it might be easier if you don't know where to look. To download everest home edition. Motherboard-bios tabs reads it right up. The actual stepping of the processor is something you'll have to read before installing the processor.


    1.5v shouldn't hurt it. I try to keep my voltages trimmed to a min.(Reduces potential EM)Better cooling also helps it to scale further . Plus they don't seem to need as much voltage. I've noticed in the past that she'll pass prime or sp with a given voltage but sometimes it seems to need just a little tiny bit more for some games. I hate annoying app errors.(Especially when people are looking.)


    What kinda' loop do you have?

  20. Ok say 2500(Total) divided by 10(CPU-Multi)=250(FSB) Divided by the divider

    whole (10) = 25 X the divideri fractional(9) = 225 mems (9/10)


    Or 2500 Divided by 10 (CPU-Multi) =250(FSB) Divided by the divider whole(6)= 41.6666666eet...ect.. Times the divider factional(5) =208.33333333....ect.. (Mems.) (5/6)


    Ok I'll go doit another way BRB.


    There ya go. 9/10 (180mhz divider)





    Math sez 225 cpuz sez 227 Heh... Oh well I gotta' go put my settin's back. LOL Somebody might have goofed. Hehee... I noticed that error in the guide myself some time ago. But what do I know? I ain't no expert er' nuthin'.. I jus' keep hittin' em' with bigger hammers... :tooth:

  21. Ok... I'm sorry... I really should find those glasses. :tooth:


    Then yulp vid. You've probably seen this before. But it can help you ease up your volts. Use a +special % gives some better stability. (More consistant cpu voltages)




    [email protected]/10 should be 225. (250dividedby10=25x9=225)

    [email protected]/6 should be 208.




    I know its my cpu because my mem was at 1/2 (edit: my ram does not even make it to 240, its stuck at 235 right now)...


    as for 5/6 vs 9/10, it using the ceiling function apparently, so 9/10 = 1.11 or 11.111(*10), which ends up being 12.... So using that formula 9/10 is the exact same thing as 5/6 (1.2 * 10 = 12), isnt it?



    cpu multi = 10

    bus = 250


    with 5/6 = 208 ram

    with 9/10 = 208 ram

  22. How do you know its the processor that not good to 250? At the current voltage. How do you know its not the mems that won't go past 240 with your current settings. Have you tried a divider yet? I really don't know. I read the thread but I don't see too well. And I may have missed it if you did. (I've killed 5 sticks of winbond UTT-BH5(2) and UTT-CH5(3) And that sounds like them okay.. [email protected] max. Oh yeah you can volt the heck outta' them for abit more but they won't last.Other peeps around here have killed more. But why? When you can put them onna' divider. Another good tip. If you havent allready done so get a fan on those mems for voltage of 2.9 or more. (For tccd its 2.7-2.8) I put fans on mems even at really low voltages. They'll error less and maybe last abit longer.


    Dividers are for tailoring your processor speed To your memorys limitations.

    Say your mems max out @240 at the dimm voltage-settings you like. With a 10 multi thats only 2400mhz. But we use the 9/10 divider we can push the procseeor to 260 or better. (while the mems are still running @nice speeds and timings they like.) With the 5/6 we can go further ect. Try a divider and work up you may need abit more vid in fact I wouild advise it.Especially if you are running those mems off the 5v rail. (To protect your cpu mem controller.)

  • Create New...