Jump to content

Death909

Members
  • Content count

    77
  • Joined

  • Last visited

About Death909

  • Rank
    Member
  1. Death909

    VX4000 overclocking question

    I had more or less the same problem, good cooling (two 120mm fans, one blowing and one sucking air across the VX, and a 120mm CPU fan hanging over the VX) and yet my VX sucks.. Terribly. I could get the VX running at 244-248 IF I run it with only one stick. With two and a 160:200 divider, I'm lucky not to get BSODs with the ram above 230mhz. The DFI NForce 4 motherboards have weak Vmem voltage controllers, they tend to get weaker.. I used to be able to get two VX sticks to run at 250mhz (VX 3200) with the memory jumper thing at 5v. Currently, the difference between 3.2v without the 5v line thing, and 3.5v with the 5v thing is negligable. There was some guy on xtremesystems who had built some sort of 'booster' he designed to run along side the mobo voltage regulator. The regulator would clamp the Vmem, so if the voltage controller didn't put out what it was meant to, the regulator would bring it back to normal. I haven't checked up on his progress recently though.
  2. The CPU temp is accurate to within 3 degrees (up or down, so 45c on software could be 42c to 48c in reality) ... This is accurate enough for most purposes, but not accurate enough for CPU to CPU comparisons (checking improvements from better cooling will be accurate to a fraction to a degree even if Since the CPU temp probe is inside the CPU, testing in different motherboards is quite pointless.
  3. Death909

    How much is too much

    If you have to jump from 1.4v to 1.65v for just an extra 100mhz, I wouldn't consider it worth it. 2.4ghtz is fine, you won't see a huge difference between that and 2.5ghtz. I can barely tell the difference between 2.45ghtz and 2.85ghtz on my 146 while playing any games. For me, it takes 1.45v (the minimum I use) for 2.8ghtz, but for an extra .075v I can get 2.9ghtz. My RAM happens to be unstable around this point, but after 2.9ghtz it would take even more voltage just to hit 3ghtz. Same with my 3200+, 1.45v was fine for 2.6ghtz, 1.525v was fine for 2.7ghtz, but even with 1.7v, 2.78ghtz was not stable (temps were 40c to 43c at load). There was an excellent guide somewhere on over-volting. One of the things it suggested was that you write down how far you go with 1.4v, then boost it to 1.425v, and so forth till you've gone as far as you feel safe. If you make a graph out of that, you will see that extra voltage gives good gains till a point, and at that point diminishing returns kick in.
  4. CPU spectrum spread, atleast on older Intel boards, would actively alter the CPU frequency to spread the EMI generated over a wider range of frequencies (so that there was less continuous EMI at specific frequencies, which would lower the interferance on devices that are only sensetive to certain frequencies). Its definitely something that should be disabled, an aluminum case and an aluminum heatsink (that is grounded to the case) will stop just about all EMI emanating from the CPU (protecting other PC components from EMI) and the case (less EMI for external devices).
  5. Death909

    Unofficial Opteron 939 Thread

    I have the exact same problem. My 146 is stable at 2.9ghtz, but using the 160/200 divider results in my VX being unstable at or above 240mhz (it can do 245mhz in memtest). I haven't been able to test this myself, but I think that by using a memory capable of higher HTT (TCCD) and dividers closer to 200/200 you could get closer to 3ghtz. My VX is quirky (not in a good way, one stick doesn't get any errors at 246mhz but it won't boot into memtest at 247mhz) and the Vdimm controller is screwed, so I have to run 3.5v though my VX just to get it stable at 245mhz. Using the 160/200 divider, it is stable at 238mhz, but anything above that gives random BSODS.
  6. Death909

    PCI eXpress Frequency

    Increasing the PCI-E frequency increases the bandwidth of the PCI-E lanes. If your card only has 1x or 2x allocated to it (certain SLI configs) then it might help. It also might help if you have top of the line video cards in 8x/8x SLI config. Other than that, it shouldn't do much at all. (Milkshake, I expect you might have changed other settings and OC'ed your system at the same time as raising PCI-E frequency. I know when I get into the BIOS that its very tempting to change numerous variables and you don't always remember what you changed afterwards).
  7. Death909

    Unofficial Opteron 939 Thread

    Next time I'm buying I'm going to ask for a specific week aswell. Any particular week I should look for? 0540 maybe?
  8. Death909

    Unofficial Opteron 939 Thread

    Well, its better than my 146 then. I've just had to boost the voltage to 1.475v to get stability at 2.86ghtz. Its better than my original estimate though, since it ran 2.7ghtz with 1.45v I was expecting a 6.66% better OC than my 3200+ (since it needed 1.55v for stability at 2.7ghtz). Once more I get a decent CPU, once more I miss out on one of those miracle ones that can OC like mine at stock voltage. I'm seriously considering buying another just to try and get 3ghtz at 1.4v. Guess I should sell the 3200+ first.. Edit: I was thinking, if dividers such as 140/200 make the CPU unstable, wouldn't higher dividers like 160/200 keep the CPU stable until higher clockspeeds are reached? As in going from 160/200 to 180/200 might get stability at higher clockspeeds? This is a wild assumption, but if anyone has TCCD and wants to test this...
  9. Death909

    Chipset Temp

    I do not know how the heatsink looks or works on your video card, so it may or may not be responsible for your high chipset temps, but you can move your videocard to the second PCI-E slot, it will work and the difference in performance is incredibly minor, something like 200-400 03marks or a fraction of a frame in games. Probably the best thing you can do to lower chipset temps is to replace the thermal pad with AS5 or Ceramique. My chipset used to get 45c-48c under load, but after cleaning the heatsink / chipset with Articlean and adding some Ceramique, it now idles at 30c (or 3c below my CPU, instead of 5c-8c above it during load and 10c-12c at idle). Running prime results in just 1c extra on the chipset (unlike 12c extra on the CPU).
  10. Death909

    Unofficial Opteron 939 Thread

    Roger_D25, I hope that thats a mis-type, 1.932 is a hell of a lot, and 1.9v has been known to kill AMD64s instantly (bug with the 7/02 BIOS ran 1.9v on some settings). bucdenny 63c is a bit much, temps like that would probably be close enough to trip the thermal protection and turn off the CPU. Better temps would definitely help, my 146 was not stable at 2.7ghtz with just 52c (left the fan on my Big Typhoon unplugged..) With 44c, and the fan turned on, its stable at 2.8ghtz (forcing my VX to play nice, 2.8ghtz seems stable).
  11. Death909

    Unofficial Opteron 939 Thread

    The 150, if it is a good stepping, will probably do much better than 144s and 146s when used with low speed, tight timing ram. Opterons seem to have alot of trouble with low dividers (like 140/200 or less). This doesn't seem to be a problem exclusive to Opterons though, its just that the most popular ones have 9x or 10x max multipliers and most if not all can hit somewhere near 3ghtz. Currently, I wish I had paid a little extra to get a 148.
  12. Death909

    Evga 6800GT Temps

    TIM = Thermal Interface Material. Its a broader way of saying thermal paste (as it encompasses any material rather than just pastes).
  13. Death909

    cpu mult isnt working right

    Half-multipliers are screwed. Its something about AMDs implementation of them (something about half-multies using dividers). Good job on finding this out for yourself, I have seen numerous others get 270FSB at 2-2-2 with VX before someone kindly pointed out that their RAM was running at 245mhz or so.
  14. Death909

    Unofficial Opteron 939 Thread

    Ok, now I'm really, really dissapointed in my VX. Two sticks, back from RMA, and both of them cannot handle voltage. Neither will do a full pass of memtest with any more than 3.4v. My 146 runs 100% stable at 2.7ghtz, but even just moving up to 2.8ghtz results in the VX running too fast, since it can't handle anymore voltage I have to change the ratio so that the VX runs slower, which is a huge problem because my 146 seems to have some problems with anything below the 166mhz/fsb ratio. Unfortuneatly, my 'good' pair of VX cannot do 250mhz, let alone agree with each other (one requires 3.4v for stability, the other wants 3.5v, and both are unstable if given the other voltage). To be honest, I certainly wasn't expecting so many problems with OCing a 146. The ability to get high clockspeeds with relative ease is no good if dividers and memory speed prevent you from going above 2.7ghtz. (Plus, I seem to have currupted a large amount of drivers trying to get stability at 2.9ghtz). Edit: Audio and video plays WAY too fast, with audio quality being terrible. Edit2: Oddly, after being unable to fix the audio distortion (despite trying just about everything), it seems to have fixed itself.
  15. Death909

    Unofficial Opteron 939 Thread

    Telling someone to go buy a dell because they are asking questions is being a bit harsh, even if they cannot be bothered to look/search through this thread (even I gave up on reading all of it after the 26th page). I can understand the reactions against sumang, WAY too many people are asking questions that cannot be answered 100%, but its still not very nice to be so harsh on him/her. I guess the point that should be made at the very begining of every OCing guide or tutorial should be that every single peice of hardware is different, and OCing is about experimentation rather than answers. (BTW, what is 'english pleb' atitude? pleb is an abreviation for?)
×