Jump to content

naddie

Members
  • Content Count

    174
  • Joined

  • Last visited

Posts posted by naddie


  1. his post above states he used an S3 pci card and i doubt it needed any drivers.

     

    That would be correct, and one of the reasons I chose an older PCI card to start the build up before my PCIe card came.

     

    So far, looped tests of 2001 and 2003 are running great all day. I am doing 2005 right now.

     

    I noticed that I still had http://www.dfi-street.com/forum/editpost.p...ost&p=35089650% CPU Throttling enabled, so I disabled it. I wonder if that is what caused the issue. I also re-downloaded 2005 just to make sure it's not the copy of 2005 itself.

     

    So far, It's running okay... but I haven't done a 2nd pass yet ( I am updating you guys live).

     

    On a side note, at 1024x768, 4x anti-ailising and anistropic filtering produced a score of 6583.


  2. Don't feel insaulted bro - I never would have sorted my problem out without a store full of "spare" parts to troubleshoot with :)

     

    btw . prime on stock voltage for almost 2 hours now yay 4 me :)

     

    good luck m8

     

    I understand. I work as a tech-on-the-go, so I have lots of spare parts. Unfortunately, because the move to PCIe was relatively recent (and I have yet to encounter a mainstream PC in use right now that has PCIe video card), I do not have any PCIe parts at all in stock.

     

    At my full time work place, we have Dells that have PCIe slots, but they are not occupied by any video cards, since they use an integrated solutions from Intel.

     

    I ran 2001, and it works fine - I should have left it looping though, seeing how it probably doesn't really stress the GPU as much as say 2005. I will do that when I get back home from work. And 2003 was being downloaded last I checked, so that will get installed when I get back as well.

     

    Haven't checked, but I wonder if the PCIe slot can turn up the voltage a bit like AGP can. I may try that if that option is available before I excercise the "get another PCIe videocard" method. :)


  3. incorrect - bridging it will not increase bandwidth. You will need 2 NICs that support bonding, and even then they would have to be the same make/model (or at least the same make anyways, depending on how the manufacturer designed their products).

     

    Intel and 3Coms are big on this. And usually they are on the more expensive server-level NICs (costs around 50-100 bucks each).

     

     

    The bridging in XP Home/Pro allows for 2 dissimilar networks to communicate. So that means sharing an internet connection, or bridging two local subnets together so that the two different networks can communicate with each other.


  4. I'll try the 2001/2003 versions first. Then I will find a way to secure another decent PCIe 3D video card (hopefully another GeForce so i don't have to reinstall the drivers).

     

    It would be an insult to take my PC somewhere - cuz I'm a techie myself. :) but as a techie, I know that there are well known issues with ANY hardware and hardware combos, so I thought I might air it out here and see if there are any other suggestions other than I tried.

     

    eVGA has a lifetime warrantee, so I'm not too worried if I have to RMA.


  5. Unfortunately that is the only PCIe card I have, and no one I know has a PCIe card either. When I first built the machine, I had to borrow an old S3 PCI video card (4MB, woot!) just to install everything (I was impatient, lol). Then when the video card came, I installed it. :)

     

    Don't worry about the above story - it's just that - a story to tell. I only used that setup to install Windows to OC and check for stability issues before my PCIe video card arrived. After it arrived, I popped it in and reformated the HD and reinstalled WinXP.

     

    I just finished reading your thread about Prime95, and now I feel I am in your position, but not as dire, since my Prime and Memtests passes with flying colors. The fact that 3DMark05 is crashing makes me worried about playing real games on that machine until it is resolved.

     

    Just in case, I am running memtest right now, and will download and try out 3DMark 2001 and 2003 tomorow.

     

    thanks for the quick reply, btw.


  6. I found out that while I can pass one session of tests in 3DMark05, if I try to start another one, it would crash. This is stock speeds.

     

    I originally found this out while it was OCed to 262 Mhz (2.62GHz), but backed it down to stock (200 Mhz, 2.0GHz) just to ensure that it was not the OC that was doing it.

     

    Memtests passed 9:30 hrs, Prime95 passed 9:30hrs (start it at night, check it in the morning before I go to work). I wonder if it is my video card now...

     

    Anyone having similar issues? My setup is in my sig. My vid card is specifically the eVGA 7800 GT CO (model 517). Factory OC 470 Mhz core, 1100 MHz memory. Besides the factory OC, I did not overclock the video one bit.

     

    I also tried to disable memory interleve, and same results. When OCed and enabled, it would crash on the first run. When OCed and disabled, I can run a test once, and the 2nd time, it would crash.

     

    I have not started playing games on it yet, as my c omputer is still a Work in Progress (as you can tell).


  7. I'd like to know that too - what is the effect of turning on CnQ when you have an OCed CPU?

     

    As for my max CPU clock and min voltage, I am currently at 2620 MHz at 1.50 * 107.3% (around 1.61v). It seems much higher than yours, and I dunno why. Anything lower and it just won't pass.

     

    My temps are nice, though - even at this voltage. I am getting 49*C under load.

     

    BTW, what is that fan/voltage monitoring program on the left? Where can I get it?

     

    ALso, for MBM users, what motherboard profile do you pick? Our specific NF4 boards are not on there!


  8. It's official - my max freq is 275 MHz at 2.71v for my G.Skill HZ sticks! memtested overnight, no issues. That means whatever my CPU can OC to, these sticks should be able to keep up at 1:1 ratio.

     

    BTW, timing is 3-4-4-8, and when I find my final freq I will use with my CPU at 1:1, I will try to reduce timing. Currently, it looks like I will be at 262 Mhz x10 (2.62 GHz CPU)


  9. Yeah i was talking about the led's the display when the cpu is processing and the cd/dvd drive is processing, and reason i posted in this section was because you had the exact same problem as me, except my computer isnt loading due to bad cmos, there is bad hardware somewhere.

     

    Ahh... you mean hard drive LEDs. Well, when I plugged them in just to check, none of them lit up except for power. The CD-ROM LED is actually not indicitive of the mothebroard's condition (unless it keeps searching for the CDROM drive, which may cause the LED to blink or glow on some models).

     

    I hope you get your problem fixed!


  10. Currently running a Manchester core at 2.61GHz stable with 1.65v. I am trying to lower it to 1.6v just to see if that will prime stable. It dropped from 52*C load to 49*C load from just the voltage drop. I hope it is stable, though!

     

    Is 1.6v and 1.65v safe for the manchester core? The temps are pretty balmy, not too hot, but I'm more afraid of frying the chip or other damage.


  11. I had 1.65v for a Manchester duel core, and I got 2.61GHz and 52*C load.

     

    I lowered it to 1.6v to see if I can prime stable. The temps dropped to 49*C load.

     

    I wonder what the Manchester duel core voltage should be max. I dont want to fry this chip. If you guys are playing at 1.5v, I wonder if that is because it is a single core or would it be the same for a duel core setup.

     

    Should I be under 1.6v? Perhaps 1.55v?

     

    BTW, it's a x2 3800+ Rev E4


  12. I've read here by Angry that SuperPi is good to use for the mini-test portion of OCing the CPU. However, not sure if he used an x2 for it, because mine would only load around 50% of the CPU. It looks like it uses about 80% of one core, and 20% of the other.

     

    When I try to run 2 instances of SuperPi to get 100% load, I always get an error from one of the instances:

     

    NOT CONVERGENT IN SQR05

     

    While the other instance continues to run. This is also true at complete stock settings.

     

    So is this error because I am running 2 instances when I am not suppose to, or is my system unstable even at stock voltages/speeds?


  13. After getting my memory speed down (after all this expermenting, I found the max speed), I tried to now do the same with my CPU. However, when I attempt to put it in a higher multiplier (resultant is only around 2 GHz, which is near stock speed for the CPU), my CPU wont boot, and the nbridge's fan keeps occillating from high to low.

     

    The LED diags show 3 LED, then 2 LEDs, then 4 LEDs blinks twice, then goes back to 3 LEDs, then 2 LEDs, then 4 LEDs blinks twice, and the cycle starts all over.

     

    I tried to use the "Safe Boot" jumper according the manual, no go. So I tried the full CMOS jumper reset proceedure, and no go there either.

     

    Is my CPU toast? Anyone I can fall back to, or do I have to bite the bullet and buy a new CPU? :(


  14. Have you decided on a PSU yet?

     

    Also, The Silverstone FM121 Fans (and the Nexus fans) are easily modded to mount on the XP/SI-120's without zip ties. Simply use a dremel or large file to cut each corner out to make it look like any other fan. :) I'm also hooking mine up to a 3.5" 6-channel Zalman fan controller instead of the stock one... (something you cant even do with a Zalman 7000! lol)...

     

    I purchased my Seasonic 600W here:

     

    http://www.newegg.com/Product/Product.asp?...N82E16817151025

     

    But at the time, there was a $30 MIR.

     

    BTW, do you have any pics of the modified FM121? I am not very hand, and am afraid to dremel it, lol.


  15. This is quite frustrating. I have the memory on all the most stable settings and it still can't go to 275. I've tried everything I can think of, but every combination gets errors in memtest at test 5 or sometimes 4. Am I going to have to go less than 1:1?

     

    I just thought of something while writing this...I could try to physicaly switch the memory....I heard that can help sometimes. I'll go try that.

     

    Any suggestions would be greatly appreciated.

     

     

    Make sure you are using the oranage slots. Some active cooling to the modules will help (I have an SI-120 HS/F to cool my CPU, so that large 120mm fan also cools my memory).

     

    And of course, try to turn up the voltage a bit. Mine is at 2.71v. Also, if you haven't done so already, relax your timing to 3-4-4-8.

     

    These memory modules are garanteed at 250MHz, so anything above that is gravy. Try lowering the freq to 270 Mhz if all else fails.

     

    My methodology is to decrease by 10 MHz and loop test 5. Then if all else is good, I'd increase by 5 MHz to see if I can find something inbetween.

     

    I use whatever is stable from the last test (if it works 5 Mhz higher, I 'll go with it, if not, i fall back the same 5 MHz), then run a FULL test. If that fails, I turn up the voltage. If that fails, I lower another 5 MHz, and bring the voltage back to DFI's stock voltage (2.66v).

     

    I currently use 2.66v as my low voltage, and 2.71v as my high. I probably won't go any higher because Samsungs seem to not respond to anything higer than 2.8v (and can sometimes react worse to the higher voltages).

×
×
  • Create New...