1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

1950XT running HOT, HOT, HOT-*crash*

Discussion in 'Graphics Cards' started by Dr. Spankenstein, Jul 25, 2007.

  1. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276
    Well, my foray into overclocking my Intel board with ClockGen and SYSTOOL has not gone well...
    Long story, short: Was able to get somewhere will SYSTOOL but none were stable (ORTHOS failed after 2 seconds @400 FSB & TAT 100% load shot temps directly to 60C and shutdown even @ 1.475V):banghead:

    I was also noticing that my card was running exceptionally hot. I have been juggling between ATITrayTool, ATITool and CCC. I like ATITrayTool the most, but it was giving me issues in Far Cry and Call of Juarez demo with some similar graphical anomalies (colored geometric shapes all converging at the center of the screen.) Then made the move to ATITool. Don't know what the issue is, but something about this will not let me overclock worth a nickel......next. Then I went back to CCC but Overclock (of whatever they call it) has too low of a maximum for me and no provision for voltage adjustments, but I didn't have the issues in Far Cry and CoJ. Not a fair enough trade-off. Back to ATT.....

    Now on to the temps...
    Last week I put a Thermalright HR-03 w/120mm fan and some Swiftech MC-14 ramsinks mouted w/AS5 & AS adhesive mix. All was good and I achived and personal best on 3DM06: 7372 @ 716/990.:D Temps were borderline but now that I've gone through some major hard reboots and CMOS resets and dinking around with SYSTOOL this thing gets fricking hot ina hurry! It's now 'round 40 idle @ 1.45v GPU and lock/reset within a minute of running an artifact scan!:mad:
    The only other thing I can think of that may have some impact would be that I tried to get some more cooling on the Vregs by removing the thermal tape and sanding down the risers/feet on either end so that the Vreg sink would touch the chips and applied a minimal ammount of AS5 to the underside of the sink. Then I attached the surplus Thermalright ramsinks that came w/HR-03 on the Pulse chips next to the Vregs. These came in contact with the Vreg sink.
    Could I have made some sort of bridge between some components so as to make some unwanted volt mod? Maybe some AS5 or metal filings made there way inbetween some resistors?

    Woud there be any particular places to test the resistances or voltages to try and find the source of a "short"?

    I know my CPU temps are running high as well, but hopefully I can wrangle those in. I'm just curious if anyone else has experienced sudden attacks from the Heat Gremlins after some failed overclocks.

    Would anyone suggest reflashing the BIOS in the mobo? the card? both?

    Thanks in advance for any advice!

    Cheers,

    Bryan
     
    Last edited: Jul 30, 2007
  2. theonetruewill New Member

    Joined:
    Nov 12, 2006
    Messages:
    2,996 (0.98/day)
    Thanks Received:
    240
    Location:
    London - Close your eyes and you'll see me
    40* idle isn't too bad- I wouldn't worry about it. I had to increase my vcore to 1.475v to get my GPU to reach it's MAX of 722MHz (up from 500). And for the crashes- try giving the vram a little bit more voltage.
     
    Dr. Spankenstein says thanks.
  3. Namslas90 New Member

    Joined:
    Aug 27, 2006
    Messages:
    4,851 (1.55/day)
    Thanks Received:
    555
    Location:
    Earth
    40 at idle isn't bad, but it should be lower than that with that cooler.
    Maybe, is anything touching that shouldn't be?
    Did you raise Vcore on GPU? if so take it down a notch.
    Take your card back to stock, and run it, check temps for normal operation. If Not normal then remove the Vsinks, and try again @ Stock Clocks. If still no change than look into re-doing the cooler/paste etc. If you can get it to run Normal temps at stock clocks then start your OC over; going in stages of 100's untill unstable, then -100 and go up by 10's until unstable; then -10 and up by 2's or 3's until you find the "sweet" spot for your card.

    :toast:
     
    Dr. Spankenstein says thanks.
  4. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276
    Hey man! Good to see you again! I have tried bumping up the mem volts to 2.102. Doesn't seem to make any difference except the ramsinks about burn my fingers when I touch them...
    Seems like 60C is the magic number for reboot on the card. Don't know why...
     
  5. theonetruewill New Member

    Joined:
    Nov 12, 2006
    Messages:
    2,996 (0.98/day)
    Thanks Received:
    240
    Location:
    London - Close your eyes and you'll see me
    Hmm, they are getting hot aren't they. Have you tried slightly lowering the Memory IO voltage (the other memory voltage option)? Sometimes you can lower this a tad to lower VRAM temperatures with no negative effects. However stability will need to be tested of course.
     
    Dr. Spankenstein says thanks.
  6. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276
    I have tried to bring it down to 1.425 and that will run for a while on the 3D render and Artifact scan, but at 60C *poof* reboot of the whole system. I should go back to stock but I was just surprised by the sudden change.

    Thanks for the advice on "tuning" the OC.
     
  7. DOM

    DOM

    Joined:
    May 30, 2006
    Messages:
    7,551 (2.34/day)
    Thanks Received:
    828
    Location:
    TX, USA
    have you tryed putting the CPU @ stock ?

    do you have 3D & 2D speed or is it always 716/990
     
  8. Namslas90 New Member

    Joined:
    Aug 27, 2006
    Messages:
    4,851 (1.55/day)
    Thanks Received:
    555
    Location:
    Earth
    Too little voltage can cause overheating as can too much voltage. Low volts and the card has to work harder to operate. Overworking can cause overheating.

    :toast:
     
    Dr. Spankenstein says thanks.
  9. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276
    No, that is generally the ONE setting I haven't adjusted. I'll give it a go tonight. Thanks!

    Did anyone have any measument points I could use to verify if the core or the mem is getting more volts than it should? This would put my mind more at ease if I knew....
     
  10. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276

    That makes sense. Honestly, I had been using 1.500v for the GPU before to reach 709MHz stable. Now if I were to run the 3D render or Artifact scan at that voltage, the temp shoots to 60C so fast I can't even abort to keep the reboot from occuring.

    In essence, I've tried all voltages from 1.425 to 1.525. No low-heat, non-reboot love, yet...

    Cheers,

    Bryan
     
    Last edited: Jul 25, 2007
  11. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276
    Has anyone had a card's BIOS borked from a bad OC? How about a motherboard's BIOS?
    How vulnerable are these to corruption from hard reboots?
    Nobody has commented on flashing the BIOs on the card. How does one go about reading the contents of one?
    ATIFLASH, RaBit....?
     
    Last edited: Jul 25, 2007
  12. DOM

    DOM

    Joined:
    May 30, 2006
    Messages:
    7,551 (2.34/day)
    Thanks Received:
    828
    Location:
    TX, USA
    Is your CPU OC in the bios or Windows or both cuz if it failed ORTHOS the load that atitool is putting is whats making you comp restart cuz the CPU isnt stable...

    why dont you just run the card at stock 3D and have the 2D when not using 3D apps.


    Oh yeah my comp only freezes when I push the card to far and its happen many of times and it still works and it reboots when the cpu it's stable
     
    Dr. Spankenstein says thanks.
  13. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276
    DOM,

    Now I am OCing in the BIOS. I have purged my system of SYSTOOL and am back on track. Currently at 3.5GHz w/1.485v CPU.
    The reason I assumed it was the card is the behavour before the restart. When it was the unstable CPU/heat issue it was instantaneous, when the vid card went there was the freeze/lock then restart. Is there a difference or am I assuming wrong?

    I should have first offered an apology. The reason I am in this state now is because I was after your 3DMark06 score! (that AND I wanted to help the ATI cause! Viva la ATI!!)
     
  14. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276
    The thing I'm trying to impress is that not much physically has changed (no reseating of heat sinks, no major mobo volt changes and ambient temps are cooler than they have been thanks to A/C), tried the last 3 ATI driver revisions, added passive cooling to Vregs, using same settings as before and the card has started flaming up like a Mormon before his mission!

    Seems to point to something on the card. I know I shouldn't have used AS5 on the chips, but I was in a hurry to get the new cooler on. I think if I could get some readings with the ol' multimeter it would help the process of elimination, but I don't know where to take such readings.

    I used to see GPU temps creep into the 60's before, now when it is about to update the temp (in ATT) before 60, it freezes. This occurs with all manner of voltages to the GPU.

    DAMN YOU HEAT! damn you...
     
  15. DOM

    DOM

    Joined:
    May 30, 2006
    Messages:
    7,551 (2.34/day)
    Thanks Received:
    828
    Location:
    TX, USA
    but 60's isnt high on these cards when I was on air I was in mid to high 60 load with the fan at 100%

    also what temps are you getting on the volt regs ?

    is there good air flow in your case ?

    cuz was it working before you put on the cooler ?

    you might of did somthing that it didnt like

    and im on water just so you know ;)
     
    Dr. Spankenstein says thanks.
  16. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276
    As was I DOM, as was I. That was before the my SYSTOOL fiasco.

    ATT only lists GPU temp and vicinity/pcb temps. I guess I could use ATITool to monitor vreg temps? Oddly, the temps (by feel) are much cooler now that I have the AS5 on them and the ramsinks on the (what are those Pulse chips called) others.

    The real change has been the massive ammount of heat being put out by the core. I switched to the HR-03 from the IceQ3 cooler and it made a noticable difference (for the better) in the temps. I would bench and it would get warm running 709 @ 1.500v or 1.525v, now it is radiating heat from 695 @ 1.475!?! The heat pipes comming out of the base are uber hot and all the radiator fins are warm. I mean, it good to know it's pullin' the heat, but why the sudden burst? (That's what leads me to believe either the ATT reported volts are wrong or the BIOS is setting it to something else.) Plus the vram sinks are quite hot to the touch. It is still scary hot when I reduce everything to near stock speeds (635/900). This teeters on the brink of freeze/restart territory (that 60C temp mark as reported by ATT). Something just doesn't add up. I must have smeared some AS5 on some resistor while I was screwing down the vreg sink of something.

    My mind is spinning.....

    p.s. I have been testing with the side off my case and a floorstanding fan blowing on "High" onto the whole enchilada. Ergo, I thing the airflow is decent.:eek:

    p.s.s. My latest scores came after I replaced the cooler. SYSTOOL has fucked up my whole world!!
     
    Last edited: Jul 25, 2007
  17. DRDNA

    DRDNA

    Joined:
    Feb 19, 2006
    Messages:
    4,805 (1.45/day)
    Thanks Received:
    573
    Location:
    New York
    I hate to even throw this in ...but did you check to see if your cards vents are clogging? Maybe a can of air is in order?
     
    Dr. Spankenstein says thanks.
  18. insider Guest

    Always use ceramique paste instead of AS5, there is about <1C difference between those paste.

    60-65C core is safe, it's when the core starts running near and above 80C load that you really ought to start to worry.
     
  19. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276
    Yeah, I put the cooler on the card last week. We don't have any pets and we don't smoke. I've had this build together since Nov. 06 and ther is barely any dust build-up on my intake fan or my Tuniq Tower.

    Thanks for kickin' in some help though. It is appreciated.

    Cheers,

    Bryan
     
  20. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276
    I know, I've been there once....;)

    It's just that now, as soon as the temp reaches 60C (on the card) it freezes then restarts.
    This is quite frustrating.

    Good point about the Ceramique. I'm an impatient bastard.....I probably deserve this!:confused:

    Regards,

    Bryan
     
  21. DRDNA

    DRDNA

    Joined:
    Feb 19, 2006
    Messages:
    4,805 (1.45/day)
    Thanks Received:
    573
    Location:
    New York
    Is the cooler still on right ...did you check to see if anomaly of mysterious un-tightening?
     
  22. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276

    Yep! It's sittin' right in front of me! Luckily for me, the Heat Gremlins lack the strength to loosen these thumbscrews!:laugh:
    Personally, I don't think I would be able to notice (by touch alone) a difference in the heat that is being generated since the SYSTOOL disaster if the cooler wan't making good contact with the core.

    I like your troubleshooting logic though. Spock would be proud!!:p

    Cheers,

    Bryan
     
  23. DOM

    DOM

    Joined:
    May 30, 2006
    Messages:
    7,551 (2.34/day)
    Thanks Received:
    828
    Location:
    TX, USA
    Do you still have the thermal tape from the Volt Regs. HS ?

    Cuz you sanded the ends was the HS all Copper or Aluminum with Copper plated ?

    I would try putting the tape back on, clean the regs with some 90%+ Alcohol see if that helps

    well I wont be on untill tomorrow GL
     
  24. insider Guest

    Ahh sounds like the PCB and/or the other components are overheating (could be due to paste shorting the board), install heatsinks on all mosfets, vrm's, controller chips etc.

    My X1950GT even with everything HS'd the PCB tops 55C, without the heatsinks stuck on using epoxy the card would have been toasted by now! :banghead:
     
  25. Dr. Spankenstein

    Joined:
    Feb 15, 2007
    Messages:
    1,920 (0.65/day)
    Thanks Received:
    276
    Well all (like you care;)),

    I got things somewhat sorted out. I took off the vreg sink, low and behold ther was some stray AS5 on more than a few of the resistors.:eek: WHOOPS!
    Cleaned it with 91% alcohol followed w/ pure acetone and thing seem to be pretty much back to normal. (Now I just need to uninstall the stinkin' 7.7 drivers!) I scavenged the "gummy" tape from the stock ICEQ3 vram sink for the vreg sink. The two uppermost chips seem to be the biggest contributor to the heat. Does anyone have a niffty solution to keeping the vregs cool. I have a feeling that my resets can be traced back to these becoming overheated.

    @ DOM,

    No, I trashed the thermal tape, but for the record ALL of the copper "colored" components are coated aluminum! I found this out when I first removed the ICEQ3 cooler to check the surface. I pulled the foam shim off and the layer of coating came with it! ZOINKS! Actually the heatsink almost looked like brass or some other alloy.

    Anyhoo, thanks everyone for your concern and suggestions.
    I love you guys (in a purely geeks-helping-geeks kind of way)!:eek:

    Cheers,

    Bryan
     

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page