Overclocking Q6600

airbornflghtairbornflght Houston, TX Icrontian
edited February 2008 in Hardware
Ok. So I started overclocking my computer.

I have a Gigabyte DS3L motherboard, and a Q6600 with 4 gigs of ram (2x2). Power supply is 520w Corsair.

Anyways. I took it to 350*9 right off the bat just to see where I'm at and it is fairly evident that I didn't apply the thermal compound correctly. Two of the cores are running dramatically hotter than the other two. So that needs to get fixed.

Right now my cores are idling around 32-39 degrees depending on the core. Again, Cores 0 and 2 are the hottest.

Core 0: 38
Core 1: 32
Core 2: 35
Core 3: 32

Do you all have some pointers? I'm fairly new to the whole overclocking game. I've heard 3.4 should be reasonable expected. That really what I'm aiming to hear. I'm a little confused on how hard I should push the memory. I did 6 minutes of prime with no error so that's a start. If I could get some general direction on what to do that'd be apreciated.

Ie what frequency to keep the ram around with dividers and the such and also voltages. What are norms and what is the safe range

Comments

  • ThraxThrax 🐌 Austin, TX Icrontian
    edited February 2008
    That's not uncommon. You have an IHS that's not mounted quite correctly.
  • SnarkasmSnarkasm Madison, WI Icrontian
    edited February 2008
    You're not even close to a problem with your temps. A lot of the quads have the same problem; I've even lapped mine and I'm still off 3-5 degrees between core pairs. You don't mention what RAM speed you have, so I'm going to guess DDR2-800, and if that's wrong, change your calculations to fit what you have. If you're aiming for 3.4, I'd start with an 8x425 FSB and use a 2x divider for the RAM to get you just a little bit over stock at 850MHz. As far as voltages go, the RAM depends on the manufacturer (OCZ is usually rated to 2.2V EVP, with 2.1V recommended, my Ballistix can take 2 or 2.1 easily), and the chip itself can handle up to 1.5V. As far as board voltages, my last board, a GA-P35-DS4, was stable with +.05V MCH, +.3V FSB, and that's it. My DQ6 has taken a little more prodding, but can take +.15V MCH and +.35V FSB, along with some other small tweaks. Unfortunately, I don't know if there's any way to tell what your fatal voltages may be, though others here might know. Really just trial and error. If something's terribly wrong, the BIOS will catch it and reset everything to defaults, though, so have a good memory or keep records of what you're doing in case you go too far and it decides to save you.

    By the way, is this a new build? If it is, or if you've recently reseated your HS, if you're using a TIM that has a curing time like AS5, nobody really recommends trying to OC before the TIM has cured, it's just interfering with the curing process. If I missed some previous threads where your build was discussed, apologies for not reading it. :)
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2008
    Airborn, those core temperature spreads are nothing unusual. I had a CPU that had a larger spread than that. Some people lap the IHS, which will usually even out the core temps some. I lapped my Q6600s and it made a dramatic difference with two CPUs and just a little difference with another.
    f you're using a TIM that has a curing time like AS5, nobody really recommends trying to OC before the TIM has cured
    TX-2 - needs no curing
    MX-2 - same, I believe
  • muddocktormuddocktor
    edited February 2008
    Airborn, idle temps don't mean much. What are your load temps looking like? That is much more important than idle temps. You can load all 4 cores with the latest version of prime95 natively (running only 1 instance of it), if you aren't a folder or other DC project person.

    BTW, you do know that we have a special forum just for overclocking and cooling issues, right? :rolleyes:
  • airbornflghtairbornflght Houston, TX Icrontian
    edited February 2008
    Refer to attachment for load temps.
  • Your-Amish-DaddyYour-Amish-Daddy The heart of Texas
    edited February 2008
    100 hard faults?! I never even got one.
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2008
    Wow, you've got a spread of 15C at load between the four cores. That's not an IHS problem. That's indicating you've got only two cores under load. There is just no way you could get that delta with all cores at full load.

    Do a screen shot of the Task Manager Performance tab window.
  • muddocktormuddocktor
    edited February 2008
    Airborn, what kind of cooler are you using? Is that with the stock hsf that comes with the Q6600? If so, check and make sure that all 4 pins are securely through the mobo and it's mounted correctly. Those kind of temps look to me like one side of the hsf isn't making good contact or has uneven pressure holding it to the cpu. BTW, if you are using the retail hsf that came with the Q6600 I recommend that you upgrade to a better cooling solution such as a Tuniq or U-120-X if you plan to overclock more than just a little. You shouldn't run into any clearance issues mounting them on that DS3L board either. When I get in from work I will be replacing the XP90-C with an U-120 on my DS3L system and doing some fan testing.

    EDIT: Leo, his attached screenshot in his first post shows the cpu under 100% load and with Prime95 showing all 4 cores with threads running. I think it looks like a bad mount myself. One side isn't getting even pressure and the 2 cores on that side are running hotter.
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2008
    Is that with the stock hsf that comes with the Q6600? If so, check and make sure that all 4 pins are securely through the mobo and it's mounted correctly.
    Ooh, yeah, hadn't thought about that. I'm so used to installing an aftermarket cooler I had forgotten about the Intel HSF mounting system. Right, just one loose cam on that heatsink mount, and you are asking for problems.
  • airbornflghtairbornflght Houston, TX Icrontian
    edited February 2008
    I have an aftermarket cooler, of the Tuniq Tower 120 persuasion. So a remount is being prescribed?
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2008
    I run two of the Tuniq Towers. They are very good CPU coolers. It's very easy when mounting the heatsink to the motherboard base hardware to not cinch (sp?) the heatsink down evenly. I would recommend reseating the heatsink and taking particular care to rotate the spring loaded pressure screws all the way. Ensure that you tighten them down incrementally, just as you should do with any fitting or structure that has nuts and bolts arrayed in a radial or square pattern. Slightly tighten a screw then move on to it's opposing counterpart, then the next screw, followed by it's opposing counterpart. I'm sure you already did this - just a reminder in case you didn't.
  • airbornflghtairbornflght Houston, TX Icrontian
    edited February 2008
    YEh, I went in a star pattern but I didn't go all the way down as a result of no instructions with the cooler and I wasn't sure how far to go down. I was afraid of damaging the processor.
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2008
    Assuming the CPU socket was installed correctly on your motherboard by the manufacturer - reasonable assumption, you should torque the four screws down all the way. Do though, stop when you feel them bottom out. You may need to experiment a little to determine bottom dead center. It took me a couple re-installations with my Tuniq 120s before I finally found the fully torqued position. Yeah, I was also afraid of applying too much torque. BTW, I recently read an article at a tech site (forgot which one) where they directly addressed the topic of force needed on CPUs for best CPU-heatsink heat. Intel's specification for optimum heat transfer was surprisingly high. No kidding. You won't damage the CPU by rotating the retention screws all the way.
  • airbornflghtairbornflght Houston, TX Icrontian
    edited February 2008
    Ok, I was worried about either cracking the die or even the motherboard pcb. Seems like an awful lot of force being applied to it.
  • ThraxThrax 🐌 Austin, TX Icrontian
    edited February 2008
    Imagine the hell that would be raised if some company released a cooler with no precise instructions on how well the sink should be seated...
  • airbornflghtairbornflght Houston, TX Icrontian
    edited February 2008
    Well, a paragraph would have been appreciated. I've never seated a processor that actually had to be adjusted before. The Socket 754 heatsinks just had a lever that tightened the heatsink down.
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2008
    I understand, believe me.
  • Mikail-CritchkiMikail-Critchki Here
    edited February 2008
    I usually use a hammer when things doesn't go my way.
  • muddocktormuddocktor
    edited February 2008
    On my Tuniq I just tighten them down until they won't tighten any more, just like Leo said. With the LGA socket and with a backplate, it would be very hard to actually crack the motherboard since the backplate is there to distribute the pressure over a very wide area of the board
Sign In or Register to comment.