MAX temp for Q6600?

Discussion in 'CPUs and Overclocking' started by LOUISSSSS, Dec 3, 2007.

  1. LOUISSSSS

    LOUISSSSS Diamond Member

    Joined:
    Dec 5, 2005
    Messages:
    8,395
    Likes Received:
    1
    anyone know whats the "Intel-suggested" max temp for the q6600? i wanna push my cpu to the max just o see how far i can get with OC
     
  2. BonzaiDuck

    BonzaiDuck Lifer

    Joined:
    Jun 30, 2004
    Messages:
    10,925
    Likes Received:
    37
    You didn't say which stepping (B3 or G0) of the Q6600.

    For the B3, with the most efficient air-cooling and a case ( or even ducting) that moves fresh air over the hot components and out the rear quickly, you should expect a 33% over-clock to maybe 3.2 Ghz at just above the intel retail-box "maximum" spec (no more than VCore_Max + 5%), and you should get temperatures that are just at, or just below 65C at load when the room ambient is close to 78F.

    With less efficient cooling, that sort of over-clock could push your core temperatures to 70C +/- 2C-degrees.

    The TCase temperature should be between 9C and 15C lower than the four-core average of core (TJunction) temperatures, and TCase is the basis for BIOS-enabled throttling when it reaches 62 to 65C.

    From what I've seen, the G0 stepping will over-clock to maybe 3.4 Ghz under the best scenario I've outlined above -- maybe farther -- and I'm just guessing from reports around here and impressions I've had that the temperatures would be at least 5C lower than for the B3.
     
  3. LOUISSSSS

    LOUISSSSS Diamond Member

    Joined:
    Dec 5, 2005
    Messages:
    8,395
    Likes Received:
    1
  4. LOUISSSSS

    LOUISSSSS Diamond Member

    Joined:
    Dec 5, 2005
    Messages:
    8,395
    Likes Received:
    1
  5. TheOtherRizzo

    TheOtherRizzo Member

    Joined:
    Jun 4, 2007
    Messages:
    69
    Likes Received:
    0
    AFAIK

    Tjunction: the temperature inside the chip
    Tcase: the temperature on the case of the chip

    Tjunction are the temps that "Core Temp" tells you. Tcase are the temps that your BIOS tells you. I think Speedfan shows both Tcase and Tjunction. What "Core Temp" calls Tjunction is in fact Tjunction Max as specified by Intel. And "Thermal Specification" is Tcase Max.

    For AMD it's similar except that AMD doesn't tell you what Tjucntion Max is so "Core Temp" is pretty useless for AMD.

     
  6. LOUISSSSS

    LOUISSSSS Diamond Member

    Joined:
    Dec 5, 2005
    Messages:
    8,395
    Likes Received:
    1
    where can i get a readin for TCase? is it the first sensor that speedfan gives u?

    and also, what is the significance of Intel's rating of 71C
     
  7. myocardia

    myocardia Diamond Member

    Joined:
    Jun 21, 2003
    Messages:
    9,223
    Likes Received:
    0
    Yes, Tcase is Temp #1 in all versions of SpeedFan. Well, for AMD, Temp #1 is the CPU socket temp.
     
  8. TheOtherRizzo

    TheOtherRizzo Member

    Joined:
    Jun 4, 2007
    Messages:
    69
    Likes Received:
    0
    It's Tcase Max.

     
  9. BonzaiDuck

    BonzaiDuck Lifer

    Joined:
    Jun 30, 2004
    Messages:
    10,925
    Likes Received:
    37
    I thought it was between 62 and 65C, but if Intel says 71C, then that's the maximum acceptable TCase value, alright. It may be that the 62/65C spec (if I'm correct) is the temperature at which the processor will throttled as enabled in the BIOS. Or it may be that I'm wrong, and it throttles at this 71C spec.

    The thing to remember is that less heat is better even as you choose that higher over-clocking is "better." The other thing to remember is that you can expect room-ambient temperature to raise both TCase and the TJunction core temperatures degree-for-degree. So if you plan to run your computer at room temperature of 80F or less, and if TCase can be held below 70C, then you won't have to complicate your over-clocking with throttling enabled in BIOS.

    Since temperatures will increase (assuming constant room-ambient) by the square of the voltage and linearly with CPU_FSB speed, I hold the view that there is some point in a parabolic or increasing-exponential schedule of temperatures where you need to limit your VCore increases especially, and your FSB (Mhz) setting generally.

    Personally, I choose to limit my over-clocking to a VCore point where the core temperatures have increased no more than maybe 10C -- and in fact, I feel more comfortable if I can over-clock with little perceptible increase in temperature at all. to my recollection, at room-ambients above 70F, the load TCase temperature on my Q6600 is just below 40C with the stock setting. Right now, the load value is around 50C, and in the summer, it can get as high as 55C with the same OC settings. This means that my "core" or TJunction values are ranging between 60 and 65+C depending on room-ambient.

    YOU'RE the one who has to decide how much risk or uncertainty you will face with the prospects of frequent processor or motherboard purchases, but then, if the normal processor lifespan is 10 years, and you're going to replace it in no more than two or three years anyway, you're the one who has to choose the parameters you're willing to live with.

    This is really all about the industrial application of statistics. If Intel chooses to print "1.35V maximum" on the retail box, they must, as a business, be pretty sure that RMA's under warranty will be nothing or next to nothing. That's their view of the failure rate under a schedule of voltages in their testing labs. Nobody can tell you (except Intel -- and they probably won't anyway) whether a 5% increase in voltage above that level means a 0% or 1% or 2% or a 5% or a 10% or a 20% increase in failure over one year, two years, etc. due to electron-migration and regardless of how well you cool the system. Similarly, there is some increasing schedule of risk when you run the system constantly at temperatures approaching the 71C spec -- increasing with each degree closer to the spec, and certainly, with each degree beyond it.

    We count on two things: that Intel's specs are based on their desire to reduce costs and customer-relations troubles, and that we're going to get a processor that is either "average" in relation to these schedules of risks, temperatures and voltages, or that we'll "get lucky" and obtain a processor at lower price performing at the level of their "flagship" models, yet was mistakenly binned for the model which reflects the lower price.

    As geeks, and generally as a society, we need to accommodate ourselves, first -- to the idea that random variation IS a form of order in Nature, and second, that uncertainty, probability and risk pervade every aspect of life, and even those aspects which derive from careful engineering and well-controlled production processes.
     
  10. myocardia

    myocardia Diamond Member

    Joined:
    Jun 21, 2003
    Messages:
    9,223
    Likes Received:
    0
    62°C is the Tcase max for B1, B2, and B3 steppings, along with the Allendale steppings that correlate to those steppings. 71°C is the Tcase max for G0 and M0 steppings. And all steppings throttle above their Tcase max point-- throttling is bad, and wears out the processor prematurely, so Intel doesn't want them throttling (and you shouldn't either).

    Throttling should never be turned off; it doesn't affect your overclocking in any way, it only protects your processor, should your heatsink's fan fail.
     
Loading...