3 Replies Latest reply on Oct 25, 2010 9:30 PM by Lucretius

    i7 980x Amperage Readings

    Lucretius

      Hello,

       

      I've been working with the Asus Rampage III Extreme ROG Connect feature and I am trying to verify the results.
      I have an i7 980x in the board running at stock for now, and I've been monitoring the amperage draw using ROG Connect.

       

      Here  are some screenshots of the ROG diagram while the CPU is  totally Idle, and at 100% load (with stress testing software).

       

      http://i53.tinypic.com/iy3gvq.png

      Thats 4.44 amps (sort-of average) while Idle

      ...and 41.89 amps (sort-of average) while at 100% load using LinPack x64 for stress testing.

       

      http://i51.tinypic.com/11w699h.png

       

      It's showing 1.164v at 100% load.
      It's showing 1.224v at Idle.

      I've verified this to be accurate using a multimeter and software monitoring tools (AIDA64, CPUz)

      (Load Line Calibration is set to 0%)

       

      I  need to know how accurate ROG Connect is for monitoring these values  for the CPU.  I've cross checked the voltage readings and they are 100%  accurate.  I can't really just assume that its accurate because ROG  Connect is a pretty new feature.  I need to have some Intel people look  at it and comment on the Amperage readings for the i980x.

       

      Using  the formula [Watts = Amperage * Volts] I'm looking to use the Amperage  and Voltage values from ROG connect to isolate the Wattage of the CPU at  a given voltage while at 100% load from the rest of the system.

       

      So  if my wattage on my wall meter reads something like ~400W I can deduct  what I find from this to see how much my system is using without the  CPU.

       

      What do you guys/gals think?  Do these Amperage readings look right?

        • 1. Re: i7 980x Amperage Readings
          Terabytes

          What GFlops did Linx show?

           

          Part of the power to drive the CPU will also be across the regulators and the efficency of the power supply. For instance if the CPU were drawing 100W then that might require 110W or more at the wall.

          • 2. Re: i7 980x Amperage Readings

            Here is a link to my test results so far.  I've tested the chip from 3333MHz to 4256MHz.

             

            https://spreadsheets.google.com/ccc?key=0AiqWJwYxhJqKdGl5dlBHRzBCSXNuVmQ4ZUFZWkJ2ZUE&hl=en

             

            LinX performance was a relative constant as I increased the multiplier, increasing from 64.312GFlops @ 3333MHz to 78.9442GFlops @ 4256MHz

             

            The purpose for isolating the CPU power consumption is for understanding how increasing Amerpage affects Max VID according to the datasheet linked below.

             

            http://download.intel.com/design/processor/datashts/323252.pdf

             

            My understanding of table 2-8 on page 25 is that as Amperage "Icc (A)" increases (i.e., the more you overclock) the lower your Max VID "Vcc MAX" tolerance becomes lower.  This is being done for the purpose of understanding the theoretical maximum "smart" overclock that yields performance benefits without exceeding specifications on table 2-8.

            • 3. Re: i7 980x Amperage Readings
              Lucretius

              To explain in more detail,


              The end result is a power profile for my CPU Overclock.  I'll explain.

              http://download.intel.com/design/pro...hts/323252.pdf
              According to page 25, figure 2-8, of this data-sheet...

              As Amperage "Icc (A)" increases (i.e., the more you overclock)
              The lower your max tolerable voltage is beginning at whatever your Max  VID is (in my case 1.375v), decreasing linearly 5A*(-0.004v)
              ...that is, if I'm reading it correctly (and I'm on Intel forum asking the same stuff to find out).

              Once I know a fairly accurate Amperage usage for my CPU at a given  overclock, I can use that information to find out what the Vcc_MAX is at  that voltage.

              If this can be accomplished there are numerous benefits.
              - Finally separating where the line is drawn between a "golden chip" and  a "crappy chip" or some "average chip" based on hard numbers.
              - People will have solid evidence for how far they can reasonably expect  to push their CPU without fear of degradation or risking failure  because this work can be replicated for every Intel CPU type.

              Here are my tests so far.
              https://spreadsheets.google.com/ccc?...ZWkJ2ZUE&hl=en

              Again the question is quite simple. How accurate do those amerpage readings look for the i7 980x?