Announcement

Collapse
No announcement yet.

Intel vs. AMD Performance-Per-Watt On Ubuntu 14.04 Linux

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #31
    Originally posted by tuke81 View Post
    Like this:
    cpu i3 2120 i5 3470 i7 3770k i3 4130 i5 4670 i7 4770k A10 7850k
    P_avg[W] 63,7 76,4 96,8 75,6 91,1 113,8 116,3
    time[s] 202,1 118,8 88,92 161,96 94,03 75,43 163,14
    Energy[J] 12874 9076 8607 12244 8566 8584 18973

    That's total energy which isn't what this article is testing exactly. Increasing time and lowering power would give the same total energy but would you actually prefer that? If so, then there are ARM and Atom chips that might give the same total energy.

    I want 1/time/energy which could also be converted to 1/time/(average power) to get performance per Watt.

    Comment


    • #32
      Originally posted by vick View Post
      That's total energy which isn't what this article is testing exactly. Increasing time and lowering power would give the same total energy but would you actually prefer that? If so, then there are ARM and Atom chips that might give the same total energy.
      Yes, total energy required to perform a computation is a reasonable thing to be concerned about if your computer can be powered off or powered down (i.e., sleep or hibernation) into a low power state after the computation is done. And I doubt there are computers with an ARM CPU that can do the same computation for less total energy than a computer with an i7 haswell CPU (remember that it is not just the CPU that requires power during the computation).
      Last edited by jwilliams; 01-18-2014, 02:43 PM.

      Comment


      • #33
        Originally posted by vick View Post
        I want 1/time/energy which could also be converted to 1/time/(average power) to get performance per Watt.
        No. energy is not equal to average power. Power is equal to energy per unit time.

        Also, one way of measuring performance is in operations per second. Watts are equal to Joules per second. So we can write:

        Code:
        Performance per Watt = ( operations / second ) / ( Joule / second ) = operations / Joule
        Note that by operations I am talking about a generalized, high-level kind of operation, not a CPU instruction. So the number of operations is the same for a given compilation, independent of the CPU. So we can ignore "operations" in the above formula and only look at the energy usage in Joules. Which is just what tuke81 showed us.

        Again, good job tuke81 !

        Comment


        • #34
          It would be really interesting to run those tests with Kaveri at 45W and 65W, says the AMD guy

          If the graphics performance scales comparably to Windows benchmarks then a back-of-envelope calculation suggests that the FPS/W numbers on Tropics would end up around 0.17 (15.7 fps / 94.8 w) at 65W and 0.19 (13.9 fps / 74.8W) at 45W, but I haven't actually run the tests and my 2-core low clock 1-stick-of-RAM Kaveri engineering box would not be a particularly good test bed.

          The interesting thing is that in the lower power settings the biggest hit from 95/100W to 45W is only about 25% (GPU shader & texturing), CPU hit is maybe 17%, and AFAICS the ROP-limited GPU functions don't drop at all because the # of CUs is reduced from 8 to 6 but the # ROPs and GPU clock seem to be unchanged.
          Last edited by bridgman; 01-18-2014, 03:00 PM.

          Comment


          • #35
            Yes, total energy required to perform a computation is a reasonable thing to be concerned about if your computer can be powered off or powered down (i.e., sleep or hibernation) into a low power state after the computation is done. And I doubt there are computers with an ARM CPU that can do the same computation for less total energy than a computer with an i7 haswell CPU (remember that it is not just the CPU that requires power during the computation).
            I have no numbers about my ARM/Atom statement. It was just to emphasize that total energy isn't the whole picture. I concede that knowing what the work load is related to that total energy is useful. Maybe Michael can run ARM and Atom tests to see how much energy is used for the same workload.

            Originally posted by jwilliams View Post
            No. energy is not equal to average power. Power is equal to energy per unit time.

            Also, one way of measuring performance is in operations per second. Watts are equal to Joules per second. So we can write:

            Code:
            Performance per Watt = ( operations / second ) / ( Joule / second ) = operations / Joule
            Note that by operations I am talking about a generalized, high-level kind of operation, not a CPU instruction. So the number of operations is the same for a given compilation, independent of the CPU. So we can ignore "operations" in the above formula and only look at the energy usage in Joules. Which is just what tuke81 showed us.

            Again, good job tuke81 !

            That's why I said converted to get a performance per Watt number. However, I can see how my wording can be confused to be seen as equal.

            I still prefer performance/Watt or operations/Joule over total energy per work load.

            Comment


            • #36
              Originally posted by tuke81 View Post
              Like this:
              cpu i3 2120 i5 3470 i7 3770k i3 4130 i5 4670 i7 4770k A10 7850k
              P_avg[W] 63,7 76,4 96,8 75,6 91,1 113,8 116,3
              time[s] 202,1 118,8 88,92 161,96 94,03 75,43 163,14
              Energy[J] 12874 9076 8607 12244 8566 8584 18973
              This is very well done. Thank you. I prefer total energy usage as you provided.

              Comment


              • #37
                Originally posted by vick View Post
                That's why I said converted to get a performance per Watt number. However, I can see how my wording can be confused to be seen as equal.

                I still prefer performance/Watt or operations/Joule over total energy per work load.
                I'd like to say that I prefer performance/Watt or operations/Joule from an active consumer use point of view. I can see energy/workload can be useful for servers and automatic background tasks but when it comes to consumer use like browsing the internet, playing a game, writing an office document, etc., the performance/Watt or operations/Joule are a better metric for a sense of efficiency. At least for me who does more regular Joe Schmo tasks.
                Last edited by vick; 01-18-2014, 03:23 PM.

                Comment


                • #38
                  Originally posted by vick View Post
                  I'd like to say that I prefer performance/Watt or operations/Joule from an active consumer use point of view. I can see energy/workload can be useful for servers and automatic background tasks but when it comes to consumer use like browsing the internet, playing a game, writing an office document, etc., the performance/Watt or operations/Joule are a better metric for a sense of efficiency. At least for me who does more regular Joe Schmo tasks.
                  I already showed that performance per Watt and operations per Joule are the same as just measuring the total energy (for a given computational task). I can make no sense of what you wrote above. It sounds like you are confused. Note that Joules are a unit of energy, and operations and computational workload are the same thing.

                  Comment


                  • #39
                    Originally posted by jwilliams View Post
                    I already showed that performance per Watt and operations per Joule are the same as just measuring the total energy (for a given computational task). I can make no sense of what you wrote above. It sounds like you are confused. Note that Joules are a unit of energy, and operations and computational workload are the same thing.
                    Maybe my English isn't very good. I'm sorry. I was using those terms like they're equal. Should I have said performance/Watt (operations/Joule) or performance per Watt / operations per Joule (this looks more confusing to me though)?

                    I still say that for some cases energy/work load can mislead people if the energy saved was at a cost of longer than desired time or performance.

                    Do you have numbers for energy/work load on recent Atom chips? Maybe there's a Phoronix article somewhere.

                    Comment


                    • #40
                      Originally posted by jwilliams View Post
                      INote that Joules are a unit of energy, and operations and computational workload are the same thing.
                      I should clarify that "workload" is not a well-defined term. Depending on interpretation, it could mean a fixed amount of work (i.e., 1000 lines of code that perform some operation), or it could be an amount work per unit time (i.e., operations per second). I suggest you not use that word if you want to improve your understanding of Performance per Watt.

                      Comment


                      • #41
                        Originally posted by vick View Post
                        Maybe my English isn't very good. I'm sorry. I was using those terms like they're equal. Should I have said performance/Watt (operations/Joule) or performance per Watt / operations per Joule (this looks more confusing to me though)?
                        Performance per Watt and Operations per Joule are equivalent, as I already showed in a previous post. But what you are apparently missing is that total energy consumed to perform a given computational task (such as a compilation) is ALSO EQUIVALENT to Performance per Watt and Operations per Joule, as long as the computational task is fixed. Which, of course, it is for a compilation (well, there could be a slight variation if the compiler has to do slightly different optimizations for each CPU, but that is likely a small enough variation to ignore).

                        Comment


                        • #42
                          Will these graphs be updated or should I do what tuke81 did for the other tests? I guess I can grab the numbers and use Octave to make myself some easy to read bar graphs.

                          Comment


                          • #43
                            Originally posted by guido12 View Post
                            Will these graphs be updated or should I do what tuke81 did for the other tests?
                            I wouldn't hold my breath. This isn't the first time Michael posts obviously wrong performance-per-watt results, people complained, provided the correct formulas to use and yet here's another bad set of results. And I'm sure it's not the last either, so if you'd like to actually post the correct results I'm also interested in that

                            Comment


                            • #44
                              Originally posted by H.F.P View Post
                              The confusion comes from the term "power consumption". Many people think "power consumption" to be equal to "energy consumption". In fact, the energy consumption is the integral of power consumption over time. Therefore your complaint is more about the fact that you would like to see a conclusion about "energy consumption". On the other hand, the problem with energy consumption comes with workloads that usually run "forever" or at least not with a fixed amount of time e.g., playing a computer game. For the latter, power consumption can be treated equivalent to energy consumption.

                              Anyhow, it can't be repeated often enough: consuming a little bit more power over a much shorter duration still holds a lower energy consumption.
                              What are you talking about? These graphs are wrong no matter what you call it. It's precisely because of what you say in your last sentence that it's obvious that the graphs are wrong. When two cpus take the same time to finish a task but one of them uses 40W more power (or energy, or whatever you want to call it) to do it, then it's obviously a lot less power efficient. These graphs show the complete opposite!

                              Comment


                              • #45
                                Originally posted by guido12 View Post
                                Will these graphs be updated or should I do what tuke81 did for the other tests?
                                I doubt it. I've never seen Michael fix an article, no matter how egregious the error.

                                I actually came here to post a comment about the graph fail, but was gratified to see that everybody is already all over it.

                                Seriously, how could those graphs in the article ever have been published? They're so obviously, obviously wrong. You don't even have to think - they leap out and slap you around the head with a wet fish.

                                Finally, (physicist here) I'd like to see two values plotted:

                                Performance per Watt = benchmark result / (average power * time / time) = benchmark result / average power over task (units of per Watt)
                                Energy needed per task = average power * time taken (units of Joules)

                                Of course, the performance per Watt formula assumes higher is better, so for those benchmarks where lower is better you'd need to use 1/result or something for the performance.

                                Comment

                                Working...
                                X