Data on geek bench of various macbook cpus and 2015 macbook air

Discussion in 'MacBook' started by gix650, May 7, 2015.

  1. gix650, May 7, 2015
    Last edited: May 7, 2015

    gix650 macrumors newbie

    Joined:
    May 7, 2015
    #1
    I made a spreadsheet to see how much of difference there was between the 1.1 and 1.3 cores. I also threw in the 2015 macbook air because that is what I have used and wanted to see how the rMB held up. On the right is a data plot blue=5y31 green=5y71 grey=Macbook air 2015 core i5 1.6. Scroll down the data set to see the actual means, standard deviations and confidence intervals for all three data sets.

    I took the first 17 scores for each processor from geek bench. I also threw out extremely weird scores that were significantly lower than the norm, for example there was one score in the 1500s for the 1.1 that I threw out and another in the 900s for another cpu.

    All the usual disclaimers apply, this is just a benchmarking tool, the sample sizes are extremely low etc. etc.

    THIS IS FOR THE 32 BIT VERSION OF GEEKBENCH 3. I ONLY USED SINGLE CORE SCORES.

    Overall the 1.3 has a very large standard deviation, some of the scores are as fast as the macbook air 2015, this is probably reflective of the unique architecture of the core M which allows it to throttle its speeds up and down. While 1.3's fastest scores are much faster than the 1.1's fastest scores; the average of the scores aren't that dramatically different for the two cores, in fact the average scores between the 1.1 and 1.3 are barely statistically significant for this small sample size.

    All of this points to the fact that the 1.3 can operate at speeds nearing the 2015 macbook air, but otherwise operates barely faster than the 1.1. How long and how well the 1.3 can operate at the macbook air's levels are all presumably dependent on heat, something which is a significant limiting factor the rMB. In the real world, how long the 1.3 can operate at high levels is unknown.

    Yes this is a terribly low sample size, and yes I was really bored. Hoped you had fun if you stuck with me!
     

    Attached Files:

  2. pookitoo macrumors regular

    pookitoo

    Joined:
    Apr 16, 2015
    Location:
    Paris
    #2
    Very interesting post ! You take the score for a website or you do the test ? When I see your result , seem the difference are very low. I hurt my head to know which one to take ...
     
  3. thitiv macrumors member

    Joined:
    Dec 3, 2011
    Location:
    Bangkok, Thailand
    #3
    Thank you very much. I have been using the base MacBook with 1.1 GHz CPU for two days. Your scores confirm my intuition: benchmark software and OS may have to be adjusted to fit the nature of Core M CPUs as well. What surprise me is that it plays MKV bluray rips smoothly and the bass audio is much better than those from my Air 11" and my former 15" retina pro.
     
  4. gix650, May 7, 2015
    Last edited: May 7, 2015

    gix650 thread starter macrumors newbie

    Joined:
    May 7, 2015
    #4
    Using traditional utilities for benchmarking these kinds of processors certainly seems like a unreliable way of going about this due to their unique nature.

    I really ended up doing this because in every post about this somebody would throw up wildly different scores for the same processor, it was very difficult figuring out how these actually were performing on average. Additionally some people would throw up gb 2 scores some 32 some multi core etc. etc. I just controlled for all of these to find a nice average and std.

    I did not do these tests myself they were taken from the geekbench website and so were performed on different machines, presumable under different circumstances.
     

Share This Page