How do I even tell what's faster nowadays?

Discussion in 'Buying Tips and Advice' started by Vexir, Aug 3, 2016.

  1. Vexir, Aug 3, 2016
    Last edited: Aug 3, 2016

    Vexir macrumors newbie

    Joined:
    Apr 6, 2013
    #1
    I'm trying to figure out what used or refurb Mac I should buy for my small business. I bought a Late 2014 Mac Mini (the 8gb RAM one) designed to replace a 2011 Macbook Pro. Turns out it's not fast enough, so I'm likely to return it and get something else. Trouble is, I'm not sure where to draw the line.

    It used to be the case that you could look at clock speed and get a pretty good idea of what was better. Or even the edition of the processor. Take a look at these two benchmarks:

    2011 Macbook Pro - http://browser.primatelabs.com/geekbench3/7629399
    Mac Mini - http://browser.primatelabs.com/geekbench3/7629333

    As you can see, the scores are almost identical. In fact, the MBP has an i7 @ 2.8, compared to the seemingly worse Mac Mini with an i5 @ 2.6. And yet, when building our software, build times were almost 50% faster with the Mac Mini.

    So my question is, what is today's method of looking at a product page, particularly for a MBP/Mini/iMac, and knowing what's faster and by how much? I need to know that to make a cost-effective decision without just blowing a bunch of $$ on the latest and greatest.

    EDIT: Also, WHAT?! Why is a 2012 Mac Mini orders of magnitude faster (with a Quad core i7 @ 2.3) than something from 2014?

    EDIT2: A benchmarking tool that actually shows a proper difference in performance would be great as well. The GeekBench scores that I linked to above clearly did not communicate the 50% difference in build times.
     
  2. theluggage macrumors 68040

    Joined:
    Jul 29, 2011
    #2
    So... if the Mac Mini you bought is 50% faster than the MBP, what's the problem? Its about the best you can hope for if you stick to the laptop/mini (the mini is basically a laptop in a desktop case) form factor - hardware hasn't improved that dramatically. The benchmark sounds about right: 2014, 28W i5 about equivalent to 2011 35W i7.

    Frankly, I suspect something other than CPU power is at work here - what hard drives do the two machines have? What RAM does the MBP have? If the Mini has a fusion drive or SSD that would explain things nicely. Have you put a SSD in your 2011 MBP? If not, you're missing out.

    Check Activity Monitor during the build process - what is maxing out? Most dev software runs multiple jobs together to take advantage of multiple cores, so you should see all 4 virtual cores being hit in the CPU History window but you need to check the memory pressure (not 'free memory' - OSX uses as much as it can get) to make sure you've got enough RAM to cope with multiple jobs. Investigate the settings in your dev software.

    For building, I'd go for something with quad-core, a SSD to keep them all fed and plenty of RAM (16GB) and not worry over-much about small differences in clock speed.

    Chicken entrails, the i-Ching, Tarot readings, the Magic 8-ball are all about as reliable as looking at the product page and trying to deduce the speed. Intel's 'i3/i5/i7' classification is useless marketing fluff and there are umpteen versions of each with different power ratings, graphics etc. and no clear rules for what the difference between (say) i5 and i7 is for a particular model. All the modern CPUs dynamically adjust their speed based on temperature and/or to conserve power, so the cooling efficiency and power management settings of a particular computer all affect performance. Good luck.

    Its well documented that Apple neutered the Mini in 2014 by dropping the options for quad i7 processors. For jobs that can take advantage of multicore (like building large software projects) the incremental improvements in processor speed over a generation or two don't make up for having half the number of cores...
     
  3. fierarul macrumors member

    Joined:
    Jun 18, 2010
    #3
    Geekbench3 is good enough but it only looks at CPU+memory speed and not disk speed.

    Obviously your builds are very IO-heavy which explains the improved build times on the mac mini.

    I believe the MacBook Pros have the fastest flash drives one can purchase nowadays. Get a 13" one.

    Nobody will have a perfect benchmark ready for your specific business use case. If you were to buy a fleet of devices and be cost-effective with it you might have tested each model beforehand. Alas, for a single item you will have to look at ad-hoc info.

    It would be interesting though to have a benchmark for developer machines.
     
  4. T'hain Esh Kelch macrumors 601

    T'hain Esh Kelch

    Joined:
    Aug 5, 2001
    Location:
    Denmark
    #4
    You could have gone for an SSD upgrade in the Mac mini. That would have made it quite a bit faster, especially for compiling or whatever it is you do.
     
  5. Samuelsan2001 macrumors 604

    Joined:
    Oct 24, 2013
    #5
    The 2012 Mac mini had a quad option because the quad and dual core mobile chips Intel made that year all fit the same socket, Intel hasn't done that since and Apple won't put the time money and resources into developing 2 different logic boards for its mini machine.

    Geekbench is fine but it is just an overall look at the speed of the system and specific workloads can have wildly different requirements and new or updated software may take advantage of firmware or new application programming interfaces to leverage more performance in some tasks, or even leverage GPU's in conjunction with the CPU.... All this means that it is pretty tough to tell what is fastest without looking at each application you want to use.
     
  6. kiwipeso1 Suspended

    kiwipeso1

    Joined:
    Sep 17, 2001
    Location:
    Wellington, New Zealand
    #6
    The 15" MBP is a quad core i7, the minis from 2012 and earlier have quad core for the i7 option
    Minis after 2012 are only dual core, so you get half the threads to use compared to older ones.
     
  7. Vexir thread starter macrumors newbie

    Joined:
    Apr 6, 2013
    #7
    Thanks for all the info folks. I'll be returning the 2014 Mini, grabbing a quad core 2012 off Craigslist, and then probably sticking an SSD into it. The Geekbench on a '12 Mini is almost the same as my 2015 15" i7 MBP, so that plus an SSD will probably make build times really fly.

    I'm still not clear on how to systematically figure out whether a system is faster than another system. Is there a benchmarking system that takes HDD into account? Even better would be a benchmarking system that was able to run tests for various use cases.
     
  8. tdhurst macrumors 68040

    tdhurst

    Joined:
    Dec 27, 2003
    Location:
    Phoenix, AZ
    #8
    Every HDD is slow as **** compared to nearly any SSD.
     
  9. maflynn Moderator

    maflynn

    Staff Member

    Joined:
    May 3, 2009
    Location:
    Boston
    #9
    How's the GPU on the 2012 compared to the current models?
     
  10. theluggage macrumors 68040

    Joined:
    Jul 29, 2011
    #10
    2012 quad core Mini sounds sensible - but there's no "probably" about adding a SSD: you want one. The SSD will probably have more effect on your build times than an extra core or two - more cores won't help if the disc drive can't keep up.

    I still wonder what adding a SSD and maybe a RAM upgrade would do to the build times on your 2011 MacBook Pro. Its probably worth punting a couple of hundred bucks to find out - and a 2011 MBP with SSD and plenty of RAM is still a useful bit of kit to have (the RAM/SSD on the MBP were designed to be DIY upgradeable so its a very easy job) .

    ...what, so you have a current-model 15" rMBP? Try running the build on that and see what the time is like, because you won't get much better with another laptop or Mini. I assume you don't want to use this as your build machine, but you can still run a test.
     

Share This Page