Apple's new Retina MacBook Pros bring faster benchmarks with slower clock speeds

Posted:
in Current Mac Hardware edited January 2014
The team behind popular benchmarking suite Geekbench has rounded up the scores for Apple's newest Retina MacBook Pro models and finds that despite slower processor clock speeds across the board, the Haswell-powered laptops outperform their predecessors by as much as 8 percent.

Latest 15-inch Retina MacBook Pro Benchmark
Apple's newest high-end 15-inch Retina MacBook Pro shows significant performance gains over its predecessor


Single-core performance for the 13-inch variants is up only two to four percent over the previous generation, which John Poole, founder of Geekbench parent Primate Labs attributes to the new revision's focus on power efficiency rather than speed gains.

The 15-inch variant sees much larger leaps in performance in its high-end configuration -- the newest 2.6-gigahertz flagship jumps eight percent in both the single and multi-core benchmarks over its 2.8 gigahertz predecessor, for example. Mid and low-end variants gain a more pedestrian two to four percent.

Interestingly, the latest high-end 13-inch MacBook Air trails the new 13-inch MacBook Pro by only five percent in the single-core benchmark, though the gap widens to thirteen percent in the multi-core benchmark. "Users with applications that only use one core won't notice much difference between the Air and the Pro," Poole writes. That does not extend to the laptops' battery life, however, as the MacBook Air still bests the new MacBook Pro by three hours in that category.

Poole calls the performance increases from the more conservatively-clocked Haswell processors "interesting," but notes that the processors are not always running at the advertised speeds. Thanks to Intel's Turbo Boost feature, the clock speeds can be increased as needed to handle more demanding workloads -- the 2.3-gigahertz Haswell chip in the newest MacBook Pro can boost up to 3.5 gigahertz, for example.

The Geekbench scores, which began trickling in on Wednesday, are based on processor and memory performance. GPU performance is not included, and as a result we have yet to see what effect the inclusion of Intel's Iris Pro graphics chip has on the Retina MacBook Pro's capabilities.
«1

Comments

  • Reply 1 of 30

    I would like to know why rMBP only has an additional hour of battery life when the CPU is turned down and Mavericks is suppose to bring big battery savings, and the entry 15" has integrated graphics. 

     

    What else changed that all these battery saving technologies only gained an hour, when the MBA gained so much more, and then an additional hour with Mavericks. Somethings seems off. 

  • Reply 2 of 30
    [quote]Thanks to Intel's Turbo Boost feature, the clock speeds can be increased as needed to handle more demanding workloads %u2014 the 2.3-gigahertz Haswell chip in the newest MacBook Pro can boost up to 3.5 gigahertz, for example. [/quote]
    And how is this different than Sammy's "benchmark booster" for which we gleefully derided the company?
  • Reply 3 of 30
    And how is this different than Sammy's "benchmark booster" for which we gleefully derided the company?

    The Intel Turbo feature is available to every application, not just benchmarks.
  • Reply 4 of 30
    Quote:

    Originally Posted by TeaEarleGreyHot View Post





    And how is this different than Sammy's "benchmark booster" for which we gleefully derided the company?

     

    OK, Jean Luc, let me explain...

     

    Samsung's "benchmark booster" feature specified certain applications under which it would "increase performance output", with those applications specifically

    named being "benchmarking apps". The rest of the time, it does little "when needed", as those performance boosts significantly degrade battery life to the point of seriously degrading product desirability. 

     

    Intel's "burst mode" appears at any time it's needed or requested, and is application-name agnostic. Because it operates efficiently, unless you're doing things like grinding heavy HD video edits in Final Cut while in the field, you won't see substantially degraded battery life, even with that kicking in from time to time as it's designed to.

     

    In other words, one is marketing hype (and only really used when needed to make the company/product look better) and the other is an actual performance/efficiency enhancement. 

     

    Get the difference? 

  • Reply 5 of 30
    And how is this different than Sammy's "benchmark booster" for which we gleefully derided the company? I will assume you are asking a serious question.

    Because the Turbo Boost feature is available to all software and therefore is a reasonable tool to use for benchmarking.

    Sammy's boosted their speed beyond that available to any apps except for the benchmarks. Therefore - they are lying.
  • Reply 6 of 30
    froodfrood Posts: 771member
    Quote:
    Originally Posted by TeaEarleGreyHot View Post





    And how is this different than Sammy's "benchmark booster" for which we gleefully derided the company?

     

    If you did a benchmark score you would get a score representative of how it would typically run (it turns up the juice as needed).  In the case of Samsung they specifically looked to see if the app name matched common benchmarks and if so automatically turned them into full power mode.

     

    Either way the scores are still representative of what the devices can do at full power.  In real usage they would burn up pretty quickly in that mode.

     

    So it is a little in how you look at them.  When cars are tested for 'maximum horsepower' on a dyno they run them to a state they would probably blow up in if left there for more than a few minutes.  It is the accepted practice.  When testing phones, what are the norms?

     

    If Apple's estimated battery life is given with Wi-Fi off and the screen brightness dimmed, is that cheating too?  Most people would say no, even though that's not how they normally use them- it is just perceived that that is an acceptable way to run that test in order to get a result that puts your best foot forward.

     

    So make up your own mind on who is guilty or not :p  IMO Samsung's move was dubious at best.

  • Reply 7 of 30
    wizard69wizard69 Posts: 13,377member
    The problem I have with this article is that they leave out an important feature of the 13" rMBP, that is the Intel GPU which is significantly better than the one in the AIRs

    In any event the benchmarks don't mean much if the processor core isn't locked to one clock rate. There is no telling what the actual clock rates are here.
  • Reply 8 of 30
    Quote:



    Originally Posted by TeaEarleGreyHot View Post





    And how is this different than Sammy's "benchmark booster" for which we gleefully derided the company?

    Horrible trolling is horrible.

  • Reply 9 of 30
    Samsung sees your Geekbench benchmark and raises you 20%...
  • Reply 10 of 30
    8 percent. Can you imagine what the reaction would have been if the A7 was only 8% faster than the A6 processor? Instead it was about twice as fast and is 64 bit. Mobile processors are about to crash into the performance curve of Intel's desktop processors over the next year or two. It should be fun to watch.
  • Reply 11 of 30
    zoolookzoolook Posts: 657member
    Quote:

    Originally Posted by Richard Getz View Post

     

    I would like to know why rMBP only has an additional hour of battery life when the CPU is turned down and Mavericks is suppose to bring big battery savings, and the entry 15" has integrated graphics. 

     

    What else changed that all these battery saving technologies only gained an hour, when the MBA gained so much more, and then an additional hour with Mavericks. Somethings seems off. 




    Because of the retina display.

  • Reply 12 of 30
    Within the same power envelope and price category Haswell offers a 2.3GHz CPU with Iris Pro 5200 graphics and a 2.7GHz CPU with HD4600 graphics. The 2.7GHz Haswell would have turned in 17% better CPU scores in Geekbench and cost Apple roughly $100 less.

    So clearly there's something else going on here and I think the Mac Pro's twin GPU is the key. I think Apple is trying to move toward a world where two GPUs are used at the same time: one for computation and one to drive the display. In such a world a 2.3GHz CPU equipped with a 5200 GPU is better than a 2.7GHz CPU equipped with a 4600 GPU even if you've got an nVidia 750M available.

    Their secondary motivation is probably to get a larger share of the customer base to accept integrated graphics as "good enough". That's sad because Apple has always criticized others for shipping products that are merely "good enough".
  • Reply 13 of 30
    Quote:
    Originally Posted by Bregalad View Post



    Within the same power envelope and price category Haswell offers a 2.3GHz CPU with Iris Pro 5200 graphics and a 2.7GHz CPU with HD4600 graphics. The 2.7GHz Haswell would have turned in 17% better CPU scores in Geekbench and cost Apple roughly $100 less.




    Quote:

    Their secondary motivation is probably to get a larger share of the customer base to accept integrated graphics as "good enough". That's sad because Apple has always criticized others for shipping products that are merely "good enough".


    You're contradicting yourself. If Apple's "motivation" was to have people think that integrated graphics are "good enough" then they would not have gone with the Iris Pro over the HD4600 since for some it is "good enough". But since that is not their motivation they went with Iris Pro which is 2-3x faster in most compute tasks over the HD4600 because to their standard it was not "good enough".




     





     

    The

  • Reply 14 of 30
    sockrolidsockrolid Posts: 2,789member

    if ([self.benchmarkName hasPrefix:@"geekBench"])

       self.turboBoostON = YES;

  • Reply 15 of 30
    Quote:

    Originally Posted by Zoolook View Post

     



    Because of the retina display.


     

    The old retina MBP did not have retina? 

  • Reply 16 of 30

    LOL... just not the case poor guy.

  • Reply 17 of 30
    john.bjohn.b Posts: 2,740member

    The PCIe-based flash storage is noticeably faster at I/O based operations compared to the last generation SSD drives (let alone spinning discs).  Booting, starting applications, loading/saving files, etc.  Geekbench is great at processor-based metrics, but IMO doesn't adequately take I/O speed differences into account.

  • Reply 18 of 30
    akqiesakqies Posts: 768member
    The old retina MBP did not have retina? 

    Sure, but it's a pretty significant part of the power needs of the device. Haswell only increases power efficiency in the CPU and chipset.
  • Reply 19 of 30
    ratsrats Posts: 19member
    Quote:

    Originally Posted by Richard Getz View Post

     

    I would like to know why rMBP only has an additional hour of battery life when the CPU is turned down and Mavericks is suppose to bring big battery savings, and the entry 15" has integrated graphics. 

     

    What else changed that all these battery saving technologies only gained an hour, when the MBA gained so much more, and then an additional hour with Mavericks. Somethings seems off. 


     

    I believe they actually shrunk the battery capacity. This allowed them to make the Macbook Pro thinner and lighter with a modest gain in battery life.  I think the trade off was good. 

  • Reply 20 of 30
    Quote:
    Originally Posted by akqies View Post





    Sure, but it's a pretty significant part of the power needs of the device. Haswell only increases power efficiency in the CPU and chipset.

     

    The screens did not change for either line. Stills seems disproportionate: 

     

    Air: 7-12 hours is 70% improvement? 

    Pro: 7-9 hours is 28% improvement? 

Sign In or Register to comment.