Any noticeable performance difference between MBP 13 TB & nTB?

Discussion in 'MacBook Pro' started by johngwheeler, Nov 27, 2016.

  1. johngwheeler macrumors 6502

    Joined:
    Dec 30, 2010
    Location:
    I come from a land down-under...
    #1
    I'm wondering whether there is much real-world difference for the majority of tasks, between the 2.0-3.1GHz CPU in the nTB and 2.9-3.3GHz processor in the TB model.

    The majority of my usage would be web-browsing, MS Office document production (Word, PPT, Excel), video playback (HD only), light photo & video editing, and occasional Java coding. I would also sometimes run VMWare Fusion VMs with fairly heavyweight servers such as Oracle database.

    Given both CPUs turbo boost to similar speeds 3.1 vs 3.3GHz, and that I would be unlikely to run tasks that demand long periods of maximum CPU load, would performance differences basically amount to the difference in turbo speed?

    I appreciate that very few of you will have access to both machines concurrently, but I would welcome your thoughts on the matter!

    Thanks,

    John
     
  2. Antairez, Nov 27, 2016
    Last edited: Nov 28, 2016

    Antairez macrumors regular

    Antairez

    Joined:
    Mar 17, 2015
    #2
    The non touch bar version is a MacBook Air replacement unit, which has only 1 fan. Even with upgraded i7 CPU, you are still stuck with 15W TDP and Iris Graphics 540, that's 300 bucks for 300 Mhz increase in clock speed. I would say just get the base model if you are going with the non touch bar unit, with its 15W TDP its going to throttle, I doubt you can utilize much of the higher clock speed all the time.

    --- updated ---
    Somehow I thought the topic was about upgrading the non TB version to i7 processor, maybe I hit my head hard or something. Anyway, to answer your question, there won't be a huge difference from day to day task between the 2, you will see about maybe 10~15% difference on benchmark (just a rough guess, don't quote on me), which is really not noticeable since day to day jobs don't exhaust your CPU that often. As for video editing, if you don't edit videos everyday then its not going to save you a lot of time either. I bought the TB version simply because I want the touch bar, and I think that's the biggest difference.
     
  3. johngwheeler thread starter macrumors 6502

    Joined:
    Dec 30, 2010
    Location:
    I come from a land down-under...
    #3
    Your statement (in bold) is exactly what I would like to see evidence of! I understand that the 28W TDP CPU in the TB model can get warmer before it throttles, but the question is how long it takes to do this. I found an interesting early review here:

    http://arstechnica.com/video/2016/11/the-2016-13-and-15-inch-touch-bar-macbook-pros-reviewed/3/

    "CPU performance increases by an amount that’s roughly proportionate to the 200MHz maximum clock speed increase, and GPU performance increases by around 10 percent. It’s not nothing... but it’s also not a huge reason not to get the $1,499 MacBook Pro if you don’t care about the other stuff that it’s missing.

    Theoretically, the 28W processor may be less prone to throttling under load than the 15W processor, but in practice that will depend heavily on how you’re using the system. If you’re only stress testing the CPU and not the GPU, both systems are able to maintain their peak clock speeds pretty much indefinitely—both laptops maintained their peak dual-core Turbo Boost speeds of 2.9GHz and 3.1GHz under 100 percent load for 30 minutes. It was only when I also fired up a GPU benchmark that I saw signs of the 15W model throttling more heavily than the 28W model, and even then the differences were small."

    I'm interested to know whether the behaviour stated in the review, i.e. CPU-only loads can maintain max turbo speeds for 30 minutes, would be applicable to my use cases. I would expect most of my heavy load use, such as running Java applications and DB servers would not use the GPU at all.
     
  4. FrozenDarkness macrumors 65816

    Joined:
    Mar 21, 2009
    #4
    Day to day you wont see a difference. Rendering a video or something that requires 100% CPU load you'll see a diff. Many reviews say that editing a 4k video is harder and choppier
     
  5. iizmoo macrumors 6502

    Joined:
    Jan 8, 2014
    #5
    Did a little bit of 4K video editing in FCPX, it was fine, though I suspect generating optimized and temporary video files used during the editing process might be slightly slower by a few %, but not that much. The only big difference for anything is if you go to the 15" with the i7 beastie and a 4GB vRAM dGPU. Otherwise at all 13" level, a theoretical 10% difference that even exists in maybe <1% of the time you're using the machine isn't very much, and isn't worth the money.

    Ran 6 hours long node.js scripting grabbing 250K web pages, core went up to around 2.8Ghz and stayed there the whole time. Fan spun, and the case got warm, but otherwise seemed to be fine. That's a task that was running 80% of 1 core, so it wasn't maxing out the CPU, just clock rate. The only scenario I can think of that would be problematic would be doing something like a multi-hours video encoding where the CPU hit near 350+%. Video playback with something like Youtube was running 4% (of 1 core). H.265 1080p at around 50%.
     
  6. xraydoc macrumors 604

    xraydoc

    Joined:
    Oct 9, 2005
    Location:
    192.168.1.1
    #6
    This is why I bought the 2.0GHz model. I won't be editing 4K video on the go (or likely at all, to be honest, but I have an iMac at home if I ever do).

    I do occasionally use an app called Osirix that benefits from a fast CPU and GPU, but I use the CPU/GPU intensive features infrequently enough in a mobile situation that I didn't feel the need to spend $300+ for better performance.
     
  7. David58117 macrumors 65816

    Joined:
    Jan 24, 2013
    #7
    I think the throttling is only if you're using the iGPU. I haven't noticed any with mine (base model non-tb), and I've ran Logic Pro x/ubuntu VM.

    I imagine anything that won't run on the non-tb, will likely run poor on the tb-version anyway.
     
  8. mikeo007 macrumors 65816

    Joined:
    Mar 18, 2010
    #8
    You'll see a significant performance increase doing pretty much anything that utilizes both the CPU and GPU at the same time.
    You'll only see a slight performance increase (~10%) if you're just maxing out the CPU. Only due to the higher turbo frequency.

    Basically, a sustained workload that utilizes both CPU and GPU will cause the 15W CPU to throttle fairly quickly. Both the 15W and 28W Skylake chips have have a good enough thermal design to sustain their turbo frequencies almost indefinitely, but the 15W is forced to throttle as soon as there is GPU activity.
     
  9. callea macrumors regular

    callea

    Joined:
    Jul 26, 2011
    Location:
    Italy
    #9
    A 15 watt CPU will generate less heat, so why it should throttle more than a 28watt CPU ??
     
  10. mikeo007 macrumors 65816

    Joined:
    Mar 18, 2010
    #10
    It generates less heat by necessity. It can generate more, which is what causes it to throttle. 15W is what the CPU is designed to use at it's base frequency of 2.0GHz. It increases the wattage to achieve its turbo frequency. This increases heat and if sustained will eventually cause the system to throttle the CPU.
     
  11. iizmoo macrumors 6502

    Joined:
    Jan 8, 2014
    #11
    More is relative. TDP is a specific way of measuring how much heat is generated based or what is considered a normalized usage pattern, which is not an average or at 50%, there's an entire standard on how CPU TDP get defined, it's by IEEE if I remember right. But basically, any CPU at max clock or running at 100% in sustained mode will generate higher heat than the listed TDP because that's not the "normalize" pattern per the standard.

    Oh, and other folks are wrong that CPU get throttled at 15 or 28watts. They don't. CPU have a defined heat point where throttling occurs to ensure it doesn't heat up so much that it burn out the CPU or the surrounding components. TDP specify how much cooling should be provided to dissipate heat in such a way that the CPU is able to operate at a specify point (defined by the TDP measurements). In term of any computer, the cooling provided can be more than the listed TDP and give the CPU quite a bit of room to heat up.

    People are arguing based on the assumption that both CPU are provided exactly 15 and 28watts of TDP cooling. Quite frankly, we don't know what Apple's internal designed specification are as far as TDP cooling provided, so mileage will varies on the throttling. It's entirely possible that the 15w can have quite a lot of breathing room above 15 than the 28w does. The 28w just have the advantage of having a natively higher TDP requirement, thus it normalized operation curve is at a much higher sustained rate 2.9Ghz vs 2.0.
     
  12. macintoshmac macrumors 68000

    macintoshmac

    Joined:
    May 13, 2010
    #12
    Hey Mike, does this mean we won't be able to game on this machine? (at any playable setting, of course, not saying maxed out or ultra-low)
     
  13. mikeo007 macrumors 65816

    Joined:
    Mar 18, 2010
    #13
    You'll be able to play basic games, but I wouldn't expect high settings even with the 550. The GPU won't stop working altogether, it will just throttle back the clock speed until it reaches a reasonable temperature. The user would experience it as inconsistent framerates. Alternatively, you could reduce graphics settings and cap the framerate to prevent the SOC from ever having to throttle.

    Here's an extreme example of how the throttling would look in a game. It starts around the 4 minute mark. Keep in mind this is a video of a ~4w TDP Cherry Trail SOC that is plagued with throttling issues, and the 15w Skylake won't be this bad, and probably won't throttle this quickly unless you're really pushing it.


    Overall, the 550 is only going to give you somewhere between 10-15% better gaming performance from the get-go. But the gap will widen when the 540 is forced to throttle while the 550 can maintain it's clock speed.
     
  14. macintoshmac macrumors 68000

    macintoshmac

    Joined:
    May 13, 2010
    #14
    In India, going the 550 route will mean an additional $388 for touch bar as well as 550. I think that is not needed at all, especially when that model will still have 8GB RAM as well as 256GB SSD only. In India, I think, the best bang for buck is the 13" non touch bar.
     
  15. linguist macrumors member

    Joined:
    Mar 31, 2009
    #15
    lower end model of anything is almost always better bang for buck . . .

    I say it depends on the use, if the machine doesn't fit the usage requirement, then best bang will be a waste and underperform.
     
  16. macintoshmac macrumors 68000

    macintoshmac

    Joined:
    May 13, 2010
    #16
    Absolutely true that. However, today's lowest rung MBP 13 is good enough for my needs at the moment. I am coming from a Late 2011 MBP 15 with 16GB 1600MHz and 256GB SSD and 1GB AMD Radeon 6770m. Apart from RAM, I think everything else is an upgrade for me.
     
  17. johngwheeler thread starter macrumors 6502

    Joined:
    Dec 30, 2010
    Location:
    I come from a land down-under...
    #17
    Thanks for the answers. It would seem that for a CPU-only load, that the only noticeable difference between the base MBP 13 and the TB model is that the slightly faster CPU turbo speed of the latter would give maybe 10-15% speed improvement in some tasks.

    Considering I would be using this as a "MacBook Air replacement" (I also have a quad-core rMBP 15 and desktop workstations), I think the base MBP 13 will be easily good enough. I'm not convinced that I would use the touch bar (I have tried it in an Apple store), and the higher spec CPU upgrade alone doesn't seem to be worth the extra cost to me (AU$500).

    John
     

Share This Page