Entry-level 2018 15" MBP is faster than top-of-the-line 2017 15" MBP

Discussion in 'MacBook Pro' started by EugW, Jul 13, 2018.

  1. EugW macrumors 68040

    EugW

    Joined:
    Jun 18, 2017
    #1
    The entry-level 2018 15" MBP is faster than top-of-the-line 2017 15" MBP, at least in multithreaded applications.

    https://www.cpu-monkey.com/en/compare_cpu-intel_core_i7_8750h-835-vs-intel_core_i7_7920hq-692

    Cinebench.PNG Passmark.PNG Geekbench3.PNG



    Interestingly, in single-threaded applications, the entry level 2018 is still faster, or at least about the same speed, despite its much, much lower base clock speed. Why? Both models have the same 4.1 GHz single-threaded Turbo Boost speed.

    People were skeptical of some of our predictions here about the CPU speed boosts this year, but it's true, at least based on CPU benches. Sure, the form factor may constrain things a bit, but nonetheless, it's still huge. This is exactly why some of us kept on telling people that the 2018 models will represent one of the biggest upgrades in the last decade, and to wait if at all possible.

    BTW, if you are looking to buy an iMac now, consider holding off for the 2018 models if performance matters to you at all. The new iMac CPUs will be pretty sweet too.
     
  2. Dave245 macrumors 603

    Dave245

    Joined:
    Sep 15, 2013
    #2
    I wonder what the score will be for the 2.6ghz core i7 on the 512GB version of the 2018 15” MacBook Pro.

    Are these speeds now getting close to say an iMac? In order words could the 15” MacBook Pro 2018 replace an iMac?
     
  3. EugW, Jul 13, 2018
    Last edited: Jul 13, 2018

    EugW thread starter macrumors 68040

    EugW

    Joined:
    Jun 18, 2017
    #3
    The entry-level 2.2 GHz 2018 15" MacBook Pro's CPU is quite a bit faster than my 2017 iMac Core i5 7600. However, the difference is my iMac has a lot more ports and it's cheap to upgrade memory. Also, it can work at full tilt for 10 minutes with no audible fan noise. It also has a faster yet quiet GPU (Radeon Pro 575).

    EDIT:

    And of course, the iMac comes with a 27" 5K Retina screen.
     
  4. EugW thread starter macrumors 68040

    EugW

    Joined:
    Jun 18, 2017
    #4
    Geekbench 4

    2017 15" Core i7-7920HQ (top of the line quad-core 15")
    Multi-core ~16800
    Single core ~5000

    2018 13" Core i5-8750H (entry level)
    No scores available yet, but I'm guessing:
    Multi-core ~22500
    Single core ~5000
     
  5. JPack macrumors 68030

    JPack

    Joined:
    Mar 27, 2017
    #5
    Benchmarks like this explain why Apple has indicated they have no plans for Macs powered solely by ARM processors.
     
  6. mmomega macrumors 68030

    mmomega

    Joined:
    Dec 30, 2009
    Location:
    DFW, TX
    #6
    If benchmark racing then the 7700K in the 5K iMac gets roughly 19,000-20,000 on GeekBench 4 MultiCore. And roughly 5,600 on Single Core.
     
  7. EugW thread starter macrumors 68040

    EugW

    Joined:
    Jun 18, 2017
    #7
    Sounds about right. My 5K iMac Core i7-7700K got just over 20000.

    Screen Shot 2018-07-13 at 8.05.10 PM.png

    I'm actually a bit surprised. My original guess was that the i7-8750H hex-core would be slower at this than it is.
     
  8. Earl Urley macrumors regular

    Joined:
    Nov 10, 2014
    #8
  9. EugW thread starter macrumors 68040

    EugW

    Joined:
    Jun 18, 2017
    #9
    Nice, but it should be noted that the 2017 15" already was over 2500 MB/s.
     
  10. vaugha macrumors 6502

    Joined:
    Nov 3, 2011
    #10
    So do we have an answer for why the base 2018 @ 2.2ghz is faster than 2017 @ 3.1ghz in single-core? This is an interesting result and I was not expecting this. People who have tested both models, how is the speed in terms real life applications? Off topic - I am also interested in the battery performance of 2018 models vs 2017 counterparts.
     
  11. electric sheep macrumors newbie

    Joined:
    Feb 26, 2017
    #11
    The difference of 3% is pretty much within margin of error for a small sample size. Any actual improvement may be down to minute refinements in architecture/microcode.
    Why wouldn't they be the same? Core architecture hasn't changed significantly and both CPUs share same 4.1 GHz boost speed. Base clocks are pretty irrelevant.
     
  12. Dave245 macrumors 603

    Dave245

    Joined:
    Sep 15, 2013
    #12
    So specs wise what would you say is the best "bang for the buck?"

    13" with
    • 2.3GHz quad-core 8th‑generation Intel Core i5 processor, Turbo Boost up to 3.8GHz
    • 2.7GHz quad-core 8th‑generation Intel Core i7 processor, Turbo Boost up to 4.5GHz
    OR the 15" with:

    • 2.6GHz 6-core 8th-generation Intel Core i7 processor
    • 2.9GHz 6-core 8th-generation Intel Core i9 processor, Turbo Boost up to 4.8GHz

    The 13" seems so nice and compact it does appeal to me, especially since there hasn't been any 12" MacBook updates, also the rumoured low cost 13" MacBook is unlikely to be getting as good of a display as the new MacBook Pro's (i was surprised that Apple added True Tone). I don't want to keep on waiting so i'm very tempted to get one of these new MacBook Pro's, i could wait until October if need be but that's the longest i'm willing to wait i really don't want to keep waiting year after year.
     
  13. EugW thread starter macrumors 68040

    EugW

    Joined:
    Jun 18, 2017
    #13
    Confirmed. 15” entry level is 5084/22222 in Geekbench 4.

    A3DA1160-DA8B-4D6D-B4FC-6B1BC917CFB8.png

    Remarkable!
     
  14. fokmik macrumors 68020

    Joined:
    Oct 28, 2016
    Location:
    USA
    #14
    I think a macook 12” with arm with macos mojave is coming and thats why they keep it for an event. Ultra portable consumer mac with maybe 20h battery life?
     
  15. leman macrumors G3

    Joined:
    Oct 14, 2008
    #15
    Why wouldn’t it be? They come with the same CPU frequency and the Coffe Lake has more cache.
     
  16. Saad-M macrumors newbie

    Joined:
    Sep 2, 2010
    #16
    That's pretty impressive! As a 2017 15" owner it goes to show that technology can still move very quickly!
     
  17. vaugha macrumors 6502

    Joined:
    Nov 3, 2011
    #17
    Elaborate? How do they have the same frequency?
     
  18. Dave245 macrumors 603

    Dave245

    Joined:
    Sep 15, 2013
    #18
    Is that for the 2.2”ghz or the 2.6ghz”? There are 2 15” the first one with 256gb storage and the 556x graphics card, then there’s the 2.6 version with the 560x graphics cards and 512gb storage which costs more, that’s before adding the i9 to either of them.
     
  19. leman macrumors G3

    Joined:
    Oct 14, 2008
    #19
    You think so? Its quite funny, since your comment shows how we grown used to the stagnation in the CPU market. I remember the times when 5-10% of performance increase between generations would be sneered at as a failure of a chip maker. These days people are happy about it.

    Of course, the Coffee Lake is a large real step in performance for mobile computing, but I'd not say that it represents how technology moves forward. Its a three year old chip, pushed to its boundaries and optimised to be as power-efficient as possible. Intel is pulling every trick in the book to slap two more cores onto the Skylake chip and make sure that the user sees as much performance per watt as possible. Still, real technological progress is in their future CPU gens.
    --- Post Merged, Jul 14, 2018 ---
    Base 2018 CPU is the i7-8750H, which boosts up to 4.1Ghz. The high-end 2017 CPU is the i7-7920HQ, which boosts up to 4.1Ghz. Geebkench is a burst benchmark, running for a very short period of time, on a single thread. Under these conditions, both of the CPUs are likely hit the same clocks. And since its essentially the same core, the performance is the same.

    The base frequency is more of a marketing thing anyway. The real clock changes over time and is subject to complex conditions. Also note that Intel has stopped publishing the turbo clock as depending on number of active cores, since that was often misunderstood.
     
  20. vaugha macrumors 6502

    Joined:
    Nov 3, 2011
    #20
    Thanks for filling in. If what you're telling me true, then geekbench scores are not reflective of real life usage.
     
  21. EugW thread starter macrumors 68040

    EugW

    Joined:
    Jun 18, 2017
    #21
    The link has the specs in it. 2.2 GHz i7-8750H.

    5084/22222

    Actually, they are reflective of some types of real life usage to a certain extent, because real life usage tends to be bursty. Most people don't peg the CPU at 100% all day long.
     
  22. vaugha macrumors 6502

    Joined:
    Nov 3, 2011
    #22
    Again, if what you're telling me also true, then the upgraded processors will have an impact on the battery life compared to the baseline 2.2ghz, no?
     
  23. Naimfan macrumors 601

    Naimfan

    Joined:
    Jan 15, 2003
    #23
    Not surprising - 6 cores vs 4, so they should be at least 50% faster on multi-core.
     
  24. EugW thread starter macrumors 68040

    EugW

    Joined:
    Jun 18, 2017
    #24
    I won't hazard to guess, but sometimes the mid-range processors have longer battery than the low end ones. Why? Because of something called race-to-sleep.

    A faster processor will accomplish a task faster than a slower one, and thus will go into low power idle mode quicker than the slower CPU. However, the power characteristics and battery life will vary from chip to chip, and from activity to activity, so it's hard to say.

    For example, a 2.6 GHz may get race-to-sleep faster than a 2.3 GHz chip, but it's also possible that particular 2.6 GHz chip model will use more power getting there, so it cancels out. And sometimes for high end top-of-the-line chips they use so much more power that they have worse battery life even if they race-to-sleep faster.
     
  25. The Mercurian macrumors 68000

    Joined:
    Mar 17, 2012
    #25
    There is literally no chance of this happening. Making a version of Mojave to work on ARM would be a HUGE undertaking. They would not do that just for a 12" MB, and most likely many apps wouldn't work either.
     

Share This Page