Tested: Thermal conditions in the 2018 i9 MacBook Pro dramatically hampering performance

135678

Comments

  • Reply 41 of 158
    cgWerkscgWerks Posts: 1,514member
    blastdoor said:
    There are two fundamental problems here from my point of view:
    1. Apple should not offer the i9 as an option if it has no real advantage. A big part of the advantage of Apple is that they don’t dump design decisions on the user. 
    There might be advantages for some, though. For example, you could run a couple VMs with 2 cores each and still have 2 cores for apps. You won't get as much maximum speed on each core, but I think it would better distribute.

    anome said:
    ... am I to understand we're not seeing similar issues with the quad-core 13"? Or the hex-core i7?
    Good question! I'm more interested in the 13", but haven't seen any poor reviews on that yet.

    zeudev said:
    Do you have any interest in running another benchmark with just the cpu engaged?  For some of the software developers in your audience, they will only care how well it performs when only the cpu is kicking out heat.
    Yes, good point... or as others have said, with an eGPU.

    caladanian said:
    The iMac Pro does not(!) have thermal issues. :) Other iMacs have (different heat flow).
    Not really thermal issues, but I thought I remember reading that it does throttle in a way... just way up there (slight reduction from max potential of the chips).
  • Reply 42 of 158
    sirozhasirozha Posts: 336member
    I have a solution for Apple. Resurrect the 2015 design of 15" MacBook Pro and bring back the missing ports. The body has more room in it, so use the extra room to dissipate the heat better. 
    jdwcgWerks
  • Reply 43 of 158
    jdwjdw Posts: 619member
    Mike Wuerthele, as an AppleInsider reader, I wish to say that I truly appreciated the testing and work that went into your excellent article.  It reports the facts of the matter and helps would-be MBP buyers be better informed.  Bravo!

    My own personal hope is that the tech media's reporting of this issue can prod Apple positively to THINK DIFFERENT when it comes to the thinness and approaches to minimalism in their notebook line.  Some say that Mac enthusiasts would not like it if Apple decided to use a thicker (and therefore better cooled) notebook design, but I strongly disagree.  My 2015 15" MBP is thicker than the late 2016 and newer models, yet my 2015 is quite thin to my eyes.  The MBP has long been about ports and performance.  The MacBook exists to strip all that away to achieve a thin and light design.  People upset by a thicker (yet so much better) MBP need only consider a MacBook.  And if Apple would kindly refresh the MacBook Air, that could be a second option for Mac portable buyers who prefer thinness and throttling over a thicker and cooler enclosure.  With 3 separate portable lines, surely Apple can come up with something insanely great for those who love the existing notebook line and also for The Rest of Us.
    edited July 18 muthuk_vanalingamgatorguycgWerks
  • Reply 44 of 158
    anomeanome Posts: 1,035member
    jdb8167 said:
    anome said:

    Cheap gags aside, am I to understand we're not seeing similar issues with the quad-core 13"? Or the hex-core i7? Or is it just that everyone bought the i9 because they wanted MOAR POWER!!!, and so we don't have anyone testing the lower tiers?

    I did an informal test with my new 13” MBP with an i7. I have a command line tool that runs multiple copies of the OpenSSL speed test that maxes out all 8 threads. I let it run for 10 minutes until the fans were maxed out and then ran GB4. I saw a minor 1% to 2.5% slow down. It doesn’t look like it is throttling much. I plan to do more testing this weekend. The core i7 13” MBP is a beast. I’m getting 5400 single core and 19100 multi core.
    Cool info. Always good to hear from people doing the experiment.
  • Reply 45 of 158
    Mike WuertheleMike Wuerthele Posts: 3,112administrator
    jdw said:
    Mike Wuerthele, as an AppleInsider reader, I wish to say that I truly appreciated the testing and work that went into your excellent article.  It reports the facts of the matter and helps would-be MBP buyers be better informed.  Bravo!

    My own personal hope is that the tech media's reporting of this issue can prod Apple positively to THINK DIFFERENT when it comes to the thinness and approaches to minimalism in their notebook line.  Some say that Mac enthusiasts would not like it if Apple decided to use a thicker (and therefore better cooled) notebook design, but I strongly disagree.  My 2015 15" MBP is thicker than the late 2016 and newer models, yet my 2015 is quite thin to my eyes.  The MBP has long been about ports and performance.  The MacBook exists to strip all that away to achieve a thin and light design.  People upset by a thicker (yet so much better) MBP need only consider a MacBook.  And if Apple would kindly refresh the MacBook Air, that could be a second option for Mac portable buyers who prefer thinness and throttling over a thicker and cooler enclosure.  With 3 separate portable lines, surely Apple can come up with something insanely great for those who love the existing notebook line and also for The Rest of Us.
    This was very much a team effort. Five staffers worked on this piece to some extent over the course of the day.

    Appreciate the kind words.
    edited July 18 jdwKITAelectrosoft
  • Reply 46 of 158
    orthorimorthorim Posts: 141member
    Yeah really good info here guys. 

    i am wondering about the 2.6... as the i9 is barely faster than the 2.2 the 2.6 might be the fastest option now. 

    6 cores will probably still make enough of a difference for me to upgrade from my mid 2012 model - aka the legend. 
  • Reply 47 of 158
    Here's how you should do this kind of testing. First get the Intel HW tool which will show you the power used by the CPU/GPU and its frequency. Second, get a GPU benchmark (Luxmark, Unigyne, etc.).  

    Run the GPU test and measure how much power the GPU is reporting. 

    Start a temperature recording utility. 

    Progressively max out each of the cores using the terminal command
    yes > /dev/null &
    Since each core has two threads, you have to enter as many of those commands as you have threads. After each command, you should see a jump in processor load and a temperature increase. 
    After you are at 100% CPU usage, let the temperature stabilize. At this point, the fans will have started to spin at high speed. Read the CPU power.

    Now start the GPU benchmark again. Monitor Temperature and power used by both GPU and CPU, as well as CPU frequency. Total power consumption should go up by the same amount you measured earlier in the GPU test. 

    De-activate each core/thread, one at a time. You should see the temperature come down, etc. This whole process will take ~15 minutes or so. 

    Let us know the readings!!!
  • Reply 48 of 158
    KITAKITA Posts: 125member

    The benchmarks have come under fire from critics, with them saying that there are "breaks" to smooth out the performance. However, that does not appear to be the case.

    While there are pauses in the test, they are done at the end of each test, not in the middle. So, individual tests are performed under full load, with no cool-down period during the test to keep processor speeds low.
    Geekbench 4 uses 2 second pauses are after each workload. Each workload only lasts for, in most cases, ~1 second and there are 50 workloads in total (25 for single core / 25 for multi core).

    Here's an example in Geekbench 4 multicore:

    MacBook Pro 15 (i9-8950HK) - 22,522
    Gigabyte Aero 15X (i7-8750H) -  19,073

    And here's the sustained performance in Cinebench R15 CPU for the Aero 15X:


    That's an average of ~1,100 for the Aero 15X.

    Meanwhile, the MacBook Pro, as per your testing, both the i7 and i9 only scored ~900 sustained.

    Clearly the Geekbench 4 results are completely misleading as even the best case scenario for the i9 MacBook Pro fell short of the i7 Aero 15X's worst run.
    Apple is hardly the only vendor dealing with i9 thermal conditions, and like we said, Premiere performs far better with Nvidia GPU silicon than AMD Radeon gear which explains most of the Dell ripping through the test. 
    Dave Lee used the same Gigabyte Aero 15X (i7 + GTX 1070 MaxQ) discussed above, not the Dell XPS 15 (i9 + GTX 1050Ti MaxQ). Although, I imagine the Dell might have also faired quite well.
    The most obvious solution right now is for Apple alter the peak speed of the processor by adjusting the power that the chip gets. Ironically, slowing the peak speed of the processor may allow it to finish tasks quicker, as it will slow down less to keep the CPU cool.
    On that topic, someone has undervolted a Core i9 Dell XPS 15 with some excellent results. Clearly the Core i9 is not an easy CPU for 4~4.5 lbs laptops to handle, but it might be possible to get some more performance out of it after all.

    Great work on the article! Lots of key points covered. Looking forward to more!



    muthuk_vanalingam
  • Reply 49 of 158
    I will have to do liquid metal cooling again.
  • Reply 50 of 158
    DuhSesameDuhSesame Posts: 281member
    sirozha said:
    I have a solution for Apple. Resurrect the 2015 design of 15" MacBook Pro and bring back the missing ports. The body has more room in it, so use the extra room to dissipate the heat better. 
    Wanna know how bad the retina performs when equipped with 4980HQ?
  • Reply 51 of 158
    doggonedoggone Posts: 161member
    Running a MacBook Pro (15-inch, 2016) 2.6 GHz Intel Core i7 at full speed initially went up to 100C at 3GHz.  Then the temp settled down to 80-85C.

    My 
    suspicion is that the current OS is not fully optimized for the new machines.  Software is an important factor in power utilization and Apple have a history of optimizing the OS to improve performance.
  • Reply 52 of 158
    DuhSesameDuhSesame Posts: 281member
    KITA said:

    The benchmarks have come under fire from critics, with them saying that there are "breaks" to smooth out the performance. However, that does not appear to be the case.

    While there are pauses in the test, they are done at the end of each test, not in the middle. So, individual tests are performed under full load, with no cool-down period during the test to keep processor speeds low.
    Geekbench 4 uses 2 second pauses are after each workload. Each workload only lasts for, in most cases, ~1 second and there are 50 workloads in total (25 for single core / 25 for multi core).

    Here's an example in Geekbench 4 multicore:

    MacBook Pro 15 (i9-8950HK) - 22,522
    Gigabyte Aero 15X (i7-8750H) -  19,073

    And here's the sustained performance in Cinebench R15 CPU for the Aero 15X:


    That's an average of ~1,100 for the Aero 15X.

    Meanwhile, the MacBook Pro, as per your testing, both the i7 and i9 only scored ~900 sustained.

    Clearly the Geekbench 4 results are completely misleading as even the best case scenario for the i9 MacBook Pro fell short of the i7 Aero 15X's worst run.
    Apple is hardly the only vendor dealing with i9 thermal conditions, and like we said, Premiere performs far better with Nvidia GPU silicon than AMD Radeon gear which explains most of the Dell ripping through the test. 
    Dave Lee used the same Gigabyte Aero 15X (i7 + GTX 1070 MaxQ) discussed above, not the Dell XPS 15 (i9 + GTX 1050Ti MaxQ). Although, I imagine the Dell might have also faired quite well.
    The most obvious solution right now is for Apple alter the peak speed of the processor by adjusting the power that the chip gets. Ironically, slowing the peak speed of the processor may allow it to finish tasks quicker, as it will slow down less to keep the CPU cool.
    On that topic, someone has undervolted a Core i9 Dell XPS 15 with some excellent results. Clearly the Core i9 is not an easy CPU for 4~4.5 lbs laptops to handle, but it might be possible to get some more performance out of it after all.

    Great work on the article! Lots of key points covered. Looking forward to more!



    MacBook Pro in 2018 does not change anything in thermal design, though.  I was wondering if adding extra airflow vent and one more heatpipe will increases the thermal capacity,  but who knows for exact.

    But looking from what Aero 15x can offer, the Core i9 will still be horrible, even undervolting and such might work much better, I'm not going to be happy until it can achieve maximum turbo boost.
    edited July 19
  • Reply 53 of 158
    chasmchasm Posts: 702member
    So it sounds like:
    1. Intel needs to admit they have a TDP problem.
    2. Apple could issue a firmware update that would instruct the fans better for the i9 model (no reason why they wouldn't, but expect this to take a while)
    3. Heavy graphics users should (surprise!) look at getting an external GPU if they're going to be pushing the machine hard, or install a manual-fan-control (not recommended) utility and do some testing to see if that eases the issue.

    I'll be interested to see what the eGPU testing from AI reveals. Based on what we know so far, I'd say Apple bears some responsibility for this, but Intel may bear a larger percentage than Dave Lee would suggest if these reports of dramatically-higher TDP turn out to be true.
    cgWerks
  • Reply 54 of 158
    People surprised about this should get a desktop, like a Mac Pro. This is nothing new. The thinner the body , the more likely the computer will throttle,to maintain temperatures. I suspect there would be even more complaints , if fans on MacBooks made more sound.
  • Reply 55 of 158
    For those anoyed with heating issue, just disable Turbo Boost. You can do it using another app like Turbo Boost Switcher or so. On redding there are even some posts reporting better performance with Turbo Boost disabled on certain tasks due to this issue.
  • Reply 56 of 158
    So can I take it that Appleinsider will now retract the story posted in response to Lee's video? (https://appleinsider.com/articles/18/07/17/test-suggests-2018-macbook-pro-cant-keep-up-with-intel-core-i9-chips-thermal-demands)

    "While thermal throttling is nothing new, especially in portables, Lee's findings are somewhat questionable in that assumptions are being made based on a single machine's performance with an unoptimized app. Making blanket statements without thorough testing is reckless at best and disingenuous at worst. 

    While Lee failed to reach out to Apple for comment, it is highly unlikely that the company would ship a flagship product without first rigorously testing its performance. That goes double for a device like MacBook Pro, considering the company's renewed vigor to serve the professional market."

    This was a pretty poor crticism from the author of that article. Appleinsider seems to now have tested along similar lines and drawn the same conclusion.....!

    edited July 19 avon b7muthuk_vanalingambala1234electrosoftprismaticscgWerks
  • Reply 57 of 158
    If the system cannot perform at the rated clock rate at all workloads, it is back to the drawing board.  
    You cannot expect the system to reach burst speeds for more than shorter periods, but I guess that is implicit in the the term burst. 

    If a system consistently underperforms (compared to stated ratings), then the manufacturer misguides the customer, who might have chosen a better performing system that meets his expectations and requirements. 
    edited July 19 cgWerks
  • Reply 58 of 158
    19831983 Posts: 1,059member
    Well this was a major oversite by Apple! They really should of noticed this problem in development. And why does Apple continue to not support NVIDIA GPUs like it does AMD ones? The former from what I understand is the better more well supported and popular design. Again if they’re serious about pro clients, both designs especially the former should be supported equally by Apple. Adobe the undisputed king of professional creative software (a monopoly too unfortunately) optimizes their products for NVIDIA not AMD apparently. As I assume do other high-end software brands for CAD and such.
    edited July 19
  • Reply 59 of 158
    asciiascii Posts: 5,778member
    This both confirms and expands upon Dave Lee's analysis. It confirms that the i9 throttles below its base clock, but adds that the 6-core i7 stays above its base clock, which is good news for Apple.

    So the 2018 MBP performs as advertised if you stick with the stock CPUs. The 2016 chassis can cool a 6-core processor provided its clocked nice and low, in the 2.3 - 2.6GHz range.

    This is all important info for potential buyers to know, so good job.
    edited July 19 cgWerks
  • Reply 60 of 158
    ascii said:

    So the 2018 MBP performs as advertised if you stick with the stock CPUs. The 2016 chassis can cool a 6-core processor provided its clocked nice and low, in the 2.2 - 2.4GHz range.

    So paying for anything above the 2.2 Ghz base config in terms of processor speed is wasted unless you really want to pay a premium for (the potential of) bursts. 
    edited July 19
Sign In or Register to comment.