fokmik
Suspended
thanks, at least it maintained the clock speed ...so under heavy load for longer period of time, will get better result than i9 and even than 2.6ghz
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
2018 i7 MacBook Pro throttling discussion.
- Thread starter iamMacPerson
- Start date
- Sort by reaction score
😱 Wow I'm averaging, like, 780. Highest score was 885. Really goes to show how much different the thermal throttling is making. I wonder how much of that is due to the 560X running hotter and taking up more of the thermal envelope.
Wow... I had a low score of 744, but it turned out I had a background task running that was almost maxing out one of the cores.
Even after running many tests in a row I'm still in the mid 800's. What do you get for Cinebench OpenGL with the 560X?
Edit: Wow I saw in the i9 throttling topic someone with an i9 that got a 865 on Cinebench... Maybe the low end i7 really is the way to go?... I was just being cheap. 😀
Edit2: And then I look up some user's Cinebench scores on the XPS 15 (9570) and see over 1250 with the same i7 processor that I have.
Last edited:
M.Rizk
macrumors 6502a
How long did you leave the test running? I think you need to run it a little longer, maybe 5 minutes.
rainking
macrumors 6502
M.Rizk
macrumors 6502a
I am a noob when it comes to such topics but I see others doing it. Maybe they keep restarting the test a few times as soon as it’s done. I did notice most are using prime95 though as they can keep it running indefinitely till they stop it themselves.
I think it needs to be restarted manually. It def need to be run for longer, the CPU was going towards potentially throttling but the test ended before it got there. Prime95 is indeed a tougher test, but also a bit unrealistic for most users. Indeed, it would be more meaningful to test with one's own actual workload rather than synthetic tests. D2D's test is pretty good in that regard.
I9 user here. Since we are using prime95 I've been using that as a base to compare with you guys.
Prime 95 Testing
Prime95 Test 1 and 3 will give me the most stable and flattest line.
Prime95 Now Test 2 and Test 4....goes ALL over the place. Give it a try and see if it happens on yours.
Cinebench
Test 1 : 859
Test 2 : 808 (Fans Kick in)
Test 3 : 824
Test 4 : 767
Test 5 : 758
Test 6 : 839
Test 7 : 744
Test 8 : 683(starts to spike all over the place)
Test 9 : 681
Test 10 :745
[doublepost=1531961642][/doublepost]curious to see if the i7's exhibit the same behaviour, esp the base 2.2
Prime 95 Testing
Prime95 Test 1 and 3 will give me the most stable and flattest line.
Prime95 Now Test 2 and Test 4....goes ALL over the place. Give it a try and see if it happens on yours.
Cinebench
Test 1 : 859
Test 2 : 808 (Fans Kick in)
Test 3 : 824
Test 4 : 767
Test 5 : 758
Test 6 : 839
Test 7 : 744
Test 8 : 683(starts to spike all over the place)
Test 9 : 681
Test 10 :745
[doublepost=1531961642][/doublepost]curious to see if the i7's exhibit the same behaviour, esp the base 2.2
Last edited:
My Cinebench results on my base 2.2 are below. My frequencies fluctuate a lot after the first 9 seconds or so, but I never see it dip below about 1.8ghz and it is typically maintaining at least the 2.2 clock speed with many spikes above it. These tests were run immediately one after another:
Test 1: 877
Test 2: 819(fans also kicked in)
Test 3: 812
Test 4: 801
Test 5: 810
Test 6: 788
Test 7: 797
Test 8: 802
Test 9: 804
Test 10: 839
I have a feeling this has something to do with the almighty software optimization. The temps aren't really that crazy, especially at the start of the test, but the freqs are still jumping around. Anyway, good to see the lower end can hang with its base clock at least.
Considering an upgrade to the 2018 MBP, and I would be interested in knowing how well the processors do if you manually ramp up the fan speeds using "Macs Fan Control" (app is free). Anyone with a 2018 MBP willing to give it a try and report how it goes? https://www.crystalidea.com/macs-fan-control
try smcFanControl https://www.eidac.de
And thanks for testing with the 2018 i7
in comparison here are my 2017 2.9Ghz i7 (re-pasted w. Kryonaut) at 26c ambient at stock fan curve
In addition to that fan control software, there is this that Lisa from MobileTechReview used to examine this behavior: https://www.crystalidea.com/macs-fan-control
Setting the fans to spin up sooner and faster for certain temperatures does seem to make a significant difference. Like some manufacturers, I guess Apple made the choice that people would want a bit more quiet over a bit more sustained performance. Since I have tried to get my computers quieter and quieter over the last half decade and now have two silent systems . . . they may have made the right trade-off. And, at least they did not lock the end user out of the option of tweaking the fan settings, even if you have to use a third-party utility to do it.
I did some more testing and personally set the fans to the read the sensor off the "CPU PECI" sensor and the fans to kick in @ 50 deg C.
Cinebench( fans @ auto) Peak temp @ 95 Deg C
Test 1 : 859
Test 2 : 808 (Fans Kick in)
Test 3 : 824
Test 4 : 767
Test 5 : 758
Test 6 : 839
Test 7 : 744
Test 8 : 683(starts to spike all over the place)
Test 9 : 681
Test 10 :745
Cinebench (with mac fan control) Peak Temp @ 85 Deg C
Test 1 : 892
Test 2 : 888
Test 3 : 898
Test 4 : 875
Test 5 : 826
Test 6 : 823
Test 7 : 873
Test 8 : 830
Test 9 : 793
Test 10 :864
then I jump over to prime 95 and i see a mark improvement.
in test 1 & 3, no change, rock solid 2.9ghz across all cores.
when we hit test 2 & 4, thats where it gets interesting. With the fans on auto, the throttle starts immediately, however using FAN SMC Controller, it hits much later, almost at the end of the test. I also noticed that it goes crazy as the thermal draw exceeds 45w and goes up, then it agressively throttles.
i'm quite happy to see that this is throttling can be mitigated to a certain degree. we will have to test more and perhaps see what happens over the course of the next few days/weeks.
will wait for more people to chime in with their observations.
update: included my prime95 screenshots
Last edited:
However in order to be meaningful need to have a consistent test & test setup.
I wonder to what extent some of the variation is due to the thermal paste - in which case there may be significant unit-unit variation.
Does anyone have a new 2018 MBP15 i7 2.2 that can run a set of 5 tests with Cinebench CPU and Intel Power Gadget? (with standard - as delivered by Apple - fan control)
Edit: specified fan control
I wonder to what extent some of the variation is due to the thermal paste - in which case there may be significant unit-unit variation.
Does anyone have a new 2018 MBP15 i7 2.2 that can run a set of 5 tests with Cinebench CPU and Intel Power Gadget? (with standard - as delivered by Apple - fan control)
Edit: specified fan control
Last edited:
Look up a few posts. I ran 10 straight Cinebench tests on my base 15" with the 2.2...
I have also since installed Macs Fan Control to set the fans to be more aggressive and now get scores ranging from 860-933.
Edit: Noticed that you mention Intel Power Gadget. I didn't take screenshots this time around, but it still throttled after about 9-10 seconds, but even when it cuts back it maintains at least the 2.2 clock speed I would say 95% of the time - while still turbo boosting occasionally to the low 3.0s. As low of a clock speed I have seen during testing would be about 1.8ghz, but that is few and far between.
Last edited:
Thank you - unfortunately without the Intel Power Gadget showing the actual behavior during the runs it's a little bit hard to draw conclusions.
I just did a test on my 2013 rMBP13 last night for comparison and interestingly I don't see any throttling - just a clean monotonic clock frequency curve.
NB I'm currently in a email exchange with a client - trying to debug a power measurement setup remotely - and temporal information is essential to be able to draw conclusions.
Last edited:
I’m getting 95.61fps on the 560X. How about your 555X? And regarding the Dell, I guess this shows that while Intel is not blameless in this, Apple really screwed the pooch on cooling this year.
ondert
macrumors 6502a
Are these 6-core i7 machines (the 2.6GHz specifically) warm to the touch when idling, word processing or using Photoshop or Illustrator?
Also, has anyone checked the CPU temps during the use cases listed above?
Also, has anyone checked the CPU temps during the use cases listed above?