Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Posted in the i9 thread too, as the same issue affected both.

From my first test running 8 threads 'yes' binary on 13inch i7 2018 MBP before and after - it's fixed:
Before:
before.png

After:
After.png


Performance greatly improved. Maintains higher than base clock speeds whereas before it was dipping under base frequency. No drop off to 800Mhz.
 
Posted in the i9 thread too, as the same issue affected both.

From my first test running 8 threads 'yes' binary on 13inch i7 2018 MBP before and after - it's fixed:
Before:
View attachment 772526
After:
View attachment 772525

Performance greatly improved. Maintains higher than base clock speeds whereas before it was dipping under base frequency. No drop off to 800Mhz.

Thanks for being so quick on it!
 
Posted in the i9 thread too, as the same issue affected both.

From my first test running 8 threads 'yes' binary on 13inch i7 2018 MBP before and after - it's fixed:
Before:
View attachment 772526
After:
View attachment 772525

Performance greatly improved. Maintains higher than base clock speeds whereas before it was dipping under base frequency. No drop off to 800Mhz.

One thing I don't understand in the Intel Power Gadget Utility, why is frequency measured as IA? What does IA stand for?
 
I can confirm with some quick benchmarks the this definitely improves things. The CPU is still allowed to pull down more than 50W for about ~10s, but is then ramped down to ~50W max draw, and multicore performance appears to stay around or above base clock. CB scores are ~1000 on a cold test, though the scores will likely still be lower after being under consistent load.
 
  • Like
Reactions: G5isAlive
I can confirm with some quick benchmarks the this definitely improves things. The CPU is still allowed to pull down more than 50W for about ~10s, but is then ramped down to ~50W max draw, and multicore performance appears to stay around or above base clock. CB scores are ~1000 on a cold test, though the scores will likely still be lower after being under consistent load.

2.6 or 2.2 i7? Please post a screenshot of the Intel tool too. Thanks :)
 
I can confirm with some quick benchmarks the this definitely improves things. The CPU is still allowed to pull down more than 50W for about ~10s, but is then ramped down to ~50W max draw, and multicore performance appears to stay around or above base clock. CB scores are ~1000 on a cold test, though the scores will likely still be lower after being under consistent load.

What machine are you testing? i7? i9?
 
  • Like
Reactions: M.Rizk
Another post-fix update: Prime95 torture test also showed no throttling behavior, although it did bring my i7 2.6 machine to its knees after about 90 seconds of sustained punishment (stabilized at ~2Ghz and ~88C).
 
  • Like
Reactions: G5isAlive
Another post-fix update: Prime95 torture test also showed no throttling behavior, although it did bring my machine to its knees after about 90 seconds of sustained punishment (stabilized at ~2Ghz and ~88C).

Which model? Please, if anyone posts results tel us which models :(
 
My results from i7 2,6 GHz postpatch
EDIT: Cinebench CPU runs

0yd4chbmuxb11.png


The last run:
eyzpxtdnuxb11.png
 
Just a quick update from a 2.6ghz, 32gb owner.

CPU temp heats up VERY FAST when plugged in and opening / running any apps. Even if utilization is < 5%, the CPU runs in turbo boost mode toward an average of 3.4 ghz (or higher) and takes the CPU temp up to 80 degrees Celsius in seconds.

Then the fans kick on and the CPU stays in turbo boost mode even though I am just using 1 Windows 10 VM (Outlook and Excel), safari, iMessage, and calendar. Utilization never gets above 8%. Temperature stays around 80 degrees celsius.

Downloaded Turbo Boost Switcher Pro and turned off the Turbo Boost. Now the computer stays at 61-63 degrees celsius (still seems too hot to me), fans are off, and utilization is still under 8%.

For whatever reason, the turbo boosting feature of these processors seems to increase CPU temperatures more than in the past, and faster than in the past. The combination is a computer that gets too hot, fans come on too often, and where CPU speeds jump around all over the place as the computer tries to cool down the CPU. However, the moment the CPU cools down a little, then turbo boost kicks back in, and the whole cycle starts over again.

Disabling turbo boost, for my very minimal workload, is a good solution. Keeps the computer relative cool and the fans off. But I feel bad for the professionals out there who rely on a computer being fast, and staying fast. The combination of the 6 Core i7 CPU and the MB Pro thermal design are a bad match.

Hopefully Apple and/or Intel can offer a software solution, maybe something akin to Windows 10's "Power Profiles", so that even when plugged in, this computer won't try to turbo boost so often when it really doesn't need to (because CPU utilization is so low).

Sorry for the long post, but thought some people considering a 2.6ghz model might appreciate the anecdotal evidence.

Best of luck to all.
 
My results from i7 2,6 GHz postpatch
EDIT: Cinebench CPU runs

0yd4chbmuxb11.png


The last run:
eyzpxtdnuxb11.png

Those scores don't seem that good... My 2.2 i7 got scores like that even straight out of the box (even with some throttling). Gets scores in the mid 900's with Volta set to power limit at 45W. Hmmm..
 
Apple definetly fixed the issues. Getting consistent scores in Cinebench without throttling. This was run on a 15" 2.2Ghz i7.

First score 1032
Second score 963
Third score 983
Fourth score 971
View attachment 772539
Now that's what I like to see... And to think I had my 2.2 15" all boxed up to possibly return this evening. Guess I'll have to get it out and install this update and see how it goes. :cool:
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.