Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

kotlos

macrumors member
Mar 20, 2017
57
50
The result in cinebench with a fixed 45W limit looked very similar between the i7 and the i9.
So I run a Geekbench test I verified that the i9 is only an unlocked i7. The geekbench score for the i9 with the fixed 45W is ~20700 for the multicore.
So while restricting the wattage can be helpful in CPU heavy tasks such as scientific computations/video rendering/etc for the average usage when all of the cores are used for less than few seconds, then you might be better off leaving the default settings.
 
  • Like
Reactions: Feenician

Feenician

macrumors 603
Jun 13, 2016
5,313
5,100
The result in cinebench with a fixed 45W limit looked very similar between the i7 and the i9.
So I run a Geekbench test I verified that the i9 is only an unlocked i7. The geekbench score for the i9 with the fixed 45W is ~20700 for the multicore.
So while restricting the wattage can be helpful in CPU heavy tasks such as scientific computations/video rendering/etc for the average usage when all of the cores are used for less than few seconds, then you might be better off leaving the default settings.

Yeah for sure. Most of the issues in general are only really observed when using crushing benchmarks, or sustained heavy real workloads though aren’t they? Or do I have that wrong?
 

kotlos

macrumors member
Mar 20, 2017
57
50
Have you downloaded the fan control software? macsfancontrol, I think it’s called?

If you couple a more aggressive fan response with what you’ve done so far, it might help even more.

I have the macsfancontrol but as you see in the test the frequency is stable across the test since there is no temperature limit.

Here is the test with the fans at max and the 45W limit:
Screen Shot 2018-07-20 at 10.17.44 PM.png
And here is with no turboboost:
Screen Shot 2018-07-20 at 10.12.49 PM.png
 

ha1o2surfer

macrumors 6502
Sep 24, 2013
425
46
I have the macsfancontrol but as you see in the test the frequency is stable across the test since there is no temperature limit.

Here is the test with the fans at max and the 45W limit:
View attachment 771865
And here is with no turboboost:
View attachment 771866
So if the fans are at full, it runs 100mhz above base clock no matter what, even with turbo boost off lol So your machine is not throttling in those pics at all. success!

Although not really as a 8750H can score 1270 in a properly cooled windows laptop
 
Last edited:
  • Like
Reactions: Eason85

kotlos

macrumors member
Mar 20, 2017
57
50
So if the fans are at full, it runs 100mhz above base clock no matter what, even with turbo boost off lol So your machine is not throttling in those pics at all. success!

The problem is that the increased performance of i9 is only achieved when the CPU is allowed to consume so much power. So I am not sure how much of a success it is :)

I am confident that apple can at least provide some remedy with a combination of wattage limitation after few seconds + some tighter coupling of fan speed and the temperatures.
 
  • Like
Reactions: Feenician

ha1o2surfer

macrumors 6502
Sep 24, 2013
425
46
The problem is that the increased performance of i9 is only achieved when the CPU is allowed to consume so much power. So I am not sure how of a success it is :)

I am confident that apple can at least provide some remedy with a combination of wattage limitation after few seconds + some tighter coupling of fan speed to temperatures.
Success or shall I say in Apple terms... what's the problem LOL
 

Feenician

macrumors 603
Jun 13, 2016
5,313
5,100
The problem is that the increased performance of i9 is only achieved when the CPU is allowed to consume so much power. So I am not sure how much of a success it is :)

I am confident that apple can at least provide some remedy with a combination of wattage limitation after few seconds + some tighter coupling of fan speed and the temperatures.

Do you think if Volta had options for higher TDP, as the Windows XTU utility does, there’d be an eventual separation between i7 and i9? You still only seem to be hitting 80 degrees, which is a full 20 less than my 2015 does before it pulls back.

Success or shall I say in Apple terms... what's the problem LOL

What do you mean? Apple are always completely forthcoming and honest about their failures :p
 

kotlos

macrumors member
Mar 20, 2017
57
50
Do you think if Volta had options for higher TDP, as the Windows XTU utility does, there’d be an eventual separation between i7 and i9? You still only seem to be hitting 80 degrees, which is a full 20 less than my 2015 does before it pulls back.

Yeah definitely. I think for best performance we could even restrict the wattage as the temperature reaches 100C. The problem is that the CPU reaches 100C instantly when >60W... Lets see how quickly Apple's engineers figure it out :)
 
Last edited:

ha1o2surfer

macrumors 6502
Sep 24, 2013
425
46
Yeah definitely. I think for best performance we could even restrict the wattage as the temperature reaches 100C. The problem is that the CPU reaches 100C instantly when >50W... Lets see how quickly Apple's engineers figure it out :)

Want to know something funny, while us windows people want big companies to stop controlling the TDP (cough HP, Lenovo...) Apple lets a fully unlocked i9 turbo to 80 watts and only limits the CPU based on thermals. Funny huh You guys are kinda lucky in a way
 
  • Like
Reactions: NBAasDOGG

Feenician

macrumors 603
Jun 13, 2016
5,313
5,100
Yeah definitely. I think for best performance we could even restrict the wattage as the temperature reaches 100C. Lets see how quickly Apple's engineers figure it out :)

That’s great. Thanks for doing all the testing. Dynamic wattage control seems like the answer. I only became aware that Apple even did this differently to everyone else from an article on notebookcheck I found a few days ago. When I was trying to find that to reply to @maflynn about temps he was seeing on his 2018 I stumbled on the new article.

It’s hard to understand how they released this in this state at all but at least there’s a practical amelioration.
 

kotlos

macrumors member
Mar 20, 2017
57
50
I did some more tests with undervolting with the Volta app. I doubt it is actually undervolting since you don't get any gains in frequency for the same wattage.
Undervolting by -55mV:
Screen Shot 2018-07-20 at 11.00.16 PM.png

Undervolting by -55mV + 45W limit:
Screen Shot 2018-07-20 at 10.58.34 PM.png

And this is for @Feenician where I let the CPU free for the first few seconds before the throttling and then restricting the wattage to 45W:
Screen Shot 2018-07-20 at 11.02.09 PM.png
As you say, if we could allow the wattage ~50-55W after the first few seconds, I believe it would get the max out of this CPU.
 
  • Like
Reactions: ksj1 and Feenician

ksj1

macrumors 6502
Jul 17, 2018
294
535
Once Apple updates the software they are going to instantly get in trouble for false advertising. That chip limited to 45 watts will not see anything above 3.5ghz when Apples site clearly states turbo boost up to 4.8

Not sure I understand why you believe this. The chip is designed with a TDP of 45 watts. It will hit a turbo boost of 4.8 Ghz single core as Intel designed it, if you give it 45 watts. It will clearly peak higher however it quickly overheats. Where are you getting 3.5Ghz?

Edit - Let me rephrase. It doesn't peak higher, they are boosting all of the cores, rather than one, so the chip gets hot, as expected.
 
Last edited:
  • Like
Reactions: Cusa

Feenician

macrumors 603
Jun 13, 2016
5,313
5,100
I did some more tests with undervolting with the Volta app. I doubt it is actually undervolting since you don't get any gains in frequency for the same wattage.
Undervolting by -55mV:
View attachment 771872

Undervolting by -55mV + 45W limit:
View attachment 771873

And this is for @Feenician where I let the CPU free for the first few seconds before the throttling and then restricting the wattage to 45W:
View attachment 771871
As you say, if we could allow the wattage ~50-55W after the first few seconds, I believe it would get the max out of this CPU.

Great stuff. Thanks again for doing all this.
 
  • Like
Reactions: kotlos

mr.anthonyramos

macrumors 6502a
Apr 25, 2015
524
380
Hong Kong
Please read the article. It’s not about fan curves at all. Even the first paragraph and/or the conclusion will tell you what you need to know.

Thanks for sharing this Btw. I think what you shared has some very good validity.

The funny thing is, people who obviously have some anger issues with Apple will read this but not see that THIS does prove that there has been some lapses at Apple but there is a proper fix that Apple can and at this point we can be very certain will release.

Thanks again. I’d say don’t mind people who see a solution in from of them but divert the conversation just so that they could continue the blame game.

Still not saying Apple has it together with Thermal dynamics with its Macbook Chassis but still we can probably see improvements soon! Thanks again!
 

ESA

macrumors member
Oct 25, 2015
83
56
Was the same with the first macbook air. Undervolting helped a lot. But I think you can´t do it now.

But will test this for sure, think 50-55W is the good place to be. and keep Turbo Boost. My 2.9 ghx i7 2016 NEVER extends 45W.

Score over 1000 in cinebench is good. Would be cool to do the same test with the i7.

And to all the negative people, please stop! You have said the same thing 100 times now, we know!
 
  • Like
Reactions: mr.anthonyramos

ksj1

macrumors 6502
Jul 17, 2018
294
535
I'm curious. Has anyone run these tests in clamshell mode using an external monitor? That way the dGPU is used instead of the iGPU?
 
Last edited:

jerwin

Suspended
Jun 13, 2015
2,895
4,651
So I run a Geekbench test I verified that the i9 is only an unlocked i7. The geekbench score for the i9 with the fixed 45W is ~20700 for the multicore.

Intel advertises it as having more cache. The unlocked multiplier is a cruel joke, though

Near%20Final_8th%20Gen%20Intel%20Core%20April%20Family%20Update%20Overview%20Deck-14.jpg


source

now, if it supported AVX-512, like some earlier "i9" parts (7900X-- an extreme desktop part with ten cores, and a $1000 price tag), that might have been of real interest to "pros".
 

Feenician

macrumors 603
Jun 13, 2016
5,313
5,100
Thanks for sharing this Btw. I think what you shared has some very good validity.

The funny thing is, people who obviously have some anger issues with Apple will read this but not see that THIS does prove that there has been some lapses at Apple but there is a proper fix that Apple can and at this point we can be very certain will release.

Thanks again. I’d say don’t mind people who see a solution in from of them but divert the conversation just so that they could continue the blame game.

Still not saying Apple has it together with Thermal dynamics with its Macbook Chassis but still we can probably see improvements soon! Thanks again!

Thank you for the nice feedback :) @kotlos did all the hard work tonight though, so they’re the one to thank. Looking forward to more results coming in from our community.
 
Last edited:

Howard2k

macrumors 603
Mar 10, 2016
5,673
5,590
I said this in the thread earlier and I'll repeat it again:

When you, the end user, purchase a device like a 2018 Apple MacBook Pro powered by an Intel i9 processor and you have to install or resort to using a third-party tool to ensure you're getting good/great performance from the hardware, that's a problem.

Can't make it any simpler than that. All that software is doing is artificially crippling performance to make it function in a more thermally effective manner, which means the end user paid for a level of performance they can't even get natively with the hardware itself but if they spend a few more bucks (a pittance compared to the cost of the hardware, I'll admit) and still don't get even with the third-party assist, that's a problem.

The new MBPs powered by the i9 appear - in many videos and reviews so far, this early on in their release - to show a massive issue with thermal efficiency and they do not appear to be performing at spec as intended because of those thermal issues. So the idea that "Oh, let me install this piece of software that will intentionally lower the performance I paid for and should be getting just so I can get at least most of the work done at a level I can deal with" is absurd on so many levels it's not even funny.


This makes a ******** of sense.

Your other post, you lost me at the first paragraph. But this one was good.
 

ksj1

macrumors 6502
Jul 17, 2018
294
535
This makes a ******** of sense.

Your other post, you lost me at the first paragraph. But this one was good.

I disagree. The software is forcing the cpu to operate within it's limits. Apple, like most manufacturers is doing two things.

- They have built the OS to support the profile of the majority of users, who will benefit from the increased short term performance for the tasks they do every day.

- They are boosting all cores rather than the single core boost the chip was designed for.

These however have implications for other users (obviously). It will be interesting to see what Apple comes up with. Or perhaps they already have given FCP's improved performance. They can't force benchmark dev's or vendors to use the things they've built to mitigate the issue.
 

Two Stacks

macrumors 6502
Apr 22, 2015
261
117
New Orleans
What was the immediate issue it had?

I changed the name of the admin account and restarted the computer as instructed by macOS. This left me without an admin account. AC+ via chat had me do a ton of troubleshooting. Last resort was reformat. Tried to do internet install of os....install from external drive, nope. Brought it back to BBY, explained issue and they swapped it out for me.
 

mr.anthonyramos

macrumors 6502a
Apr 25, 2015
524
380
Hong Kong
Here's another test:


This is by far one of the best explanations/reviews I’ve heard so far. Yeah sure it throttles but is it a slouch? No. Is the 2017 top spec 15 inch faster? No. Is the 2018 i7 faster? No.

Will the i9 run hotter? Yes. Is there an issue with Apple’s thermal build on the chassis? Yes. Is the i9 the best performing MacBook Pro you can buy? Yes. Is the i9 needed for everyone...of course not.

@stringerhye I think we can keep an eye on this guy who has done a flurry of different tests with different applications and different workflows.
 

semajm85

macrumors member
Nov 30, 2012
81
88
This is by far one of the best explanations/reviews I’ve heard so far. Yeah sure it throttles but is it a slouch? No. Is the 2017 top spec 15 inch faster? No. Is the 2018 i7 faster? No.

Will the i9 run hotter? Yes. Is there an issue with Apple’s thermal build on the chassis? Yes. Is the i9 the best performing MacBook Pro you can buy? Yes. Is the i9 needed for everyone...of course not.

@stringerhye I think we can keep an eye on this guy who has done a flurry of different tests with different applications and different workflows.

totally agree with the observations made in the video.

sadly i've decided to return my i9 and get the 2.6 i7....i tried everything from disabling turbo, to putting a fixed TDP and all with good results....but my brain couldnt wrap around spending another $400.

i wanted a lower thermal envelope as well as better battery life. a good buddy of mine with a similar workflow sees that with his i7 2.6 so thats how i decided.

good luck i9 users. i certainly hope that Apple will release something to mitigate this without intervention from the users. till then, peace.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.