2018 i9 MacBook Pro throttling discussion [merged]

[Sterkenburg]

Right now now I'm mulling it over as the GPU is the weakest link, equally there's always something better around the corner...

The 550-560X is by far the weakest upgrade in the new MBPs, I agree. However, if your needs call for a powerful GPU not even an architecture jump to Vega would make a big difference on that front, as Apple would still use a ~35W mid-range card.

All considered, I think if one is a power user there's no better time to upgrade than now. Even with all the issues, this is probably the biggest leap in performance we'll see from Apple for a while. I'll thoroughly test my i9 as soon as it arrives, but if I see good results using Volta with a 45W limit I'm likely going to keep it.

 

[mr.anthonyramos]

In the video he is comparing his tests to Dave Lee and mentions Dave Lee a lot, but forgets to mention that he tests the i9 against a completely different i7 machine than Dave Lee did. I see a lot of people on MacRumors seem to think this video disproves the test of Dave Lee, while it does not.
Also the comments to this video on YouTube itself mention this.

All I am saying is: the discussion is not quite over yet.

So what results are you implying that the 2017 machine will get over the 2018 machines? This video doesn’t directly compare with Dave Lee’s videos because this is so much more substantial. Different videos renderings, different softwares, work with photos which are more than what Dave Lee did with a Single test with a single application.

If I were to be honest, you are all too fixated on Dave Lee’s videos. What this newer more comprehensive video shows is, real life work performance, different applications, does show that the i9 although it does throttle still bests the other Macbook Pros.

Are we still going to say that the 2017 i7 top spec will deliver better results than the 2018 models?

Because here, based on the other videos as well comparing the 2017 15 inch top of the line, the i7 and the i9, also show performance on the i9 does seem to be better albeit not to its full expectations.

I think after all the other videos doing tests between the different MacBook Pro models, we need to look beyond Dave Lee’s video because sometimes it’s beginning to sound like someone who can’t get over an ex-girlfriend who did you wrong, hence all other girls who come along will also be crappy experiences. lol Sorry for the analogy.

So, moving on from this fixation, what is most interesting to me now is how the i7 2.6 will perform against the i9 and how much of a difference in performance these two would have between them and if the upgrade to i9 will be worth it or not.

Does this make sense?

 

[Petetastic]

Now I need to find someone that has one of the latest Dell Precision workstation laptops powered by an i9, have them install macOS in a VM with VMware Workstation, then have them run the gamut of the same tests everybody else is using and see if it throttles or gets better performance (which is actually quite easy these days using VMs, believe it or not).

Might just have to do this if I can find someone.

Wait, whoa. You can install MacOS as a VM? Where? How?? This is news to me!

 

[M.Rizk]

I've just tried volta and limiting TDP to 45w. It works. cinebench is now consistently scoring 1000+ on my i7 2.6

Volta doesn't allow higher than 45W, right? I am sure we can get even better results with 50W~55W and the cooling system can still handle it just fine.

 

[AndyMacAndMic]

So what results are you implying that the 2017 machine will get over the 2018 machines? This video doesn’t directly compare with Dave Lee’s videos because this is so much more substantial. Different videos renderings, different softwares, work with photos which are more than what Dave Lee did with a Single test with a single application.

If I were to be honest, you are all too fixated on Dave Lee’s videos. What this newer more comprehensive video shows is, real life work performance, different applications, does show that the i9 although it does throttle still bests the other Macbook Pros.

Are we still going to say that the 2017 i7 top spec will deliver better results than the 2018 models?

Because here, based on the other videos as well comparing the 2017 15 inch top of the line, the i7 and the i9, also show performance on the i9 does seem to be better albeit not to its full expectations.

I think after all the other videos doing tests between the different MacBook Pro models, we need to look beyond Dave Lee’s video because some times it’s beginning to sound like someone who can’t get over an ex-girlfriend who did you wrong, hence all other girls who come along will also be crappy experiences. lol Sorry for the analogy.

Does this make sense?

I agree with you, but if it is true that the MacBook pro i9 is not faster than the previous generation MacBook pro i7 it seems quite important information. It certainly is something I would want to know before I spend a lot of money on an i9, don't you agree? The more facts available the better the decision. I am not fixated on Dave Lee per se, but I also did not see anything yet that disproves his test. Waiting for more conclusive/reliable tests seems a logical path to follow.

 

[Queen6]

I've just tried volta and limiting TDP to 45w. It works. cinebench is now consistently scoring 1000+ on my i7 2.6

Not as good as it could be, equally not too shabby for the MBP's form factor.

I'd try with an undervolt, you should get more out of it. I'd also play with the power limits

Q-6

 

[mr.anthonyramos]

I agree with you, but if it is true that the MacBook pro i9 is not faster than the previous generation MacBook pro i7 it seems quite important information. It certainly is something I would want to know before I spend a lot of money on an i9, don't you agree? The more facts available the better the decision. I am not fixated on Dave Lee per se, but I also did not see anything yet that disproves his test. Waiting for more conclusive/reliable tests seems a logical path to follow.

Okay...Let me break it down for you again. Answer me this, do you think the i7 base model 2.2 model is slower than the top of the line 2017 15 inch?

 

[Mockletoy]

Very interesting. Check this video out. I think it's the latest update of the i9 throttling issue from a famous YouTuber.

How does any of that matter?

I have a gaming notebook with a 7700HQ in it. That 7700HQ has a 2.8Ghz base clock. I can put that machine under literally any sort of load, even something artificially, stupidly unrealistic like running 8 threads of Prime95 torture test and FurMark at the same time, and it will turbo boost to 3.4Ghz and maintain that boost indefinitely with temps that never get within 20 degrees of thermal throttling. Not at base clock, at all-core boost clock.

During my normal usage, like gaming, the CPU temps hover in the 50’s and 60’s and, again, it’ll hold its maximum boost clocks 24 hours a day if I want it to. That’s how this is supposed to work. That is what I paid for when I bought the machine, and that is what I got.

That video is 10 minutes of someone saying, essentially, “You’re using it wrong.”

Wow.

 

[Queen6]

Volta doesn't allow higher than 45W, right? I am sure we can get even better results with 50W~55W and the cooling system can still handle it just fine.

Maybe approach the Dev. On the Windows some are talking of 52W being optimal. I've left my GL703GS on the 45W default for PL-1 as it's a good chip with great tolerance, delivering knockout performance.

Q-6

 

[augustya]

Ladies & Gentleman, Boy's & Girls this guy Dave Lee has sort of become a Villain on the entire YouTube community and the Apple rMBP audience equally ! As expected with release of all Apple Products people are just dying to find faults in Apple Products !! kind of do Nit Picking ! This guy Dave Lee sounds one of them !! I mean cmon what he does is get a 5K RAW Video uses Adobe Premiere Pro which has always been at loggerheads with the MacBook Architecture and tries to convert it in 4K Video, a perfect recipe to crash the machine !! And then he like CNN goes all over the world broadcasting hey guys ! You know what Apple 2018 15" rMBP i9 machine is Thermal Throttling !! I mean what the F ?? So my question to the audience here is how much of him has to be really believed ?Adobe Premiere Pro is not friendly with the rMBP Architecture and on top of that you still kind of tease it so this is inevitable to happen isn't it ?? What do you guys think about this ? I am asking because perhaps I may just but the 2018 15" rMBP with the i9 Processor and 32 GB of RAM ! Would love to have some opinion about this from people here !!

 

[M.Rizk]

Maybe approach the Dev. On the Windows some are talking of 52W being optimal. I've left my GL703GS on the 45W default for PL-1 as it's a good chip with great tolerance, delivering knockout performance.

Q-6

Someone else mentioned emailing the dev already so lets hope they get a response soon. I am hoping Apple applies such measures in an update soon because I do not want to disable the System Integrity Protection first thing on a new Mac.

 

[Beachguy]

Now English is not my first language

You did a great job- very clear.

 

[iMacDragon]

Someone else mentioned emailing the dev already so lets hope they get a response soon. I am hoping Apple applies such measures in an update soon because I do not want to disable the System Integrity Protection first thing on a new Mac.

Indeed, the evidence is clear that it works and massively improves the situation. They NEED to implement this in FW update.

I just updated my previous post to show a power gadget graph showing some before/after enabling limit to show how clear difference is.

 

[mr.anthonyramos]

How does any of that matter?

I have a gaming notebook with a 7700HQ in it. That 7700HQ has a 2.8Ghz base clock. I can put that machine under literally any sort of load, even something artificially, stupidly unrealistic like running 8 threads of Prime95 torture test and FurMark at the same time, and it will turbo boost to 3.4Ghz and maintain that boost indefinitely with temps that never get within 20 degrees of thermal throttling. Not at base clock, at all-core boost clock.

During my normal usage, like gaming, the CPU temps hover in the 50’s and 60’s and, again, it’ll hold its maximum boost clocks 24 hours a day if I want it to. That’s how this is supposed to work. That is what I paid for when I bought the machine, and that is what I got.

That video is 10 minutes of someone saying, essentially, “You’re using it wrong.”

Wow.

The point of the video is not to compare with “gaming” laptops/machines but to compare the MacBook Pro models based on people’s claims that the 2018 i7s will perform better than the i9 due to its throttling.

And as he mentioned in the video, he will also try and show tests between the 2018 models with the top of the line 2017 model just for clarity sake.

This is not about Apple NOT having any issues because he says it very clear in the video, the 2018 models do suffer from throttling due to the hotter chips and the same ineffieient thermal architecture of the chassis.

 

[AndyMacAndMic]

Okay...Let me break it down for you again. Answer me this, do you think the i7 base model 2.2 model is slower than the top of the line 2017 15 inch?

I think you don't understand what I am trying to say and your breakdown has nothing to do with what I was claiming.

I think the i7 base model 2018 is slower than the top model i9 2018. That is to be expected because the i7 and the i9 2018 are basically the same 6 core processors. Also that is measured in the mentioned video of Jonathan Morisson. That point is clear.

However... it is still very possible that the top of the line i9 2018 is not faster than the previous top of the line i7 2017 according to Dave Lee. The video of Jonathan Morisson (how good and informed it may be) does not give an answer to that.

I personally do not want to spend more money on an i9 if it is not faster than the previous generation i7. As long there is the slightest chance this can be the case I am waiting until more information emerges. Nothing more and nothing less.

Edit: some auto correction typo's fixed.

 

[Mockletoy]

The point of the video is not to compare with “gaming” laptops/machines but to compare the MacBook Pro models based on people’s claims that the 2018 i7s will perform better than the i9 due to its throttling.

And as he mentioned in the video, he will also try and show tests between the 2018 models with the top of the line 2017 model just for clarity sake.

This is not about Apple NOT having any issues because he says it very clear in the video, the 2018 models do suffer from throttling due to the hotter chips and the same ineffieient thermal architecture of the chassis.

My point wasn’t to compare the machine to mine, either, it was to point out how CPUs are supposed to work and to express my distaste at people trying to find literally anyone or anything else to blame for why their $3000+ computers are not working to their full potential.

I wasn’t comparing one machine to another except in the sense that I was comparing the behavior of a functioning machine to the behavior of a malfunctioning one.

 

[mr.anthonyramos]

I think you don't understand what I am trying to say and your breakdown has nothing to do with what I was claiming.

I think the i7 base model 2018 is slower than the top model i9 2018. That is to be expected because the i7 and the i9 2018 are basically the same 6 core processors. Also that is measured in the mentioned video of Jonathan Morisson. That point is clear.

However... it is still very possible that the top of the line i9 2018 is not faster than the previous top of the line i7 2017 according to Dave Lee. The video of Jonathan Morisson (how good and informed it may be) does not give an answer to that.

I personally do not want to spend money on an i9 if it is not faster than the previous generation i7. As long there is the slightest change this can be the case I am waiting until more information emerges. Nothing more and nothing less.

Oh gosh. I’m holding in my sarcasm really hard here. Answer my question...do you think that the 2018 i7 2.2 performs better than the 2017 top of the line?

And again we’re fixated on Dave Lee, the ex-girlfriend that you all can’t get over. Go ahead and answer my question.

 

[Mockletoy]

Oh gosh. I’m holding in my sarcasm really hard here. Answer my question...do you think that the 2018 i7 2.2 performs better than the 2017 top of the line?

And again we’re fixated on Dave Lee, the ex-girlfriend that you all can’t get over. Go ahead and answer my question.

That question is silly.

The question is not, “Is this CPU faster than this or that other CPU,” the question is, “Is this CPU as fast as it should be?”

Is it working as designed and advertised? Is it functioning properly?

The answer certainly appears to be no. No, it is not. Adobe is not to blame. Intel is not to blame. Users are not to blame. Apple’s engineering decisions are to blame.

 

[mr.anthonyramos]

My point wasn’t to compare the machine to mine, either, it was to point out how CPUs are supposed to work and to express my distaste at people trying to find literally anyone or anything else to blame for why their $3000+ computers are not working to their full potential.

I wasn’t comparing one machine to another except in the sense that I was comparing the behavior of a functioning machine to the behavior of a malfunctioning one.

Yes and the guy I the video was not talking about how CPUs were supposed to work. Again, spelt out for you, he was dispelling any false info that the i9 is slower than the base 2018 i7 15 inch. He wasn’t doing a video trying to justity the CPU power of the 2018 macs. Which is why, to answer your question of “why does it matter?” Because some people here are making a purchase decision between buying the i7 and the i9 because there have been reports of the i9 being slower then the base i7. Do you get it?

 

[AndyMacAndMic]

Oh gosh. I’m holding in my sarcasm really hard here. Answer my question...do you think that the 2018 i7 2.2 performs better than the 2017 top of the line?

And again we’re fixated on Dave Lee, the ex-girlfriend that you all can’t get over. Go ahead and answer my question.

I was as factual, respectful and friendly as I could be. You force a question on me that had nothing to do with what I was stating.
If you are incapable of talking factual and going to be sarcastic/belittling about a 'fixation' and 'ex girlfriends' I have nothing more to discuss with you.

 

[M.Rizk]

I think you don't understand what I am trying to say and your breakdown has nothing to do with what I was claiming.

I think the i7 base model 2018 is slower than the top model i9 2018. That is to be expected because the i7 and the i9 2018 are basically the same 6 core processors. Also that is measured in the mentioned video of Jonathan Morisson. That point is clear.

However... it is still very possible that the top of the line i9 2018 is not faster than the previous top of the line i7 2017 according to Dave Lee. The video of Jonathan Morisson (how good and informed it may be) does not give an answer to that.

I personally do not want to spend money on an i9 if it is not faster than the previous generation i7. As long there is the slightest change this can be the case I am waiting until more information emerges. Nothing more and nothing less.

While I can't test Premier Pro since I do not own any of the two devices, here is a Cinebench 15 comparison between the Apple MacBook Pro 15 2017 (2.8 GHz, 555) Intel Core i7-7700HQ (ORANGE) and the Apple MacBook Pro 15 2018 (2.6 GHz, 560X) Intel Core i7-8850H (RED) . According to NotebookCheck Cinebench 15 has been left on a loop on both devices and results were recorded.

Now this is all done on default settings on both MBPs. The 2017 i7 MBP is stable at around 732 points. The 2018 i7 MBP which is throttled and requires more optimization is at around 840 points after sustained workload.

Now how about we power limit the CPU to 45W and test the 2018 i7 again?
Orange = MBP 2018 i7 2.6 GHz after 45W limit
Red = MBP 2018 i7 2.6 GHz without power limit on Windows 10
Gray = MBP 2018 i7 2.6 GHz without power limit on Mac OS

As you can see the score was raised to around 1010 after applying power limit beating the i7 2017 by over 250 points. This is not even full optimization because as been noticed by all reviewers, the 2017 fans kick in much earlier than the 2018 models. If we raise the fan speed in the 2018 i7 model I am sure we can get a few more points.

Now imagine what the i9 can do which already beats the i7 when in the current default non optimized state?

 

[ESA]

If I´m done my research right..

Undervolt -140 mV
Keep Turbo Boost
Set the W to 55
Set the fans between 50-80 degrees.

Then it would work as best I think.
Would be nice to compare that between i9 2.9 and i7 2.6

 

[mr.anthonyramos]

That question is silly.

The question is not, “Is this CPU faster than this or that other CPU,” the question is, “Is this CPU as fast as it should be?”

Is it working as designed and advertised? Is it functioning properly?

The answer certainly appears to be no. No, it is not. Adobe is not to blame. Intel is not to blame. Users are not to blame. Apple’s engineering decisions are to blame.

Of course it isn’t silly, silly. The poster mentions the possibility of the 2017 model beating the i9 in performance. So, the question is, does he think the base i7 perform better than the top spec 2017? Because if he says of course if does, then I’ll have him watch the video again and if the 2018 i7 does perform better than the 2017 i7, does he still think the i9 will perform slower than the 2017 model. You’re over reading.

 

[AndyMacAndMic]

While I can't test Premier Pro since I do not own any of the two devices, here is a Cinebench 15 comparison between the Apple MacBook Pro 15 2017 (2.8 GHz, 555) Intel Core i7-7700HQ (ORANGE) and the Apple MacBook Pro 15 2018 (2.6 GHz, 560X) Intel Core i7-8850H (RED) . According to NotebookCheck Cinebench 15 has been left on a loop on both devices and results were recorded.

View attachment 771976

Now this is all done on default settings on both MBPs. The 2017 i7 MBP is stable at around 732 points. The 2018 i7 MBP which is throttled and requires more optimization is at around 840 points after sustained workload.

Now how about we power limit the CPU to 45W and test the 2018 i7 again?
Orange = MBP 2018 i7 2.6 GHz after 45W limit
Red = MBP 2018 i7 2.6 GHz without power limit on Windows 10
Gray = MBP 2018 i7 2.6 GHz without power limit on Mac OS
View attachment 771974

As you can see the score was raised to around 1010 after applying power limit beating the i7 2017 by over 250 points. This is not even full optimization because as been noticed by all reviewers, the 2017 fans kick in much earlier than the 2018 models. If we raise the fan speed in the 2018 i7 model I am sure we can get a few more points.

Now imagine what the i9 can do which already beats the i7 when in the current default non optimized state?

Thanks! Those are facts I can work with

 

[mr.anthonyramos]

I was as factual, respectfull and friendly as I could be. You force a question on me that had nothing to do with what I was stating.
If you are incapable of talking factual and going to be belittling about a 'fixation' and 'ex girlfriends' I have nothing more to discuss with you.

I too am being respectful as well as trying to state out facts and using common sense here. If the base 2018 hexagon core performs better than the 2017 i7, wouldn’t it be common sense, and even after watching some of these videos, that the i9 would too perform better generally than the 2017 i7?

No intention of belittling but it is true how everyone seems too fixated on a single video when there are now tons of over videos online, yes confirming throttling on the 2018 models, but also still confirming that 2018 models do perform better than 2017 models. Whether or not there is an issue with the throttling, thermals...of course there is, no one is arguing with that. I agree that there is. But focusing on the point of performance between the Mac systems, the 2018 models do perform better.