Busting the "throttling is evil" myth

[richinaus]

The thermal management has to be addressed.

Bootcamp is horrid!

There has always been a lack of thermal management when running Bootcamp. However, it has never been to the point of thermal throttling out of the gate and causing such erratic performance.

Bootcamp will be pushed off and dismissed by many with the following statement. "If you want to run Windows, buy a PC"

To my response is, If your going to have a feature called Bootcamp, make sure the feature works correctly.

I had assumed this would be the case, but had hoped it wouldn't be. Your post has probably tipped me over the edge to get a windows laptop now, as I have had terrible experiences with macbookpro's and bootcamp.

 

[leman]

I understand that you may not see that much of a difference, but it significantly helps those with workloads requiring the cores.

Let me be more precise: I do obviously see a healthy difference in performance (and that is also why I got the 2018 model to replace my 2016), what I don't really see is any noteworthy technological progress.

Come on now... taking the architecture from 45w to 15w and blowing the door open for ultra portables to harness the power is more than a rebadge.

True. Turbo boost (and all things that came out of it) was a very smart thing to do and it really changed what one can do with modern machines. At the same time, it also makes stuff much more complex to the user. Look at a burst benchmark of a modern 15W chip and it almost matches a premium 45W performance part. In real sustained workloads, its still going to be substantially slower though. All this creates potential for abusive marketing and user confusion...

 

[Maven1975]

I had assumed this would be the case, but had hoped it wouldn't be. Your post has probably tipped me over the edge to get a windows laptop now, as I have had terrible experiences with macbookpro's and bootcamp.

Hang on a moment...

I have not verified this myself so take it with a grain of salt.

Was just speaking with a coworker that has a 2018 2.3 13" and he claims that his system is running better under bootcamp today after applying the Speaker/Bridge OS fix.

I have no idea what's going on but perhaps Apple just fixed things inadvertently. (Or labeled the fix incorrectly as the issues it was to address still exist)

Will not be able personally test till tomorrow :/

 

[leman]

To my response is, If your going to have a feature called Bootcamp, make sure the feature works correctly.

Its a bit surprising that Apple didn't do it.. I mean, setting the TDP limit is just two lines of code in the bootloader...

 

[Maven1975]

Let me be more precise: I do obviously see a healthy difference in performance (and that is also why I got the 2018 model to replace my 2016), what I don't really see is any noteworthy technological progress.



True. Turbo boost (and all things that came out of it) was a very smart thing to do and it really changed what one can do with modern machines. At the same time, it also makes stuff much more complex to the user. Look at a burst benchmark of a modern 15W chip and it almost matches a premium 45W performance part. In real sustained workloads, its still going to be substantially slower though. All this creates potential for abusive marketing and user confusion...

What do you think the performance real world between the 2.3 and 2.7 28w 13" chips are? I understand we are talking about 400mhz base clock, but under sustained load with boosts, would you think the 2.3 would hold load close to the 2.7?

I read that a user has been able to see their 2.7 hold at 4.1. That's a darn good boost

 

[maflynn]

Bootcamp is horrid!

Yes, I know

To my response is, If your going to have a feature called Bootcamp, make sure the feature works correctly.

Agreed, its a supported feature, and clearly Apple hasn't done anything to address this to the detriment of their customers

 

[StellarVixen]

BootCamp performance has been bad since I am using BootCamp.

Grand Theft Auto: San Andreas on BootCamp runs at 15-20 FPS at medium settings.
Same game made with WineSkin runs at 35-40 FPS on Mac OS.

Java projects compile faster on Mac than on BootCamp Windows.

Same hardware different OS, different performance.

I think it is drivers, and I do not know why Apple provides such bad drivers.

 

[leman]

What do you think the performance real world between the 2.3 and 2.7 28w 13" chips are? I understand we are talking about 400mhz base clock, but under sustained load with boosts, would you think the 2.3 would hold load close to the 2.7?

My bet would be around 5%, maybe

 

[maflynn]

BootCamp performance has been bad since I am using BootCamp.

I use windows, but I don't play games, so I'm thinking of using Vmware over bootcamp just because of the thermal management. There's also the advantage of keeping macOS running

 

[Queen6]

Technically the CPU's are not "throttling" in many cases, however what's certain is that the current MBP's cooling and power solutions are inadequate for optimal performance, and that is far more important to the majority of users rather than semantics...

Intel's 8750H in conjunction with a competent cooling solution...

Ambient 26C, CPU peak 89C, reduces as fans spool up, six core at 3.9GHz, holds 3.4GHz irrespective of duration & load, averaging low 70C.

Q-6

 

[1096bimu]

I don't think there is a lot of people who think like this - and to be fair the main issue came about because the throttle issue resulted in the MacBook's going below the base clock (which has now been fixed). Truthfully, I think you are talking about a non-issue as I rarely have seen complaints about throttling over the years (and the recent spike as stated earlier was due to a VRM issue). Even on Notebookreview.com on the gaming notebook subforums, it is pretty much accepted that the laptops won't hold max clock speed all the time.

So I already quoted Linus Tech Tips, here's another from NotebookChek.com's review of the 13" MacBook

The advantage of the optional Core i7 should be very small, and it might even be slower than the i5 due to higher temperatures (especially under sustained workloads.

Lots of people don't understand how processor power scaling works. It's also hilarious how they get the base model of the 13" with the i5 and complain that it's not as fast as the i7 in the Windows machines. AND they have since refused to update the review with the i7 model because the i7 model beats all other 13" Windows laptops.

Never underestimate the desire for new excuses to bash the latest Apple products. Just last year we had the PWM thing being blown out of proportion just because the iPhone X now uses an OLED screen, even though most laptops today still have a PWM frequency of around 200Hz and Samsung has been using PWM on their phones since 2008.
This year the latest excuse is throttling, people will kill to bash the latest MacBook, even if the excuse blatantly goes against objective facts.

What is a valid cricitism (this isn't Apple related, but I am talking laptops in general including Windows) is comparing average frequency when stressed. Some laptops can manage say the max turbo boost for 1-2 seconds before going down, while another laptop may be able to handle it for a minute. Some manufactuers can greatly improve how much CPU they utilise by quality cooling solutions/heatsinks (it doesn't have to become a 5kg behemoth). For example, an i5 XPS 13 has managed to out perform many Window laptops on the i7, just because it offered a better cooling solution (without being larger).

Yes, when you allocate more space to the cooling system you will get more cooling and therefore more performance. But I don't think that is a criticism because everyone will have a different idea of what is the ideal size/weight for a 15" laptop and what percentage of its internals should be allocated to cooling, as opposed to speakers, battery or logic board. You may disagree with Apple's decision but that doesn't make it wrong.
[doublepost=1535630111][/doublepost]

Technically the CPU's are not "throttling" in many cases, however what's certain is that the current MBP's cooling and power solutions are inadequate for optimal performance, and that is far more important to the majority of users rather than semantics...

Define "Optimal" and "adequate".

 

[maflynn]

Technically the CPU's are not "throttling" in many cases, however what's certain is that the current MBP's cooling and power solutions are inadequate for optimal performance, and that is far more important to the majority of users rather than semantics...

Agreed, due to heat, the MBP is not achieving optimal performance, many people will call it throttling, its really doesn't matter. The fact remains that heat is causing the CPU not to run at its highest ability.

The fact remains computer makers (not just dell), slapped the Coffee Lake processor into last year's design and expected everything to be hunky dory, but that's just not the case, especially when people like Dell, Apple and others put in the I9 into the thin laptop.

Many computer makers got lazy and now we consumers are paying the price.

I will say my 2018 MBP runs cooler, quieter, and can be used on my lap over my prior Razer. Granted the Razer had a beast of a GPU, but the fact remains I'm getting better usage out of the Mac, so I'm not upset about this

 

[Queen6]

Many computer makers got lazy and now we consumers are paying the price.

I will say my 2018 MBP runs cooler, quieter, and can be used on my lap over my prior Razer. Granted the Razer had a beast of a GPU, but the fact remains I'm getting better usage out of the Mac, so I'm not upset about this

There's a solution to that, which certainly doesn't involve rewarding them for these poor efforts...

Q-6
[doublepost=1535631106][/doublepost]

Define "Optimal" and "adequate".

Full Turbo for the PL-1 limit, no thermal throttling for the PL-2 limit at 45W/52W . These CPU's are designed to boost to high clock frequencies, not be held back by poor/lazy design choice. If your willing to settle for less performance it's ultimately moot...

Q-6

 

[1096bimu]

The fact remains computer makers (not just dell), slapped the Coffee Lake processor into last year's design and expected everything to be hunky dory, but that's just not the case, especially when people like Dell, Apple and others put in the I9 into the thin laptop.

Many computer makers got lazy and now we consumers are paying the price.

You can't just appeal to ridicule by saying "lol put an i9 into a thin laptop" because there is absolutely no reason to think that is in any way bad practice.

A processor acts exactly the same as an electric heater to the thermal system, regardless of what it is called or how many cores it has. A 45W i9 is exactly the same as a 45W i5 to the thermal system so why can't you replace one with the other?

You could argue that given unlimited power and cooling, the i9 has more potential than last year's i7. But that's irrelevant because we don't have unlimited power and cooling.

 

[leman]

Intel's 8750H in conjunction with a competent cooling solution...

Ambient 26C, CPU peak 89C, reduces as fans spool up, six core at 3.9GHz, holds 3.4GHz irrespective of duration & load, averaging low 70C.

You should probably mention that its in a chassis thats 1kg heavier and 1cm thicker then the MBP... while being around twice the overall volume of the MBP

 

[1096bimu]

Full Turbo for the PL-1 limit, no thermal throttling for the PL-2 limit at 45W/52W . These CPU's are designed to boost to high clock frequencies, not be held back by poor/lazy design choice. If your willing to settle for less performance it's ultimately moot...

Q-6

And why are you the authority on what should be considered "Optimal"?

No they are not designed to boost to high clock frequencies constantly. Intel only recommends that they be able to run at the base frequency constantly. Not even that is a technical limitation, it's just arbitrary and it is perfectly possible to run a processor at below TDP specifications.

I am not settling for less performance I am settling for less size and weight, which determines performance through the laws of physics.

automatically allowing processor cores to run faster than the rated operating frequency if they’re operating below power, current, and temperature specification limits.

 

[maflynn]

A 45W i9 is exactly the same as a 45W i5 to the thermal system so why can't you replace one with the other?

Because I believe the i9 goes beyond the 45w in turbo boost higher then the i7, its also running at a higher clock speed.

I'm no chip expert, but there's plenty of anecdotal evidence that the I9 does run hotter then its i7 counterpart. Its just not an apple thing, just look at any laptop and its reviews of the i7 vs i9. We see a lot of discussions on how the i9 throttles more then the i7 in thinner bodies, such as the dell.

 

[leman]

The fact remains computer makers (not just dell), slapped the Coffee Lake processor into last year's design and expected everything to be hunky dory, but that's just not the case, especially when people like Dell, Apple and others put in the I9 into the thin laptop.

I don't understand this. The i9 has the same thermal specification as any other 45Watt CPU... so if your thin and light chassis can take a gen-5/6/7 45W CPU, then it should be able to take the i9, right?

Many computer makers got lazy and now we consumers are paying the price.

The problem is moreover that Intel sells a CPU which can go WAY beyond its TDP (we are talking 150%!) in order to reach/maintain the upper range of its turbo boost frequency.

The real problem of course is that users expect desktop-level performance from a laptop chip...

 

[maflynn]

I don't understand this. The i9 has the same thermal specification as any other 45Watt CPU... so if your thin and light chassis can take a gen-5/6/7 45W CPU, then it should be able to take the i9, right?

Doesn't the i9 shoot higher in GHz/wattage in turbo mode then the i7? I'm under the impression it does.

 

[leman]

Because I believe the i9 goes beyond the 45w in turbo boost higher then the i7, its also running at a higher clock speed.

Doesn't the i9 shoot higher in GHz/wattage in turbo mode then the i7? I'm under the impression it does.

What is relevant is that the i9 is supposed to run at at a frequency not lower than 2.9Ghz, all cores, all day, when consuming not more then 45 watts of power. That is the official CPU specification and that is the only thing you can really rely on. If your i9 meets these characteristics, its operating normally and does not throttle. Everything else is speculation.

And of course, when you are talking about upper ranges of turbo boost, the CPU can briefly consume much more power (MSI has measured close to 150 watts). Obviously, its not something that Intel openly documents. In another words, they have set the upper frequency limit so high that you need an equivalent of an upper-range desktop cooler in order to maintain it. Personally, I think this is misleading description for a laptop product that is otherwise marketed as a drop-in replacement for other chips in the line.

 

[1096bimu]

Because I believe the i9 goes beyond the 45w in turbo boost higher then the i7, its also running at a higher clock speed.

I'm no chip expert, but there's plenty of anecdotal evidence that the I9 does run hotter then its i7 counterpart. Its just not an apple thing, but just look at any laptop and its reviews of the i7 vs i9. We see a lot of discussions on how the i9 throttles more then the i7 in thinner bodies, such as the dell.

Yes the i9 has a higher ceiling of power consumption, so what? Both the i7 and the i9 will exceed 45w on full turbo. It says nothing about their average power consumption and power consumption to do the same work.

Which by the way if you are wondering, the i9 does more work with the same amount of energy expended, which is the same as saying it generates less heat doing the same job. Yes I said LESS. So if you picture trying to cook an egg with your MacBook, you need to render more videos on the i9 as compared to last year's i7 to fully cook that same egg.

so-called "anecdotal evidence" are massively biased and have always been that way. Much like how people say that every iOS update make the phones run slower (which it doesn't).

https://www.notebookcheck.net/Apple-MacBook-Pro-15-2018-2-6-GHz-560X-Laptop-Review.317358.0.html
https://www.notebookcheck.net/Apple-MacBook-Pro-15-2017-2-8-GHz-555-Laptop-Review.230096.0.html
Professional reviews and hard data would say that the 2018 MacBook Pro 15" is not any hotter than the 2017 version.

 

[maflynn]

And of course, when you are talking about upper ranges of turbo boost, the CPU can briefly consume

I don't think we know the logic in how the CPU decides to go into Turbo so while I won't disagree with the apples to apple comparison (wattage/ghz). We cannot dismiss the body of evidence that suggests that the I9 runs hotter then the i7.

There's just too many reviews that document thermal issues with the i9. We've seen many benchmarks and reports of the base 2.2 MBP beating out the i9 because the thermal problems.

so what?

So What? higher boost equates to higher heat, pure and simple.


I'm not going to beat this horse to death. People can believe what ever they want, you're of the opinion that there's no thermal difference between two identical laptops one running i7 and the other an i9, and I believe there is based that I've seen and read.
[doublepost=1535632559][/doublepost]

so-called "anecdotal evidence" are massively biased and have always been that way. Much like how people say that every iOS update make the phones run slower (which it doesn't).

And your opinion on this is also biased. My opinion is biased. So what

 

[1096bimu]

The problem is moreover that Intel sells a CPU which can go WAY beyond its TDP (we are talking 150%!) in order to reach/maintain the upper range of its turbo boost frequency.

So? Having the option to go way higher is just better than not having the option because you loose nothing by having it.

So What? higher boost equates to higher heat, pure and simple.

So when you are using FCP you rather have it lag a bit more, than to have the machine reach it's working thermal conditions 10 seconds faster during your 3 hour work shift? Why would you want that? Just so that your laptop can be cooler for the first minute?

 

[leman]

I don't think we know the logic in how the CPU decides to go into Turbo so while I won't disagree with the apples to apple comparison (wattage/ghz). We cannot dismiss the body of evidence that suggests that the I9 runs hotter then the i7.

If you run the i9 at higher clock than the i7 can do (4.3Ghz), yes, it will run hotter. At the same frequency, they will produce exactly the same amount of heat or the i9 will be cooler (due to binning).

There's just too many reviews that document thermal issues with the i9. We've seen many benchmarks and reports of the base 2.2 MBP beating out the i9 because the thermal problems.

Frankly, I haven't seen a single review of the likes you say (and I read a lot of them). The cases of 2.2. beating the i9 I know of were from the first week of testing, before Apple fixed the power management bug (which indeed used to cause the i9 to drop boost prematurely).

 

[maflynn]

Just so that your laptop can be cooler for the first minute?

Yes, to be quite frank