Vapor chamber in macbook pro ?

[Queen6]

And about this: back then, Intel published max full core turbo boost. They don't do it anymore. I have no idea if they changed anything under the hood or whether its just a marketing thing, but "turbo" with the new CPUs is basically what max single core turbo used to be.

Full Turbo on an Intel 8th Gen 8750H is 3.9GHZ across all six cores, or 4.1GHz on a single. Seems perfectly clear the majority of people What I see in software reading the CPU registers mirrors, with 1-2 Cores active 4.1Ghz, 3-4 cores active 4.0GHz & 5-6 cores active 3.9Ghz.

Most I assume would be interested in the performance of all cores under load, given the small 0.2GHz differential. What's more important is the long-term Turbo performance. In a decently cooled chassis 3.6GHz across all six cores should be expected, unless the long-term power limit is raised from 45W to say 52W which is manageable by the CPU, then 3.8GHz is possible. For example Gigabyte's Aero 15X runs the long term TDP at 52W, my own Asus GL703GS is set to 45W, although it may be possible to raise the same limit.

Q-6

 

[leman]

We are all entitled to our opinions, and mine is that the Razer with its vapor chamber offers superior cooling then the MBP's heat pipes.

Let me expand my previous answer a bit before it gets misunderstood.

First, there is no doubt that Blade has a much more powerful cooling system (anybody arguing to contrary must frankly, be an... well, not an intelligent person). And with the new cooling design, Razer has achieved an impressive feat of engineering.

At the same time, projecting this on Apple and saying, oh, if only Apple also used a vapour chamber is a massive oversimplification. Compare the design of the two laptops. Razer Blade is literally built around the idea of cooling down that massive GPU. Its heat spreaders are huge compared to the MBP, its fans are larger, it has large air grills on the bottom though which the fans such in the air directly. All this together is what makes it great at cooling, not just the utilisation of the vapour chamber. And it of course has its drawbacks. Increased weight, size, less space for other components, requirement for the bottom to be unobstructed etc. etc.

I'm certainly not saying that Apple shouldn't seek to improve their cooling system. But they will never use Blade's solution, since they are not building a showcase of a gaming machine. And it's not clear where exactly the bottleneck of the MBP's cooling solution is. A vapour chamber only helps if its the heatsink to heatpipe connection. Which is rather unlikely if you ask me.

 

[Newtons Apple]

Why not just make the laptop thicker ?

Because Apple feels like we are willing to sacrifice anything for thin!

I would not care if they added a 1/2" to the thickness. More battery and better cooling!

We have scarificed much for this "thin". Enough already!

 

[Queen6]

Because Apple feels like we are willing to sacrifice anything for thin!

I am not and would not care if they added a 1/2" to the thickness. More battery and better cooling!

Same I'm confident most professional's would far prefer an uplift in performance, usability & reliability, equally Apple's not a provider of professional grade hardware, just nice computers to impress...

Q-6

 

[maflynn]

However, better cooling or not, it doesn't actually do anything to improve the CPU performance,

My point is vapor chambers offer better thermal management, pure and simple.

 

[Newtons Apple]

Same I'm confident most professional's would far prefer an uplift in performance, usability & reliability, equally Apple's not a provider of professional grade hardware, just nice computers to impress...

Q-6

Apple hardware is and always will be my favorite hardware!

 

[Queen6]

Apple hardware is and always will be my favorite hardware!

True, however once it impacts $$$$ things change...

Q-6

 

[leman]

My point is vapor chambers offer better thermal management, pure and simple.

Vapor chamber offers better thermal dispersal, not more and not less. In a system where dispersal is not a bottleneck, vapour chambers won't do a thing.

P.S. I think before thinking about redesigning the thermal system, Apple should get their stuff together and finally make sure that the thermal paste is applied properly + use better quality thermal compound. I am sure that already this will make a healthy difference.

 

[1096bimu]

Well apparently a vapor chamber is much more efficient than heat pipes because there is more surface area to transfer heat. Maybe the issue is with the heat transfer from the cpu to the heatsink and not the fans, in which case a vapor chamber would help.

No it isn't.
No it isn't, it's the heatsink and fans.

 

[Howard2k]

Full Turbo on an Intel 8th Gen 8750H is 3.9GHZ across all six cores, or 4.1GHz on a single. Seems perfectly clear the majority of people What I see in software reading the CPU registers mirrors, with 1-2 Cores active 4.1Ghz, 3-4 cores active 4.0GHz & 5-6 cores active 3.9Ghz.

Most I assume would be interested in the performance of all cores under load, given the small 0.2GHz differential. What's more important is the long-term Turbo performance. In a decently cooled chassis 3.6GHz across all six cores should be expected, unless the long-term power limit is raised from 45W to say 52W which is manageable by the CPU, then 3.8GHz is possible. For example Gigabyte's Aero 15X runs the long term TDP at 52W, my own Asus GL703GS is set to 45W, although it may be possible to raise the same limit.

Q-6

How come you didn't mention the base clock at all? The base clock is 2.2Ghz. Anything above that is managed by the processor based on the thermal conditions isn't it? I could be wrong, but I don't think there is anything in the processor data sheet that shows that the processor should sustain 3.6GHz across all six cores. The specs do mention a pretty interesting methodology for managing the cooling all the same. Well worth a look, and maybe I missed it in there somewhere.

This is how they described their Turbo technology:

Compared with previous generation products, Intel Turbo Boost Technology 2.0 will increase the ratio of application power towards TDP and also allows to increase power above TDP as high as PL2 for short periods of time. Thus, thermal solutions and platform cooling that are designed to less than thermal design guidance might experience thermal and performance issues since more applications will tend to run at the maximum power limit for significant periods of time.
​

TDP, which is elsewhere defined as:

Thermal Design Power (TDP) represents the average power, in watts, the processor dissipates when operating at Base Frequency with all cores active under an Intel-defined, high-complexity workload. Refer to Datasheet for thermal solution requirements.​

Which is 45W.

So unless I'm misunderstanding, above 2.2Ghz, all bets are off. Sure you can define your own limits like "3.6Ghz, sustained" but I think that's purely a personal decision right?


Meaning that it's up to the vendor to make the cooling solution as they see fit. It better maintain 2.2GHz, but above that all bets are off. Again, unless I'm misunderstanding.

 

[Thysanoptera]

TDP, which is elsewhere defined as:

Thermal Design Power (TDP) represents the average power, in watts, the processor dissipates when operating at Base Frequency with all cores active under an Intel-defined, high-complexity workload. Refer to Datasheet for thermal solution requirements.

They forgot to update the popups. I guess. Because at 2.2 those CPUs are nowhere near 45W, they hit it around 2.9 though, which strengthens my belief that they all come from the same wafers and are just stamped differently for marketing reasons, with the i9 designation reserved for the ones having no errors in whole cache area.

 

[Howard2k]

They forgot to update the popups. I guess. Because at 2.2 those CPUs are nowhere near 45W, they hit it around 2.9 though, which strengthens my belief that they all come from the same wafers and are just stamped differently for marketing reasons, with the i9 designation reserved for the ones having no errors in whole cache area.

So Intel got their own processor specifications wrong? In both the popups and the data sheet?
Looking at the data sheets I'm just not finding that.

Isn't the more likely scenario that Intel is getting better yields that advertised on 14nm and just marketing them as lower end processors (and pricing them as such) even though the silicone is more capable? So Intel is presenting a particular CPU as a 2.2GHz base and marketing it as such, though the silicone happens to give a better return.

 

[Queen6]

How come you didn't mention the base clock at all? The base clock is 2.2Ghz. Anything above that is managed by the processor based on the thermal conditions isn't it? I could be wrong, but I don't think there is anything in the processor data sheet that shows that the processor should sustain 3.6GHz across all six cores. The specs do mention a pretty interesting methodology for managing the cooling all the same. Well worth a look, and maybe I missed it in there somewhere.

This is how they described their Turbo technology:

Compared with previous generation products, Intel Turbo Boost Technology 2.0 will increase the ratio of application power towards TDP and also allows to increase power above TDP as high as PL2 for short periods of time. Thus, thermal solutions and platform cooling that are designed to less than thermal design guidance might experience thermal and performance issues since more applications will tend to run at the maximum power limit for significant periods of time.
​

TDP, which is elsewhere defined as:

Thermal Design Power (TDP) represents the average power, in watts, the processor dissipates when operating at Base Frequency with all cores active under an Intel-defined, high-complexity workload. Refer to Datasheet for thermal solution requirements.​

Which is 45W.

So unless I'm misunderstanding, above 2.2Ghz, all bets are off. Sure you can define your own limits like "3.6Ghz, sustained" but I think that's purely a personal decision right?


Meaning that it's up to the vendor to make the cooling solution as they see fit. It better maintain 2.2GHz, but above that all bets are off. Again, unless I'm misunderstanding.

Context being Turbo frequencies Correct 3.6GHz for the 8750H is shall we say a best guess as holding maximum Turbo is unlikely given the typical 90W/45W PL2, PL1 power limits. Only notebook that I'm aware of that can sustain the i9 is a massive Asus with two power supplies

The TDP is now at the base frequency or Base TDP in this case 2.2GHz, so OEM's should be fully aware that the Base TDP of 45W is a minimal not the maximal.

Intel Turbo Boost Technology 2.0 Frequency

To determine the highest performance frequency amongst active processor IA cores, the processor takes the following into consideration:

• The number of processor IA cores operating in the C0 state.
• The estimated processor IA core current consumption and ICCMax register settings.
• The estimated package prior and present power consumption and turbo power limits.
• The package temperature.
• Sustained turbo residencies at high voltages and temperature.

Any of these factors can affect the maximum frequency for a given workload. If the power, current, Voltage or thermal limit is reached, the processor will automatically reduce the frequency to stay within the PL1 value. Turbo processor frequencies are only active if the operating system is requesting the P0 state. If turbo frequencies are limited the cause is logged in IA_PERF_LIMIT_REASONS register.

Intel is indeed guaranteeing a base minimum frequency, with upper limits clock locked for the H series CPU's, what happens in the middle is very much governed by the OEM's implementation. As can be seen above there are multiple factors that govern the CPU reaching & sustaining full Turbo Boost.

Personally I'm confident Apple could have done a better job with the cooling solution, maybe not capable of sustaining full Turbo, however less likely to roll back the CPU frequency due to thermal constraints. Overall I'd personally prefer to be able to sustain at 3.6GHz than roll back to near base clock frequencies.

Q-6
[doublepost=1536147447][/doublepost]

So Intel got their own processor specifications wrong? In both the popups and the data sheet?
Looking at the data sheets I'm just not finding that.

Isn't the more likely scenario that Intel is getting better yields that advertised on 14nm and just marketing them as lower end processors (and pricing them as such) even though the silicone is more capable? So Intel is presenting a particular CPU as a 2.2GHz base and marketing it as such, though the silicone happens to give a better return.

Intel have always binned down & clock locked higher performing CPU's for numbers, hence why some CPU's like my own 8750H performs far better than average. The "Silicon Lottery" running cooler, sustaining higher clock speeds etc.

Q-6

 

[Thysanoptera]

So Intel got their own processor specifications wrong? In both the popups and the data sheet?
Looking at the data sheets I'm just not finding that.

Isn't the more likely scenario that Intel is getting better yields that advertised on 14nm and just marketing them as lower end processors (and pricing them as such) even though the silicone is more capable? So Intel is presenting a particular CPU as a 2.2GHz base and marketing it as such, though the silicone happens to give a better return.

That's exactly what I said, so I'm not sure how's your opinion any different. The comment about 'forgetting' to update spec was a snarky one, sorry if that wasn't obvious. Intel has been listing the same TDP for CPUs from the same family for a while now, this way they can create confusion and false expectations that you're getting 30% more performance/watt (which would be an equivalent of several CPU generations) and charge for it accordingly.

 

[leman]

Overall I'd personally prefer to be able to sustain at 3.6GHz than roll back to near base clock frequencies.

I think most of us would prefer that. In old-school terms, this is called "overclocking". You want a machine that is able to perma-run in an overclocked state. Which is fine, but nor really within the official spec. I think we are a bit lost in semantics here. In that aspect, no, you probably won't see Apple running their CPUs at 60+ watts since its agains their vision for portable computers (which has been unchanged for last 12+ years).

 

[Queen6]

I think most of us would prefer that. In old-school terms, this is called "overclocking". You want a machine that is able to perma-run in an overclocked state. Which is fine, but nor really within the official spec. I think we are a bit lost in semantics here. In that aspect, no, you probably won't see Apple running their CPUs at 60+ watts since its agains their vision for portable computers (which has been unchanged for last 12+ years).

Yes and no as Intel supports the feature and it's reasonably attainable in the PL1 window of 45W as long as the cooling solution is adequate, nor by default does this require a notebook to be 1"thick.

The i7 is not overclockable being H series and clock locked, the i9 is overclockable as the chip is unlocked. Your tending to fall back on narratives that support your argument of which there really isn't one, as the performance speaks for itself with the 2018 MBP not having adequate cooling to offer sustained performance at higher levels.

Right now unfortunately it's deadlock as Apple wont budge and Intel is struggling to meet it's timelines. If Apple solves this in 2019, I'm certain once again as per the 32Gb debate the narrative will reverse once again, after all who cares about being 20%-25% slower

Q-6

 

[leman]

Yes and no as Intel supports the feature and it's reasonably attainable in the PL1 window of 45W as long as the cooling solution is adequate

I don't understand this since my laptop can't really hit frequencies of 3.6 Ghz sustained at 45W continuous consumption (which the MBP can comfortably dissipate).

The i7 is not overclockable being H series and clock locked, the i9 is overclockable as the chip is unlocked.

The "narrative" has changed with turbo boost. Its nothing else but dynamic overlooking that is built into the CPU. But as with old-school overclocking, the coming things is that there are no guarantees. That is what Intel states explicitly. The only thing you can rely on are stock clocks. If your particular CPU happens not to hit the desired 3.6Ghz even with good cooling, the product is not defective.

To be honest, I don't really think that you and me are in disagreement. Its just that we look at things from different angles and offer different interpretations. Your point is about the actual real attainable performance of these chips. My point is about how they are marketed and what expectations arise from this.

 

[Queen6]

I don't understand this since my laptop can't really hit frequencies of 3.6 Ghz sustained at 45W continuous consumption (which the MBP can comfortably dissipate).

I think this is likely to be a factor of cooling and/or the CPU Core & Cache voltage which Apple has now locked since 2016 I rather doubt that Apple is applying anything other than Intel's stock voltages, which are proving to be very much on the high side for the hex core chips.

Best to explain is on my Asus the CPU is power limited to 45W on PL1, however the cooling system obviously has far more headroom. Your i9 may simply be pulling more power and generating more heat saturating the 45W cooling solution far quicker. From what I've read most i9's PL1 is 52W, so not being able to hold say a sustained clock of around 3.6GHz would not surprise me. The MBP is not in isolation by any means with many Windows notebooks suffering similar issues due to poor design.

On the Windows side there are at least solutions for some hardware with sensible usage of ThrottleStop (Intel XTU is too flaky, it although can unlock a few more CPU registers related to PL1 & PL2). For the MBP your stuck only being able to reduce the CPU's power limit with Volta, which equally reduces performance and importantly for some temperature, well until Apple gets round to locking those MSR's

The "narrative" has changed with turbo boost. Its nothing else but dynamic overlooking that is built into the CPU. But as with old-school overclocking, the coming things is that there are no guarantees. That is what Intel states explicitly. The only thing you can rely on are stock clocks. If your particular CPU happens not to hit the desired 3.6Ghz even with good cooling, the product is not defective.

To be honest, I don't really think that you and me are in disagreement. Its just that we look at things from different angles and offer different interpretations. Your point is about the actual real attainable performance of these chips. My point is about how they are marketed and what expectations arise from this.

Agree just a matter of perspective, and given how the CPU's are produced and binned you can see why Intel is reluctant to present certain data points with every chip being literally unique. Frankly we need AMD to challenge Intel for us to start seeing significant progress and break the monopoly.

Q-6

 

[leman]

Your i9 may simply be pulling more power and generating more heat saturating the 45W cooling solution far quicker. From what I've read most i9's PL1 is 52W, so not being able to hold say a sustained clock of around 3.6GHz would not surprise me

Just a quick note: I did check the registers back then and if I remember correctly, both PL1 and PL2 were set to 100 watts. Not that it matters, as Apple seems to use their own independent system for controlling power delivery to the CPU...

BTW, I did some quick tests and it does seem that at least in my machine, the heatsink is not able to dissipate the heat quickly enough. In a single core 100% load scenario, the CPU consumes close to 30 watts, but does hit the 100C ceiling without attaining the max boost (it moves around 4.3-4.5 Ghz). I interpret this as CPU generating heat quicker then it can be transferred to the heat exchanger (but I might be wrong). If so, a vapour chamber might indeed help Or its again the case of badly applied thermal compound...

 

[Howard2k]

Apple hardware is and always will be my favorite hardware!

No offence, but this is crazy. If the next MBPro has an entire keyboard made of "touchbar" it will still be your favourite? What if it the next iPhone drops the battery life by 4 hours in order to make it thinner?

Why not make them work for your money?
[doublepost=1536168802][/doublepost]

That's exactly what I said, so I'm not sure how's your opinion any different. The comment about 'forgetting' to update spec was a snarky one, sorry if that wasn't obvious. Intel has been listing the same TDP for CPUs from the same family for a while now, this way they can create confusion and false expectations that you're getting 30% more performance/watt (which would be an equivalent of several CPU generations) and charge for it accordingly.

I think the difference is that I'm saying that Intel proposes the base specifications (say 2.2GHz for a particular unit) and says "thermals allowing, you can go above that".

But you and Q-6 seem to be saying that the base should not be 2.2GHz, since the silicone is far more capable, but something like 3.6GHz or whatever. However, that's not how Intel is marketing and selling the ship.

Or maybe I am misunderstanding.

IMO That's a marketing and positioning issue for the laptop manufacturer, not a technical issue for Intel. Does it make sense for Intel to market the chips this way? Yeah, I suspect it does since it allows them to get the chips in lower end machines at a lower price point but still up-charge for more demanding users. Sure the silicone perhaps practically identical, but that's ok.

And the obvious problem that we're all seeing is that sure the chip may support a consistent 3.6GHz, but the chassis won't. It'll only support a consistent 2.2GHz or whatever. Maybe more than that, but at the minimum it supports the base clock speed.

I don't think we're materially disconnected, but I think perhaps you're looking at it more from a technical aspect and missing the market positioning issue.

 

[Queen6]

No offence, but this is crazy. If the next MBPro has an entire keyboard made of "touchbar" it will still be your favourite? What if it the next iPhone drops the battery life by 4 hours in order to make it thinner?

Why not make them work for your money?
[doublepost=1536168802][/doublepost]

I think the difference is that I'm saying that Intel proposes the base specifications (say 2.2GHz for a particular unit) and says "thermals allowing, you can go above that".

But you and Q-6 seem to be saying that the base should not be 2.2GHz, since the silicone is far more capable, but something like 3.6GHz or whatever. However, that's not how Intel is marketing and selling the ship.

Or maybe I am misunderstanding.

IMO That's a marketing and positioning issue for the laptop manufacturer, not a technical issue for Intel. Does it make sense for Intel to market the chips this way? Yeah, I suspect it does since it allows them to get the chips in lower end machines at a lower price point but still up-charge for more demanding users. Sure the silicone perhaps practically identical, but that's ok.

And the obvious problem that we're all seeing is that sure the chip may support a consistent 3.6GHz, but the chassis won't. It'll only support a consistent 2.2GHz or whatever. Maybe more than that, but at the minimum it supports the base clock speed.

I don't think we're materially disconnected, but I think perhaps you're looking at it more from a technical aspect and missing the market positioning issue.

The base CPU frequency is indeed 2.2GHz for the 8750H, and optimally it can Turbo Boost up to 3.9GHz on all six core, however unless all the factors below line up you will not reach or sustain maximum boost. There's no time limitation, although PL1 & PL2 come into play, rather more the OEM's implementation in the case of MBP it simply doesn't have the thermal headroom.

Intel Turbo Boost Technology 2.0 Frequency

To determine the highest performance frequency amongst active processor IA cores, the processor takes the following into consideration:

• The number of processor IA cores operating in the C0 state.
• The estimated processor IA core current consumption and ICCMax register settings.
• The estimated package prior and present power consumption and turbo power limits.
• The package temperature.
• Sustained turbo residencies at high voltages and temperature.

Any of these factors can affect the maximum frequency for a given workload. If the power, current, Voltage or thermal limit is reached, the processor will automatically reduce the frequency to stay within the PL1 value. Turbo processor frequencies are only active if the operating system is requesting the P0 state. If turbo frequencies are limited the cause is logged in IA_PERF_LIMIT_REASONS register.

Q-6
[doublepost=1536170903][/doublepost]

Just a quick note: I did check the registers back then and if I remember correctly, both PL1 and PL2 were set to 100 watts. Not that it matters, as Apple seems to use their own independent system for controlling power delivery to the CPU...

BTW, I did some quick tests and it does seem that at least in my machine, the heatsink is not able to dissipate the heat quickly enough. In a single core 100% load scenario, the CPU consumes close to 30 watts, but does hit the 100C ceiling without attaining the max boost (it moves around 4.3-4.5 Ghz). I interpret this as CPU generating heat quicker then it can be transferred to the heat exchanger (but I might be wrong). If so, a vapour chamber might indeed help Or its again the case of badly applied thermal compound...

The single core performance looks pretty solid, although I doubt max frequency is possible as that additional boost is very tightly controlled by the thermals - Thermal Velocity Boost (TVB) if I'm not mistaken, allowing up to 4.8GHz without deliberately overclocking the processor.

I believe it's a limitation of the design, a more efficient thermal compound will likely help, although no miracles. Personally speaking Apple should have bolstered the cooling solution to lessen the thermal impact of the 8th Gen CPU's. I find it rather naïve to think that the 2017 cooling solution would cope with faster, more aggressive silicon that clearly has a far higher TDP under heavy loads.

Apple like all is governed by the laws of thermodynamics, equally I think they need to try harder on this one. This not a leading question more curiosity; how much faster is your 2018 over the previous model? Given that one held all four cores at maximum Turbo Boost.

For myself is was very significant uplift, using something that is very generic and available to all CineBench R15 results jumped from over 750CB - (7700HQ) to over 1250CB (8750H), with both notebooks being in the same class, although the newer is a good deal smaller & lighter.

Q-6

 

[leman]

This not a leading question more curiosity; how much faster is your 2018 over the previous model? Given that one held all four cores at maximum Turbo Boost.

About 20-25% faster compared to my previous machine (high-end 2016 MBP) on CPU-only workflows relevant to me. So a huge upgrade.

 

[Queen6]

About 20-25% faster compared to my previous machine (high-end 2016 MBP) on CPU-only workflows relevant to me. So a huge upgrade.

A decent uplift, In certain specific workflow's as much as 40% with the 8750H, admittedly the system is also running 2666 RAM which has an impact, although it minor.

Q-6

 

[Newtons Apple]

No offence, but this is crazy. If the next MBPro has an entire keyboard made of "touchbar" it will still be your favourite? What if it the next iPhone drops the battery life by 4 hours in order to make it thinner?

Why not make them work for your money?

I can't see Apple doing any of your "what ifs" but if the time comes when they make me unhappy with what they offer they will no longer be my favorite.

Loved all my hardware other than the 2016 MBPro which after three keyboard changes, they refunded my money in full. Can't be unhappy with that.

Apple does fine "working for it's money",

 

[Howard2k]

I can't see Apple doing any of your "what ifs" but if the time comes when they make me unhappy with what they offer they will no longer be my favorite.

Loved all my hardware other than the 2016 MBPro which after three keyboard changes, they refunded my money in full. Can't be unhappy with that.

Apple does fine "working for it's money",

ok cool. That doesn't correlate with your "always will be" statement, but you're on the side of common sense so all good!