Waiting for Haswell MBP Mega Thread

[Vanilla35]

So in regards to 13" rMBPs getting 16 gigs of RAM with Haswell, is there a potential limitation due to the chipset or the fact that the machine would probably need two 8 GB sticks which may not be possible, or something along those lines?

Basically is there a physical limitation/reason that the Haswell 13" couldn't have 16 gigs of RAM or would it just be an intentional Apple gimp?

Yes. 15% chance of apple douchebagging, 85% chance it's limited in some way shape or form. i.e. physical space, or cost of switching to 8 GB sticks instead of 4 GB, in order to fit it in, etc.

----------

So who thinks that the announcement is gonna be on September 10th along with the iPhone? That's what all the reports are saying. I'm not getting my hopes upand I'm still thinking October is gonna be when, but it does make sense for the release to be earlier in September.

I would be happy if that were the case too, but I just have a feeling it's not gonna happen. Either way I'm not in a rush to buy, so doesn't matter to me

----------

11%, 3%, 9%. all of those seem pretty insignificant to me. I don't know what percent of power CPU+GPU takes up in a computer like the rMBP, but I would guess that it's much less than 50% making the difference in battery life < 5.5%, 1.5%, or 4.5%. it looks to me like there won't be much of a battery life improvement due to dropping the GPU, but there will still be Haswell related improvements because of the low-power states, and don't forget the possibility, however unlikely, that IGZO will add to those improvements

Why does everyone keep thinking IGZO is gonna happen this year

 

[mcarling]

11%, 3%, 9%. all of those seem pretty insignificant to me. I don't know what percent of power CPU+GPU takes up in a computer like the rMBP, but I would guess that it's much less than 50% making the difference in battery life < 5.5%, 1.5%, or 4.5%. it looks to me like there won't be much of a battery life improvement due to dropping the GPU, but there will still be Haswell related improvements because of the low-power states, and don't forget the possibility, however unlikely, that IGZO will add to those improvements

The factor that is usually underestimated is memory. A 16GB system uses considerably more power than an 8GB system (assuming twice as many of the same memory chips). That would probably be at least as big a factor in power consumption than integrated versus discrete GPU.

 

[bill-p]

Why does everyone keep thinking IGZO is gonna happen this year

Because IGZO doesn't cost Apple a fortune to buy... Oh wait...

Seriously, though, it's because if they don't do anything to that screen, improving battery life on the next rMBP is harder than trying to squeeze an elephant through a pin hole.

And if they don't give us better battery life, then... what good is Iris Pro? Higher OpenCL performance for very select few people, same CPU performance, worse 3D performance?

The factor that is usually underestimated is memory. A 16GB system uses considerably more power than an 8GB system (assuming twice as many of the same memory chips). That would probably be at least as big a factor in power consumption than integrated versus discrete GPU.

Wait... what!!?

Um... I think you may want to double check that statement.

 

[koyoot]

Are you saying the Haswell rMBP will include a dGPU?

No. Im Saying only that Iris Pro Will not increase battery life. And in case of performance - it still can be weaker than GT650M.

In two of the three tests, the i7 HD4600 + GT750M consumed substantially more power than the Iris Pro CPU. In the remaining test, the Iris Pro CPU consumed insignificantly more.


TDP and power draw are not exactly the same but are closely related. The integral of both, over a long time, is the same. TDP is essentially a rolling average of recent power draw. Power draw can fluctuate very quickly, for example between 1W and 90W, for any period of time during which TDP may remain in, for example, the 45W to 47W range.

That doesnt change the fact that Iris Pro WILL NOT increase battery Life.

TDP is Way different than Power Draw, even if they are related. TDP indicates what type of thermals company must use to cool down the CPU. Thats All. It doesnt indicate how many Watts it draws. Because - like i said before, CPU can draw 90W of power and fit 47W TDP. Understand?


And again. Iris Pro will not increase Battery life. Low power states of CPU will.

 

[Vanilla35]

Because IGZO doesn't cost Apple a fortune to buy... Oh wait...

Seriously, though, it's because if they don't do anything to that screen, improving battery life on the next rMBP is harder than trying to squeeze an elephant through a pin hole.

And if they don't give us better battery life, then... what good is Iris Pro? Higher OpenCL performance for very select few people, same CPU performance, worse 3D performance?

I feel like they'll be able to squeeze 3-4 hours out of their ass somehow without IGZO. And they're not about to add IGZO to the line-up considering it's current notoriety. I can definitely see it for next year though if that is their intention.

And regarding that last part, you're just assuming Apple has nothing else up their sleeve haha. But yeah, I get your point

 

[APPLE-FANATIC]

If Apple has held off this long with updating their MacBook pro line, Thunderbolt and Cinema displays, there has to be some big changes coming. Im not too sure about the IGZO displays, but Apple has been investigating this since the iPad 3, but due to them being unable to mass produce these displays, is has yet to occur. I will say, this would be a real nice upgrade for the MacBook Pro line if they were to include IGZO+ Haswell, I honestly dont understand the delay of a "refresh" by Apple. If nothing significant was going on behind the scenes, I think there would have been an announcement at WWDC. Just my opinion

Early report about IGZO and Apple:

http://appleinsider.com/articles/13...isplays-apple-called-a-prime-candidate-to-use

 

[mcarling]

TDP is Way different than Power Draw, even if they are related. TDP indicates what type of thermals company must use to cool down the CPU. Thats All. It doesnt indicate how many Watts it draws. Because - like i said before, CPU can draw 90W of power and fit 47W TDP. Understand?

It's clear that you don't understand. Every joule (one watt for one second) in power consumed must be dissipated in heat. There are no other sources of heat so, over time, power draw and thermal dissipation must match. The only reason they are not necessarily exactly equal at any moment is that thermal dissipation can lag power draw in time (in which case internal parts are rising in temperature) but that can happen only in the short term (because temperatures can rise only within limits).

 

[kaellar]

No. Im Saying only that Iris Pro Will not increase battery life. And in case of performance - it still can be weaker than GT650M.



That doesnt change the fact that Iris Pro WILL NOT increase battery Life.

TDP is Way different than Power Draw, even if they are related. TDP indicates what type of thermals company must use to cool down the CPU. Thats All. It doesnt indicate how many Watts it draws. Because - like i said before, CPU can draw 90W of power and fit 47W TDP. Understand?


And again. Iris Pro will not increase Battery life. Low power states of CPU will.

I'll just wait for more comprehensive and reliable review to build my statements upon it, and will do it when rMBPs arrive. From my POV that German article either miscalculates somewhere or simply takes the numbers from the ceiling.

And why is everyone so hung on the battery life? Sure it will be improved because of low-power CPU states. But the most important reason of using Iris Pro is lowering the requirements for cooling system. Current gen rMBP barely handles the proper load, and if both CPU and GPU are fully loaded can even throttle while being hot as hell. Reducing the TDP by almost two times will show its benefits here for sure.

 

[bill-p]

It's clear that you don't understand. Every joule (one watt for one second) in power consumed must be dissipated in heat. There are no other sources of heat so, over time, power draw and thermal dissipation must match. The only reason they are not necessarily exactly equal at any moment is that thermal dissipation can lag power draw in time (in which case internal parts are rising in temperature) but that can happen only in the short term (because temperatures can rise only within limits).

You're the one that doesn't understand.

The rMBP isn't a space heater. So not every watt of power consumed must be dissipated.

A part of the power consumed is converted to heat, but everything else must be used to move things around the machine. A computer is basically a mechanical rig of insanely small parts, and you need energy to move those parts around. Whatever energy is used to do those tasks would eventually be absorbed or "swallowed up" in the process. So it won't come back out as heat at all.

And even a space heater would consume more power than the amount of heat it pushes out due to inefficiencies in the process of power conversion.

Look that up.

 

[mcarling]

You're the one that doesn't understand.

The rMBP isn't a space heater. So not every watt of power consumed must be dissipated.

A part of the power consumed is converted to heat, but everything else must be used to move things around the machine. A computer is basically a mechanical rig of insanely small parts, and you need energy to move those parts around. Whatever energy is used to do those tasks would eventually be absorbed or "swallowed up" in the process. So it won't come back out as heat at all.

And even a space heater would consume more power than the amount of heat it pushes out due to inefficiencies in the process of power conversion.

Look that up.

The only things in a rMBP that are not solid-state are the fan and the speakers, which consume an absolutely insignificant portion of total power. Except those and the light produced by the display, all power consumed by a rMBP is converted into heat. The discussion was about chips, for which exactly all the power consumed is converted into heat.

 

[bill-p]

The only things in a rMBP that are not solid-state are the fan and the speakers, which consume an absolutely insignificant portion of total power. Except those and the light produced by the display, all power consumed by a rMBP is converted into heat. The discussion was about chips, for which exactly all the power consumed is converted into heat.

Wow... I think you may want to start reading up on the subject... and look at more data before you make statements like these.

Seriously, I'm not kidding.

 

[mcarling]

Wow... I think you may want to start reading up on the subject... and look at more data before you make statements like these.

Seriously, I'm not kidding.

I used to write circuit simulation software for the CPU designers at Intel. We periodically produced silicon to calibrate our simulations, so I am intimately familiar with such data. You may not be kidding, but you're dead wrong. The reason is obvious to anyone who understands the principle of conservation of energy.

 

[CmdrLaForge]

PowerbookG5!

I still remember this user. He was very active around 2005/2006.

 

[hellric]

French site Macgeneration says nvidia 755M GT In rmbp 15, from some code found in Mavericks...

 

[mcarling]

French site Macgeneration says nvidia 755M GT In rmbp 15, from some code found in Mavericks...

You mean this article?
http://www.macg.co/news/voir/260977/os-x-10.9-indiscret-sur-les-cartes-graphiques-des-futurs-mac

It doesn't say that. It says that the current 15" rMBP uses an Nvidia chip in addition to the integrated graphics and that the successor to the Nvidia chip in the current 15" rMBP is the 755M. The paragraph that talks about the graphics drivers in Mavericks only mentions the Intel 4000 and 5000 series HD integrated graphics.

 

[GSPice]

I used to write circuit simulation software for the CPU designers at Intel. We periodically produced silicon to calibrate our simulations, so I am intimately familiar with such data. You may not be kidding, but you're dead wrong. The reason is obvious to anyone who understands the principle of conservation of energy.

I'm an engineer.. I think people are getting consumption and dissipation confused.. and that energy is *universally* conserved, but in a specific system, ie a laptop, energy is lost in heat, etc.

 

[mcarling]

I'm an engineer.. I think people are getting consumption and dissipation confused.. and that energy is *universally* conserved, but in a specific system, ie a laptop, energy is lost in heat, etc.

I'm not sure what the source of the confusion is. For any system, such as a laptop, that is not converting energy into mass (via either a chemical or nuclear reaction), the energy that goes in must eventual equal the energy that comes out. The energy that goes in is in the form of electricity. In the case of a rMBP, the energy that comes out is mainly in the form of heat, secondarily in the form of light from the display (and the backlit keyboard), and tertiarily in the form of kinetic energy (airflow caused by the fans), sound pressure waves caused by the speakers, and little electromagnetic radiation from the Wifi antenna. That's all there is unless there are devices connected which draw power from the laptop.

 

[bill-p]

I used to write circuit simulation software for the CPU designers at Intel. We periodically produced silicon to calibrate our simulations, so I am intimately familiar with such data. You may not be kidding, but you're dead wrong. The reason is obvious to anyone who understands the principle of conservation of energy.

So you're saying... the CPU takes in energy and releases all of that energy out as heat?

How does it transmit information then, if not by using a fraction of that energy to send that information to other components on the board?

No, please, enlighten me. How does it send information if it doesn't have any energy to do so? Because as you said, that energy would be converted to heat completely.

In fact, why are these things even called CPU? They don't do anything other than taking in energy and converting all of that energy to heat.

 

[sofianito]

My understanding is that the TDP is the energy needed to dissipate the heat generated so that every component functions properly. The power draw is the energy needed by CPU, GPU, and other components to perform their job...

----------

I'm an engineer.. I think people are getting consumption and dissipation confused.. and that energy is *universally* conserved, but in a specific system, ie a laptop, energy is lost in heat, etc.

Yeah, Lavoisier law

The rule of thumb is that components need energy (Power draw) to perform tasks, but when performing tasks they generate heat. If the heat is not dissipated components will likely melt and stop functioning properly. For that, energy is needed to dissipate that heat (TDP).

 

[Vanilla35]

You mean this article?
http://www.macg.co/news/voir/260977/os-x-10.9-indiscret-sur-les-cartes-graphiques-des-futurs-mac

It doesn't say that. It says that the current 15" rMBP uses an Nvidia chip in addition to the integrated graphics and that the successor to the Nvidia chip in the current 15" rMBP is the 755M. The paragraph that talks about the graphics drivers in Mavericks only mentions the Intel 4000 and 5000 series HD integrated graphics.

Actually, it says pretty plainly that the 15" rMBP "should" get the 755M. That is their opinion of course, but that's what it says.

"The 15-inch also has a graphics chip from NVIDIA and the new model expected to fall should include the GeForce GT 755m , the successor to the current chip." (Google translation)

Unless you know french or something and this has been translated wrong, the article says that they think the successor will come with the 755M.

 

[sofianito]

Actually, it says pretty plainly that the 15" rMBP "should" get the 755M. That is their opinion of course, but that's what it says.

"The 15-inch also has a graphics chip from NVIDIA and the new model expected to fall should include the GeForce GT 755m , the successor to the current chip." (Google translation)

Unless you know french or something and this has been translated wrong, the article says that they think the successor will come with the 755M.

It says the 15" will have the 755M this fall...

 

[kaellar]

the article says that they think the successor will come with the 755M.

.. which has no difference from the GPU in current gen. Pretty lame, if you ask me. A TDP just 5w more and you could put 760m into it.

 

[Vanilla35]

.. which has no difference from the GPU in current gen. Pretty lame, if you ask me. A TDP just 5w more and you could put 760m into it.

I doubt their assumptions have any clout behind them. They're just kind of assuming it will have a bumped dGPU. And that's the card that comes directly after (since it's a refresh of the same card basically). I don't know Apple's history regarding their cards, but the 755M came out in June, which is really new. Does Apple usually do that with their cards? Or do they use more mature ones? I'd be happy with this card, as it has more memory bandwidth and higher clockspeed. The 650M is enough power as it is, and the case can only handle so much, so faster clockspeed and bandwidth is as good as it's gonna get really

 

[thundersteele]

So you're saying... the CPU takes in energy and releases all of that energy out as heat?

How does it transmit information then, if not by using a fraction of that energy to send that information to other components on the board?

No, please, enlighten me. How does it send information if it doesn't have any energy to do so? Because as you said, that energy would be converted to heat completely.

In fact, why are these things even called CPU? They don't do anything other than taking in energy and converting all of that energy to heat.

Yeah, this is correct. In the end computers are smart space heaters. All the energy that goes in has to come out in some form. Since information is not a form of energy, it all goes into heat (except for the small fraction that radiates away as light, and maybe sound waves and EM waves from WiFi etc).

 

[mcarling]

Unless you know french or something and this has been translated wrong, the article says that they think the successor will come with the 755M.

I did graduate studies at the University of Paris 1 Pantheon-Sorbonne. My French is sufficient to read the article. The author of the article speculates that the Haswell rMBP will get the 755M, but most certainly does not claim that the drivers in Mavericks so indicate. The author doesn't make any claim to support that belief other than saying that the 755M is the successor to the Nvidia chip in the current rMBP.

----------

So you're saying... the CPU takes in energy and releases all of that energy out as heat?

How does it transmit information then, if not by using a fraction of that energy to send that information to other components on the board?

No, please, enlighten me. How does it send information if it doesn't have any energy to do so? Because as you said, that energy would be converted to heat completely.

In fact, why are these things even called CPU? They don't do anything other than taking in energy and converting all of that energy to heat.

The CPU and surrounding chips communicate by setting the voltage on interconnecting lines to 0V (to represent a zero) and 1.2V (to represent a one). All, yes all, of the energy needed to do this eventually dissipates as heat. Regardless, the energy consumed by these interconnect lines is a tiny tiny fraction of 1% of the energy consumed by and dissipated by the chips themselves.