Waiting for Broadwell MBP Thread

[Eauboy]

Future proofing for 4K displays

I'm currently running a 2009 MBP, which is really starting to shows its age. I know that I could replace it with one of the 2014 rMBPs and get a huge performance increase as well as a nicer display.

My question though centers around future proofing for purposes of external displays. I expect to buy a 4K display within a year or two. If between now and then Apple and the other display manufacturers begin to deliver 5K displays I would certainly give those a look. From what I've read, it sounds like the current rMBPs can handle 4K displays but only barely, and don't handle 5K displays at all.

Am I correct that it's a bit of a gamble waiting for Broadwell? It sounds like at least a possibility that the next round of rMBPs based on broadwell won't have any better support for high density displays (in the form of the next generation connector standards) than the current generation does. I don't want to wait until Skylake, so if Broadwell doesn't buy me anything in this particular department I might as well buy one of the current models.

I welcome your thoughts.

 

[iKrivetko]

It's fairly likely that there will be no Broadwell in the first place.

 

[mcarling]

My question though centers around future proofing for purposes of external displays. I expect to buy a 4K display within a year or two. If between now and then Apple and the other display manufacturers begin to deliver 5K displays I would certainly give those a look. From what I've read, it sounds like the current rMBPs can handle 4K displays but only barely, and don't handle 5K displays at all.

Am I correct that it's a bit of a gamble waiting for Broadwell? It sounds like at least a possibility that the next round of rMBPs based on broadwell won't have any better support for high density displays (in the form of the next generation connector standards) than the current generation does. I don't want to wait until Skylake, so if Broadwell doesn't buy me anything in this particular department I might as well buy one of the current models.

I welcome your thoughts.

If I were betting, I'd give it about a 10-20% chance that Apple will introduce Displayport 1.3 and Thunderbolt 3 to the MBP with Broadwell. About 80% chance with Skylake.

Based on your needs, I suggest you buy now.

 

[MareLuce]

steppings?

The 14nm process turned out to be more difficult than expected. However, the steppings for Broadwell are ready, so Intel will roll out Broadwell chips as fast as they can rebuild their fabs for the 14nm process. The steppings for Skylake are not ready yet. It is possible that some of the steppings for Skylake might be validated while the 14nm rollout is still underway, in which case Intel might skip some Broadwell parts and go directly to Skylake, but I don't expect that. It will be easy to make both Broadwell and Skylake parts in the same fab. I'm not aware of any technical or business reason why Skylake couldn't follow Broadwell by four to six months.

Could you elaborate on what you mean by "steppings"?

 

[mcarling]

Could you elaborate on what you mean by "steppings"?

http://www.cpu-world.com/Glossary/C/Core_stepping.html

 

[Gildarts]

If I were betting, I'd give it about a 10-20% chance that Apple will introduce Displayport 1.3 and Thunderbolt 3 to the MBP with Broadwell. About 80% chance with Skylake.

Based on your needs, I suggest you buy now.

Thunderbolt 3 is definitely not coming to Broadwell.

See this link.

 

[mcarling]

Thunderbolt 3 is definitely not coming to Broadwell.

Intel are definitely not including Thunderbolt 3 support in the Broadwell CPUs. That does not mean Apple cannot provide Thunderbolt 3 support in the Broadwell MBPs. If that's not obvious, consider that there will be Thunderbolt 3 devices which do not contain an Intel CPU.

 

[Gildarts]

Let's put it another way: I prefer Broadwell over Haswell and I would prefer Cannonlake over Skylake. My current machine has an Ivy Bridge CPU. I prefer to buy on the Tick cycle when the design bugs have all been worked out. Thunderbolt 3 does not require any support in the CPU. It could be implemented (at significantly higher cost) with a Core2Duo.

As far as I recall Ivy Bridge was 22nm, Haswell was 28nm and Broadwell is 14nm. Ivy bridge had no problems. Haswell had no problems. Broadwell is having a lot problems to the point where it's late.

Intel has stated Skylake, which like Broadwell will also be 14nm, to be on schedule and problem free.

Considering they had to shrink the design from Haswell 28nm to Broadwell 14nm it seems that the problems with Broadwell might be related to the shrinking of the design. It's not an unreasonable assumption.

Seeing as Skylake is 14 nm and has been stated to be problem free I would go as for as to say that the design bugs were in Broadwell and Skylake will not suffer from that (Intel has already stated so).

With Cannonlake being 10nm I ought it more likely to see design bugs with Cannonlake though the difference is only small with Skylake-Cannonlake (4nm) compared to Haswell-Broadwell (14nm).

----------

Intel are definitely not including Thunderbolt 3 support in the Broadwell CPUs. That does not mean Apple cannot provide Thunderbolt 3 support in the Broadwell MBPs. If that's not obvious, consider that there will be Thunderbolt 3 devices which do not contain an Intel CPU.

If Intel's not doing it. What incentive would Apple have to do it? Like you said, it's a cost issue.

It seems cut and dry to me that if there are any Broadwell MacBooks they will not have Thunderbolt 3.

 

[Gildarts]

I don't think I would notice the difference. 802.11n has no difficulty keeping up with my Internet connection..

WiGig is said to be 3 times faster than 802.11ac en 10 times faster than 802.11n according to Intel (who is on the Board of Directors of the WiGig Alliance).

According to this wiki article WiGig can be up to 50 times faster than 802.11n.

So I'm guessing it would be noticeable but since you said you you're limited by the bandwidth speed provided by your ISP it seems this would not benefit you. But in any case, in the name of future proofing it would be a handy feature to have.

However, the wiki article does mention that the 50X speed that WiGig will provide will be compatible with existing wifi devices. So I guess that means you don't need to have it your device as long as your router has it?

Quote:

WiGig tri-band enabled devices, which operate in the 2.4, 5 and 60 GHz bands, deliver data transfer rates up to 7 Gbit/s, about as fast as an 8 antenna 802.11ac transmission, and nearly 50 times faster than the highest 802.11n rate, while maintaining compatibility with existing Wi-Fi devices.

 

[mcarling]

As far as I recall Ivy Bridge was 22nm, Haswell was 28nm and Broadwell is 14nm.

No. Westmere and Sandy Bridge used the 32nm process. Ivy Bridge and Haswell use the 22nm process. Broadwell and Skylake will use the 14nm process now being rolled out to Intel's fabs.

Ivy bridge had no problems. Haswell had no problems. Broadwell is having a lot problems to the point where it's late.

Each new process had many problems, but they are generally not publicized and they were overcome more or less according to schedule. The 14nm process took longer than expected to overcome all the problems, but it's already in production at a few Intel fabs.

Intel has stated Skylake, which like Broadwell will also be 14nm, to be on schedule and problem free.

Considering they had to shrink the design from Haswell 28nm)0 to Broadwell 14nm it seems that the problems with Broadwell might be related to the shrinking of the design.

Seeing as Skylake is 14 nm and has been stated to be problem free I would go as for as to say that the design bugs were in Broadwell and Skylake will not suffer from that (according to Intel itself).

That is the crux of your misunderstanding. There were no design-related delays with Broadwell. The design differences between Haswell and Broadwell are limited to two types: a) adding more of stuff: more EUs, more cache, etc. to improve performance, and b) fixing design bugs (errata) in the Haswell microarchitecture. All the reason why Broadwell was delayed is due to the 14nm production process, not the design. The reason why the delays affect Broadwell but not Skylake (which use the same 14nm process) is that the process was ready after Intel planned to start producing Broadwell parts but still well before Intel planned to start producing Skylake parts.

With Cannonlake being 10nm I ought it more likely to see design bugs with Cannonlake though the difference is only small with Skylake-Cannonlake (4nm) compared to Haswell-Broadwell (14nm).

Don't think about the process shrinks along a linear scale. Think about them along a logarithmic scale. Each process shrink is roughly a factor of the square root of two. One reason why they are becoming more difficult is that they are now approaching the atomic scale. Obviously, it's not possible (using anything like current ICs) to make transistors smaller than one atom wide. Even several atoms wide involves quantum effects that were not an issue for past process shrinks.

If Intel's not doing it. What incentive would Apple have to do it? Like you said, it's a cost issue.

What incentive would Apple have to offer new features to their customers? Really?

It seems cut and dry to me that if there are any Broadwell MacBooks they will not have Thunderbolt 3.

You think Apple (a company that just spent over $6B in one year on R&D) cannot afford the cost of integrating a Thunderbolt 3 controller into a Broadwell system? Cash flow problems? Or you think Apple don't believe their customers want Thunderbolt 3? Or???

I'm not sure how Apple would weigh the costs and benefits here. I'm puzzled why you are certain how Apple will have chosen.

 

[Gildarts]

No. Westmere and Sandy Bridge used the 32nm process. Ivy Bridge and Haswell use the 22nm process. Broadwell and Skylake will use the 14nm process now being rolled out to Intel's fabs.


Each new process had many problems, but they are generally not publicized and they were overcome more or less according to schedule. The 14nm process took longer than expected to overcome all the problems, but it's already in production at a few Intel fabs.


That is the crux of your misunderstanding. There were no design-related delays with Broadwell. The design differences between Haswell and Broadwell are limited to two types: a) adding more of stuff: more EUs, more cache, etc. to improve performance, and b) fixing design bugs (errata) in the Haswell microarchitecture. All the reason why Broadwell was delayed is due to the 14nm production process, not the design. The reason why the delays affect Broadwell but not Skylake (which use the same 14nm process) is that the process was ready after Intel planned to start producing Broadwell parts but still well before Intel planned to start producing Skylake parts.


Don't think about the process shrinks along a linear scale. Think about them along a logarithmic scale. Each process shrink is roughly a factor of the square root of two. One reason why they are becoming more difficult is that they are now approaching the atomic scale. Obviously, it's not possible (using anything like current ICs) to make transistors smaller than one atom wide. Even several atoms wide involves quantum effects that were not an issue for past process shrinks.


What incentive would Apple have to offer new features to their customers? Really?


You think Apple (a company that just spent over $6B in one year on R&D) cannot afford the cost of integrating a Thunderbolt 3 controller into a Broadwell system? Cash flow problems? Or you think Apple don't believe their customers want Thunderbolt 3? Or???

I'm not sure how Apple would weigh the costs and benefits here. I'm puzzled why you are certain how Apple will have chosen.

Forgive me, I wrote design but I was referring (or trying) to the manufacturing process. I do not know anything about chipdesign.

I don't know where I got that Haswell was 28nm, I had checked the wiki. I might've been confused.

In any case it doesn't affect my main point. Which was that the manufacturing difficulties of Broadwell are likely due to the differences in nm between Broadwell and Haswell.
This seems to be confirmed by the fact that Skylake has not suffered any manufacturing difficulties. Broadwell was the first 14nm, they struggled to figure out how to get it right at the expense of the schedule. Now Skylake doesn't need to the same fate.

Regardig Thunderbolt 3, you might be right but then again. To me it seems unlikely. You yourself said that the chance of Broadwell having Thunderbolt 3 is about 10-20%. I just went out on a limb and said it's definitely not happening.

I don't think even 0,1% of the 6B$ spent R&D has been spent at looking how to integrate TB3 into Broadwell.

Like you said, it's a cost issue. Why would you spend R&D money on figuring out how to get it to work with Broadwell when you will basically get it on a silver platter with Skylake?

The Broadwell delay has been known for a well amount of time, since about Q4 2013. And Apple has probably knew about it way before then so.

If costumers want TB3 they can buy Skylake about 6 months after Broadwell (if they release BW MacBooks at all in 2015).

So to me, it makes no sense that they would spent R&D money figuring out how to get TB in BW when Intel could basically just tell them "Don't worry about, I got you covered with SL".

It's not that they can't afford it. It just seems to make no sense. (So I guess it's not so much a cost issue as it is an efficiency issue. It's not efficient to spen R&D trying to figure out how to get TB3 ontoBW when intel basically has it covered in SL.

Edit 1: I hadn't proof read my comment. I edited to make the text sound better.

Edit 2: Added some text.

 

[mcarling]

Forgive me, I wrote design but I was referring (or trying) to the manufacturing process. I do not know anything about chipdesign.

I don't know where I got that Haswell was 28nm, I had checked the wiki. I might've been confused.

No problem.

In any case it doesn't affect my main point. Which was that the manufacturing difficulties of Broadwell are likely due to the differences in nm between Broadwell and Haswell.

You have cause and effect reversed. The major difference between Broadwell and Haswell is the manufacturing process used, which results in the different die sizes. Let's put it this way: If I were given a set of Broadwell and Haswell designs without any reference scale for physical size, I would not be able to sort them into two subsets (one for Broadwell and one for Haswell) let alone figure out which subset would be which. On the other hand, I could easily do that for Ivy Bridge vs Haswell. Other than the manufacturing process used, the differences between Broadwell and Haswell are absolutely minor. The correction of errata (bugs) and sometimes different choices about number of EUs, cache sizes, number of cores, etc.

Seeing as Skylake has not suffered or at least not to the extent of Broadwell is because the Broadwell manufacturing process served as a learning experience for Skylake.

Not exactly. It's not that Intel broadly learned about 14nm manufacturing processes with Broadwell and so will have an easier time developing 14nm manufacturing processes for Skylake. Not at all. They will be built with the same 14nm process.

Regardig Thunderbolt 3, you might be right but then again. To me it seems unlikely. You yourself said that the chance of Broadwell having Thunderbolt 3 is about 10-20%. I just went out on a limb and said it's definitely not happening.

I don't think even 0,1% of the 6B$ spent R&D has been spent at looking how to integrate TB3 into Broadwell.

Like you said, it's a cost issue. Why would you spend R&D money on figuring out how to get it to work with Broadwell when you will basically get it on a silver platter with Skylake?

The Broadwell delay has been known for a well amount of time, since about Q4 2013. And Apple has probably knew about it way before then so.

If costumers want TB3 they can buy Skylake about 6 months after Broadwell (if they release BW MacBooks at all in 2015).

So to me, it makes no sense that they would spent R&D money figuring out how to get TB in BW when Intel could basically just tell them "Don't worry about, I got you covered with SL".

It's not that they can't afford it. It just seems to make no sense. (So I guess it's not so much a cost issue as it is an efficiency issue. It's not efficient to spen R&D trying to figure out how to get TB3 ontoBW when intel basically has it covered in SL.

Do you have any sense of the engineering involved to make a Thunderbolt 3 controller work with Broadwell versus with Skylake? Can you tell us what the major difference(s) would be? In other words, what specific engineering challenge(s) would Apple face integrating a Thunderbolt 3 controller into a Broadwell MBP that they would not face with a Skylake MBP?

 

[Gildarts]

You have cause and effect reversed. The major difference between Broadwell and Haswell is the manufacturing process used, which results in the different die sizes. Let's put it this way: If I were given a set of Broadwell and Haswell designs without any reference scale for physical size, I would not be able to sort them into two subsets (one for Broadwell and one for Haswell) let alone figure out which subset would be which. On the other hand, I could easily do that for Ivy Bridge vs Haswell. Other than the manufacturing process used, the differences between Broadwell and Haswell are absolutely minor. The correction of errata (bugs) and sometimes different choices about number of EUs, cache sizes, number of cores, etc.

I understand this explanation. This is exactly what I've been saying (trying to say). You must forgive me. English is not my native language and I may not always instantly understand that a word is misused or gives different meaning/context to what I'm saying (or trying to say).

But basically what I've been trying to say for the last few comments is that the difficulties of Broadwell are caused by the differences in the manufacturing processes between Haswell and Broadwell.

Not exactly. It's not that Intel broadly learned about 14nm manufacturing processes with Broadwell and so will have an easier time developing 14nm manufacturing processes for Skylake. Not at all. They will be built with the same 14nm process.

I get that. The point remains that the process was new (or different) when manufacturing Broadwell (as opposed to Haswell) and any kinks that may have recurred have been resolved which will give Skylake a problem free ride through the same process.

Do you have any sense of the engineering involved to make a Thunderbolt 3 controller work with Broadwell versus with Skylake? Can you tell us what the major difference(s) would be? In other words, what specific engineering challenge(s) would Apple face integrating a Thunderbolt 3 controller into a Broadwell MBP that they would not face with a Skylake MBP?

If it weren't clear already I have absolutely no engineering sense of TB3. That's not the point though. From a logical business perspective it doesn't make sense for Apple to invest any effort in how to get TB3 to work in BW.

The reason for this lies in the fact that Apple already knows, and has known, that there is a solution ready for the next generation SL. So why on earth invest any time in figuring out how to get it on BW.

I don't understand why you keep returning to the TB3 stuff when you yourself have acknowledged that it's very unlikely, estimating the chance of this happening at 10-20%.

Edit: Took out some spelling errors.

 

[MacSumo]

Why doesn't intel just skip broadwell and go straight to skylake? Broadwell sounds like a flop.

 

[Gildarts]

Why doesn't intel just skip broadwell and go straight to skylake? Broadwell sounds like a flop.

There's a discussion about the likelihood of that happening here.

You seem to be right on the money.

 

[puercaeli]

rMBP Broadwell: more battery life focus?

I am currnetly looking at buying rMBP13 to replace my MBA 2011 and has been following Broadwell saga very closely. My main reason for waiting was better GPU promised by Broadwell graphics.

Recently Broadwell U (bound for current design MBA and rMBP ) SKU was posted by many tech sites.(below is table of rMBP bound broadwell U)

Notable differences include: upgrade to iris 6100, TDP reduction to 15W(some website has 28W but this Intel leak seems more reliable), base clock increase of 0.1Ghz, Max GPU frequency is reduced from 1200 to 1100.

Similarities include: turbo boost frequency essentially remains same.


So the main message I am getting here is that Broadwell will be similar CPU and GPU performance(as IPC increase in CPU is said to be may be 5percent, GPU may be 20 percent ) but at substantially reduced TDP. - ie longer battery and ability to sustain the higher frequency longer(if same design is maintained - god forbid apple does another iPad 2 style size reduction!)

This tells me another 6 or more month wait for next refresh may not bring that much performance increase(i am pretty happy with battery on current gen model). And I guess I could wait for more substantial Skylake update, i have a feeling that 28W/ 15W will probabaky not launch until very end of next year and only Y series will launch in 2H next year(please correct me if I am wrong).

 

[Grudderdude]

Next Macbook Pro Update + GPU (2015 Broadwell)

Soo.... I am deeply curious as to what people think will be on the inside of the next MBP 15" with dedicated GPU. I know Nvidia has announced the 970m and 980m, which claim some pretty amazing power for a mobile GPU. I bet no major upgrades will take place until the Broadwell stuff is launched, but I just want to hear what the tech community is thinking. I was going to buy the more current 15" rMBP with the 750m, but seeing that it just pushes out 30fps on low graphics at 1080p for games like Assassin's Creed Black Flag, I wanted to wait to see what the next generation of GPU's would look like. There are some pretty bad ass games coming out, like Witcher 3, Dragon Age: Inquisition, Far Cry 4...

I know that laptops are not the best thing to be gaming on, and for the price a person could build a raging desktop dedicated gaming machine, but mobile computers just fit my life better and I am a sucker for the MBP and OS. I'd love to hear what people are speculating around this. The idea of a machine that is mobile, low profile that also games well is a dream at this point. Especially in the sleek Mac form factor. Does anyone think we will get something this grandiose the coming 2015 year?

 

[Gata]

Soo.... I am deeply curious as to what people think will be on the inside of the next MBP 15" with dedicated GPU. I know Nvidia has announced the 970m and 980m, which claim some pretty amazing power for a mobile GPU. I bet no major upgrades will take place until the Broadwell stuff is launched, but I just want to hear what the tech community is thinking. I was going to buy the more current 15" rMBP with the 750m, but seeing that it just pushes out 30fps on low graphics at 1080p for games like Assassin's Creed Black Flag, I wanted to wait to see what the next generation of GPU's would look like. There are some pretty bad ass games coming out, like Witcher 3, Dragon Age: Inquisition, Far Cry 4...

I know that laptops are not the best thing to be gaming on, and for the price a person could build a raging desktop dedicated gaming machine, but mobile computers just fit my life better and I am a sucker for the MBP and OS. I'd love to hear what people are speculating around this. The idea of a machine that is mobile, low profile that also games well is a dream at this point. Especially in the sleek Mac form factor. Does anyone think we will get something this grandiose the coming 2015 year?

I hope.

With the advances in iGPU technology, it could be possible that Apple jumps for the 970, and then implements a gpu switching algorithm that more aggressively favors the integrated card, only switching once the integrated card is really hitting its limits.

The potential price though ;-;

 

[Grudderdude]

Oh barf... I know. They milk us dummies for all we are worth! I still prefer their stuff though. If they built it, I'd buy it.

 

[CmdrLaForge]

Correct. I will replace my early-2013 15" rMBP with the first MBP to support Displayport 1.3 and able to drive a 5K monitor. I don't know whether that will be Broadwell or Skylake.

Despite formerly working for Intel as a programmer responsible for the circuit simulation software the CPU designers used to verify their designs, I try to make my purchase decisions based on end-user features, not geeky CPU-level features.

Interesting. And I will replace my 2013 13" rMBP once Appl manages to offer a quad core CPU in the 13".

 

[chleuasme]

Image

Notable differences include: upgrade to iris 6100, TDP reduction to 15W(some website has 28W but this Intel leak seems more reliable), base clock increase of 0.1Ghz, Max GPU frequency is reduced from 1200 to 1100.

Similarities include: turbo boost frequency essentially remains same.


So the main message I am getting here is that Broadwell will be similar CPU and GPU performance (as IPC increase in CPU is said to be may be 5percent, GPU may be 20 percent) but at substantially reduced TDP

No real surprise here.
Exactly what to expect with a die shrink and the continual run for smaller footprint devices, both physically and thermally. That's what was meant when was said "30% power consumption reduction with Broadwell" all along; rMBP13 Haswell long-term performances will now be accessible to an MBA13-like notebook without throttling as much.
And that's why people speculate on the potential merging of the MBA13 and the rMBP13, and then all these rumors about a possible 12" retina notebook in place of the MBA11, with even lower TDP chips now also more capable with Broadwell and SkyLake in the future. (I link to the concurrent thread in the Air section: https://forums.macrumors.com/showthread.php?p=19111753#post19111753).
Of course, the other possibility is no merging, as is probable lower TDP chips around 25W could now be quad core, and could then be imagined a Pro line only with quad core CPUs, and the small footprint Air line with lower TDP dual core chips.

 

[augustya]

Hey Guys,

So is there any new news on the Broadwell Release Date for the rMBP ? If it is the first quarter of 2015 I guess by now we should have some rumors floating that it may most likely be released in the first quarter of 2015 anyone here recently heard anything new and latest on this ?

 

[puercaeli]

Hey Guys,

So is there any new news on the Broadwell Release Date for the rMBP ? If it is the first quarter of 2015 I guess by now we should have some rumors floating that it may most likely be released in the first quarter of 2015 anyone here recently heard anything new and latest on this ?

Broadwell U headed for MBA and rMBP 13 is rumored to be launced at CES 2015(second week of January). Previously leaked schedule note shiping to customers in week 5-10 2015.

So probabaly looking at feb/march.(may be will launch around iWatch launch)

I have a feeling though rMBP 13 will wait until June so 15 can be updated simultaneously.

 

[augustya]

Broadwell U headed for MBA and rMBP 13 is rumored to be launced at CES 2015(second week of January). Previously leaked schedule note shiping to customers in week 5-10 2015.

So probabaly looking at feb/march.(may be will launch around iWatch launch)

I have a feeling though rMBP 13 will wait until June so 15 can be updated simultaneously.

Hmmm...Interesting information what about the Broadwell inclusion in rMBP 15 inch ? Did you say June-2015 ?

 

[Rollo220]

Waiting for Broadwell MBP Thread

From the performance of the 5Y70 in the Yoga 3 Pro we can clearly see Intel is not ready with their 14nm process. We can also tell that transistor density is in fact worse than TSMC 20nm planar from die sizes and published transistor counts. The Yoga 3 Pro with Core M 1.1/2.6 2C4T 4MB L2 is appallingly bad. It's clear that instead of actually improving real world performance, Intel has actually managed to make it worse!!


It's time to switch to Apples own processors. Hopefully Broadwell-U won't be a dismal failure like Core M but if I wanted an x86 Mac, I'd buy Haswell now. A 6 core version of A8X with higher clocks would easily hold its own against i5s and i7s, if not in single thread then at least multi thread.


I am appalled at what Intel has done deceiving shareholders with the lama mountain concept and its canned benchmarks using s huge heatsink that could never be replicated .



That heat sink is what allowed Core M to appear to be viable. Intel should not be canning benchmarks that don't reflect the (terrible) real world performance of its products. The Yoga 3 Pro throttles constantly and gets at best 6 hours of battery life, why would apple consider putting this architecture in its laptops?


I sincerely hope, for Intel AND apple's sake, that Lenovo and the 32nm PCH are to blame for all of this. Because otherwise this is securities fraud.