Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Anyone Else Planning On Skipping Haswell And Instead Aim For Broadwell MBP?

dusk007 said:
You are forgetting one big thing here and that is platform power.
Broadwell is supposed to bring Quad Cores and basically all mobile chips into the fold of SoC designs. Haswell will only do that for the ultra mobile dual cores.
That won't make anything faster but it goes a long way in pushing total platform power as low as possible. Next to the Display that is where the biggest potential for increased battery life lies.
For high end quad core notebooks battery life should change a lot with Broadwell. It should end up almost at the same point as ulv haswells.
Quad Core Haswells get the new idle i0x state but aside from that the chipset and everything will largly be just as power hungry as with Ivy Bridge. While it may run more frequently on idle power consumption that idle power itself won't be too different. The SoC ULV are different and Broadwell will move all mobile chips to SoC.

Also die shrinks always deliver performance. What architecture changes does Haswell deliver. The CPU is the same just a bit wider same as from Sandy to Ivy. The GPU is mostly the same as in Ivy only bigger. The GPU from Sandy to Ivy is more different in architecture then now from Ivy to Haswell. The power managment and SoC for ULV is the only architectural difference. That is what Broadwell will do to only for the non ULV chips.
Core was a big difference in architecture for performance and Nehalem everything after Nehalem is iterative. Only on the GPU side one can speak of changes. Sandy to Ivy the EU efficiency of hte GPU climbed by 100%. With Haswell it is mostly more EUs and a more mature 22nm process. So the tick tock doesn't really hold true anywhere here. ;)
Click to expand...

There is some speculation in all of this.

Broadwell is supposed to make quad-cores standard at 35W/37W, but that doesn't necessarily mean increased power. 45W chips will continue to exist, and those will still be more powerful. We don't know how much power the 35W Broadwell quad-cores will pack, but they certainly won't be ultra-fast. Performance will depend on factors such as the clock speed and cache memory, and we know nothing yet on these quad-core chips. The performance of these quad-cores may be very similar to the performance of dual-cores.

There will be changes to the performance of both CPU and GPU, and to battery life, but this is only speculation at this point. There is nothing specific on it. Haswell is not even out yet, and people don't really know the exact power of its CPU and GPU, and its battery life. Of course Broadwell will be better, but so will be Skylane and Skymount. The real question is: is Haswell good enough? Are there any specific changes that Broadwell will bring that will overcome the shortcomings of Haswell?

I don't see a reason for waiting for Broadwell right now. There will always be improvements in performance, efficiency, battery life, and so on. But is it worth waiting over a year to get the hands on these processors?
 
Nobody needs to wait for anything but if you are like me and upgrading any 3-4 years one year more or less doesn't make a difference.

Max TDP and all that doesn't matter.
What we know is that only the ULV up to 28W Haswells sport the SoC design. The Quads and some Dual Cores won't. That doesn't help performance or make any difference there.
The thing is platform power, What the system consumes when it just sits there with the cpu mostly power gated, is better with SoC design. Also the size of the logic board which both helps price and/or space for battery and what not.
There are never specifics this early but history shows that some rumors are usually quite right if they make technological sense.

Why shouldn't they offer Quad Core SoCs? And it is quite obvious that Intel would like to go all BGA wherever possible.

Haswell is good enough for the Dual Core Ultrabooks now but I think Broadwell will be what Haswell is for the higher end bigger notebooks.
Like you didn't see many Ivy Bridge Tablets/Hybrids. Haswell will change that. So will Broadwell the more performance oriented bigger notebook market.
Tablet like battery life on a 15" with all the performance (aside from gaming) and not 10 pounds of battery to make it happen.
For the 15" MBP Haswell is just a new chip that is slightly better but nothing special. With Broadwell a really quite powerful 15" MBA like notebook can be built. If you need a well performing machine yet nothing for serious gaming. You would rather have it much thinner and lighter.

Just waiting for a new chip which is slightly better isn't worth it. Waiting for a new class of devices which isn't possible until a new generation of chips hits is something else IMO.
i.e. the new Intel Baytrail Atoms or the rumors of the next Motorola smartphone with supposedly a 4000mah battery. That would be something that doesn't exist now. All smartphones go for thinness over a big battery. While I would rather give up a good camera or a 1080p screen for a seriously long lasting smartphone. All newer ARMs got faster but not really more efficient (just hold Status Quo). If Intel reverses this trend (as they claim) and at the same time Google finally makes good on their claim to take smartphone battery life more seriously that is a new generation worth waiting for IMO.
Space wise most of todays smartphones could double their battery life if they added 2-3 mm to their thickness and fill all the added space with battery cells. For some reason they don't. Shoot the focus groups. :mad:

I am still on my iphone 3GS. Aside from the battery life and the miserable screen (poor contrast tiny size) it is good enough for me. A better camera would be nice but nothing I would pay a serious amount of money for an entire new device. Performance is also quite poor with some apps but something I can still live with. The day they release some compelling long battery life smartphone I would like to have it.
 
Basic requirements:

- Quad-core Hyperthreaded CPU with 4GHz base speed (or 4 equivalent AMD modules)
- DDR4 (not soldered, of course)
- hUMA
 
Well, to be fair;

Both Sandy, Ivy and now Haswell has had features postponed to a later generation, so I wonder if some of the things in Broadwell won't end in there until Skylake?




I think the only real answer is; what do you need and when? If you upgraded 2 years ago, it would make sense for most users to wait another year. if you wated 3 or 4 years, you might just want to go with haswell.


I am looking forward to seeing the IGPU performance of Broadwell combined with Nvidia Maxwell. That should be interesting. I wonder if we will be looking at 680mx quality graphics? That might be too optimistic, but not entirely possible.
 
Simplicated said:
Anyone planning on skipping the MacBook Pros released from 2013 to 2017 and aim for the mid-2018 MBP? :rolleyes:
Click to expand...

Jokes aside I'll most likely be purchasing my next computer by that time frame :p

I would honestly would still be using my 06' Macbook but it can't run anything on that powerful Intel 950GMA

I would like to assume that the 650M/Intel 4000 would last longer than that beast of a GMA.
 
I waited until Ivy Bridge MBPs were released and then replaced my 2008 MBP with a Sandy Bridge system for a few hundred off of its regular price, and used some of the money saved to upgrade the RAM and hard drive (swap to SSD). The strategies of a semi-poor student who doesn't require the absolute latest and greatest.
 
I think you can pretty much rule out that AMD manages anything near comparable single threaded performance. They lack funds and process technology. They would need a miracle.
hUMA by Intel seems more likely. QuickSync in a way runs on something similar. According to a Intel Dev the working data never leaves the package. It all stays in the L3 Cache and the GPU fixed function works on it the CPU and the GPU EU array until they get some transcoded data.
I think it is mostly a driver issue and needs some locks. Lots of things can go wrong if you mess stuff up with shared memory. I doubt Broadwell will come with a publicly available way of a hUMA like architecture. Some OpenCL functions may get some short cuts in the system. I think Intel will do that step by step and maybe at a much later time all the way.
 
AMD has been fabless for quite some time now. The ones doing heavy investment in process are TSMC and Globalfoundries, who have a lot of money.
 
cube said:
AMD has been fabless for quite some time now. The ones doing heavy investment in process are TSMC and Globalfoundries, who have a lot of money.
Click to expand...

doesnt matter really, you have to design the chips to work at said fab process

what dusk said is true, you have to worry more about the actual implementation of said feature than the actual availability

both are heading towards the same path if you look at how the cache and the igpu design from intel is right now. i want to see what broadwell brings to the table
 
Of course it matters, AMD was at a serious capital disadvantage when they operated their own fabs.

Designing a chip is not cheap, but the main obstacle to competition has been largely overcome.
 
My decision to buy/replace is more fiscally driven and need driven than just waiting for a newer technology. I usually do the best I can at the time, trying to get the most power I can for the price I can afford, while meeting need in the timeframe that need must be met.

The desire to wait for a new release is only an issue for me when a new release is imminent, such as now for new rMBP's and (maybe) cMBP's. If I needed one today and could wait, I would.
 
SMDBill said:
My decision to buy/replace is more fiscally driven and need driven than just waiting for a newer technology. I usually do the best I can at the time, trying to get the most power I can for the price I can afford, while meeting need in the timeframe that need must be met.

The desire to wait for a new release is only an issue for me when a new release is imminent, such as now for new rMBP's and (maybe) cMBP's. If I needed one today and could wait, I would.
Click to expand...
thats my take as well. if a pc is hindering you of doing what you need to do, or frustrating you in the process and making you produce less, its time for a change

the new tech is just there to justify the change and for you to enjoy that
cube said:
Of course it matters, AMD was at a serious capital disadvantage when they operated their own fabs.

Designing a chip is not cheap, but the main obstacle to competition has been largely overcome.
Click to expand...

doesnt really matter.

amd built more fabs than what they needed dumped money on useless things and overall was a very bad managed.

the point that matters in the discussion is how to improve adoption for huma and power for the amd cpus. inherently both are tied
 
Mr MM said:
thats my take as well. if a pc is hindering you of doing what you need to do, or frustrating you in the process and making you produce less, its time for a change

the new tech is just there to justify the change and for you to enjoy that

doesnt really matter.

amd built more fabs than what they needed dumped money on useless things and overall was a very bad managed.

the point that matters in the discussion is how to improve adoption for huma and power for the amd cpus. inherently both are tied
Click to expand...

It matters.

AMD could not keep pace with Intel, who is building newer fabs the whole time.
 
I've got a mid-2009 MBP (Core 2 Duo), and it's just starting to show it's age a little when opening my now huge iPhoto library. The processor is the bottle neck of my system.

I'm saving up (also a poor student) very slowly with money from my holiday job, so I'll probably be able to afford one in about 12 months time. By that time, the Haswell rMBP will probably be ready for a refresh so I'll just wait the extra few months for that.
 
AMD could really capitalize on their Athlon 64 architecture in no small part due to illegal business practices of Intel. For which Intel paid a ludicrous fine compared to the money they made. I don't really get what is the point of fines if they don't even deter because you earned multiple times the fine with illegal activity. :confused:
Anyway after the A64 they didn't have a follow up and Intel got off their lazy ass and dumped their bad P4 architecture for Conroe. Barcelona and Bulldozer didn't quite work out. Barcelona was okay but far from great. Bulldozer a two edged sword that lacked considerably in some important areas.
AMD would have had more money and more use for their fab capacity if A64 yielded more but still their follow ups just didn't cut it. That is the main problem. If they were just worse but not by as big of a difference like P4 vs. A64 things would be fine. Problem is Intel is just very far ahead for some reason.

The way I see it AMD didn't keep pace with Intel in architecture. The Integer Performance of their cores just lacks and single threaded IPC. They don't have to dump money into fab tech anymore but I don't see how that will fix their architectural mess. That has been the case for a while now. What is next after Bulldozer that will really change the game again. All I read about is incremental improvements are planned but nothing really new. Unless Intel falls asleep at the wheel that won't be enough.
 
As both TSMC and GlobalFoundries are working on advanced structures and nodes, I see AMD catching up quite a bit in the medium term (if they want).
 
cube said:
It matters.

AMD could not keep pace with Intel, who is building newer fabs the whole time.
Click to expand...

it doesnt matter. are you actually aware that their fabs got a lot of unused time? that they started to produce for others because of that?

it doesnt matter. if you produce more than people are buying, you are going to lose money on 2 ends.

the point is that they didnt keep up with the tech race, the die shrink race is a mere consequence of that, if you dont sell you dont use your production capacity, if you dont use that capacity you are losing money on stocks and on costs, and so on and so forth.

the problem here is that you identify fabs as the pinnacle, they arent, they are a mere consequence and in the amd case one of their money throw away pile.
 
My point is that to keep up with the tech race, you need upgraded fabs.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back