Worth waiting for Haswell?

[boomhower]

Haswell isn't going to be some miracle. It will be a larger upgrade than IB was but nothing earth shattering. It will be faster, cooler, and more power efficient. Nothing so big that new apps won't be able to run on IB hardware, that kind of change will never happen from one year to the next. They are saying that Haswell is going to be out Aprilish next year so probably early summer for the new computers and that's assuming that it's not delayed which is a very very real possibility. If you need a new computer now, get one. If you don't, wait. Personally I would wait if you don't need one. Getting the first model of a completely new model anything is generally a bad idea.


Intel IGP's rock. They aren't the most powerful thing in the world but can certainly get the job done. With the ability to play 1080P video it has eliminated the need for a discrete GPU for a large percentage of the population. Most people use computer for Facebook, email, and videos. Professionals and gamers that need a discrete GPU know they do but in general the masses don't. With Intels new GPU the masses are saving a lot of heat, battery power, and noise. I give props to Intel for how far they've come with them since they started making it a focus a couple years ago.

 

[leenak]

Processors are always improving. If you can wait, you'll always get a machine with better specs for a better price. At some point, you have to buy though and stop waiting. Only you can figure out that threshold.

 

[Jeff3f]

my take on this is that in a year the apps will be caught up better (ie most popular apps will be retinized with perhaps a few puzzling exceptions). The newer models will be Mountain Lion at the get-go, and there will be fewer wrinkles in the second generation rMBP experience. But it'll be 2013. There'll be newer and better stuff in 2014, and 2015 will probably mark a next generation that will be much better positioned to drive retina level pixel counts.

But, the current rMBP sounds like it might be vulnerable...the Anandtech review notes that there were performance issues even in Safari, though these were largely solved by Mountain Lion beta. To me, this means that even if they fix this, it's still GPU and/or CPU limited for what it has to accomplish in the retina GUI...so this may mean that rMBP will have a shorter supported life (ie no OSX "Chesire Cat" for you!). This will not be like the first (late 2008) unibody macbook (which is still extremely capable with aftermarket SSD), or windows XP (still using in 2012), or iphone 3gs (still being sold to folks new...).

I'm going to buy one, I wish I could get by with the base model as I don't think this is the machine to keep for more than 2 years...but I need more than 256GB. I think if one can get by with the base config, then this is a better deal, though it won't have future legs. I hope I'm wrong about future legs though.

 

[mattopotamus]

if you are really worried about what is to come, do what i do.....buy the entry level model to whatever mac you want and sell it when the new comps come out. If you keep it in excellent condition, you will only take a $300-$400 dollar hit. That, to me, is worth the price of having the newest every year.

 

[nitro230]

I am trying to make the same decision. Do I buy the rMPB now or wait and get the Haswell version next year.

I in no way "need" a new laptop right this second but I did want one and was all for getting this years prior to the announcement. I do however have concerns over the lag issues and other first gen problems.

Other than the obvious spec bump/energy efficiency associated with Haswell (which will occur on every new model), my main argument for wanting to wait is price. If this year was the last of the classic MBP's we may see the rMBP's drop in price by a few hundred (which, in my opinion, is reason enough to wait). I am also hopeful that next years rMBP will allow you to choose a 512ssd on the base model.

So basically, I'm torn and will probably make my final decision in about a month after I see what people are saying about their rMPB running on ML

 

[pandamonia]

Oh really? The HD3000 made entry level discrete GPUs redundant for most people and the HD4000 improved on that by ~50% a growth rate you don't see in the discrete GPU market. All of this with no price increase for the consumer.

They aren't competing with power sucking high end cards, they are competing with low end cards and they are doing a damn good job.

LOW END?

You mean the LOWEST END.

IGP is a joke and it makes me laugh watching people trying to play even the most basic modern 3D games.

Intel has less experience than AMD and even AMD can't get half decent power in their IGP.

Look at Larabee Intels silver bullet. - it got canned just like every other less than stellar attempt at GPU.

Intel 4000 has finally managed basic GPU level performance.

It took them nearly 5 years.

 

[dusk007]

I'm waiting for 4nm! 2022 baby!

I think you will be waiting a long time then. Getting over the physical 10nm barrier will be hard enough as it is. Getting down to 4nm might not be possible at all with anything that is remotely similar to the current transistors, or just take a lot longer with more and more research cost. 2022 would be very optimistic.
Nanotubes and stuff might help below 10nm but fabricating them on huge billion transistor chips is not going to be a walk in the park. It might be so much more expensive that it just takes a few years just to be economical enough.

I really wonder where we are in 10+ years. I think there will be less shrinking and more other stuff. Different types of non Si transistors that just perform better even in the same size. Voltage optimizations and Architecture.

 

[torana355]

the new rmbp is like the mba when it came out: a test subject. it is not completely polished, but it shows what the future can hold.

Exactly, im sure the second and third revisions of the rMBP will be the ones you will really want to get. First rev hardware is something i usually stay away from. Will use my 2011 MBA for another year or two then upgrade to a rMBP.

 

[Bubba Satori]

Discrete GPU's are 10 years ahead of anything intel even can dream of today.

Wrong.

 

[beamer8912]

I think you will be waiting a long time then. Getting over the physical 10nm barrier will be hard enough as it is. Getting down to 4nm might not be possible at all with anything that is remotely similar to the current transistors, or just take a lot longer with more and more research cost. 2022 would be very optimistic.
Nanotubes and stuff might help below 10nm but fabricating them on huge billion transistor chips is not going to be a walk in the park. It might be so much more expensive that it just takes a few years just to be economical enough.

I really wonder where we are in 10+ years. I think there will be less shrinking and more other stuff. Different types of non Si transistors that just perform better even in the same size. Voltage optimizations and Architecture.

Graphene transistors....I work on them

 

[Aftershocker]

There's always something better around the corner, and if you apply that logic you will never buy anything.

Personally I've ordered the retina macbook.. I will last me 2-3 years and then I'll upgrade the the next best thing...

In that 2-3 year gap I'll may swap out my phone and ipad to the newer models, although at the moment I have no desire to swap out my 4s or ipad 2.

 

[cmdrmac]

There's always something better around the corner, and if you apply that logic you will never buy anything.

Personally I've ordered the retina macbook.. I will last me 2-3 years and then I'll upgrade the the next best thing...

I'm in the same boat. My 2007 macbook was getting old and ML will not work on it. Plus, my macbook was slow, battery is getting hot and looks 'warped', and I wanted a machine with a bigger screen. Since Apple came out with new MBPs this year, I figured it's best to pull the trigger and go with it. Sure next year new MBPs will have newer features and maybe even a slimmer profile, but I don't think I could have waited till next year.

 

[dusk007]

Graphene transistors....I work on them

How long till mass production?
I ever only here they are great and nice and all but if you read the right new you hear that a lot. Between "that is great and awesome in labs" and "we have an upcoming product" is sometimes a lot of time. I am thinking OLEDs here. They have been around a long time but we are still mostly on TFT.
Shrinking known silicon production every few years is one thing, switching to something very different is another.
Would be interesting to know what is all the stuff Intel has in mind for 10,7 and 5nm. Apparently they think they can do it but they give not much details beyond that. 10nm could still be Si after that stuff should change as much as I understand not really coming from the semiconductor field.

 

[beamer8912]

How long till mass production?
I ever only here they are great and nice and all but if you read the right new you hear that a lot. Between "that is great and awesome in labs" and "we have an upcoming product" is sometimes a lot of time. I am thinking OLEDs here. They have been around a long time but we are still mostly on TFT.
Shrinking known silicon production every few years is one thing, switching to something very different is another.
Would be interesting to know what is all the stuff Intel has in mind for 10,7 and 5nm. Apparently they think they can do it but they give not much details beyond that. 10nm could still be Si after that stuff should change as much as I understand not really coming from the semiconductor field.

Still quite a ways off. Perfecting graphene production on any sort of macro-scale is a ways off. There's tons of graphene oxide production, and its really easy, but not useful for high-quality electronics.

There was a small breakthrough called graphene printing recently, that basically allows inkjet printing of graphene sheets. Given the right graphene ink, they could start making some products. 2022 sounds a little unrealistic though, unfortunately.

 

[magbarn]

The biggest advantage for Haswell will be much improved igp. This will allow your rMBP to remain in igp mode more often which is a big improvement in battery life over discrete. Yes, the 650m blows away the intel igpus but at the expense of up to 50% battery life. No, I'm not even talking about gaming which is a bigger loss. Intel's previous igpu have been more hype than performance, but apples continued flogging of them to produce better igpus is bearing fruit. The hd4000 in the IB mbp is faster than nvidias best 320m of yesteryear.

 

[psykick5]

I'll just upgrade to Haswell when it comes out. I don't mind selling then getting the new one, but only if it's significantly better.

 

[nuckinfutz]

Cheers for the correction, only done a quick google for haswell+intel so nice to see they are working hard on the lower tdp versions as well.

Haswell should make for some good server class CPU as well so you were right and I wasn't sure if anyone had discussed it's GPU in detail. We'll have to see if this Crystal Well stuff is the real deal.

 

[Panini]

Glad to see other people are in the same situation as me.

I want a new laptop now, but don't need one. That's the best way to put it.

Like others said, since this is the first of its kind, I'm leaning more towards staying away from it. But on the other hand, if next year is only going to bring a better iGPU and consequently better battery life, I'm not that excited - I use my laptop at home and don't take it wherever I go.

I know there aren't going to be any significant improvements to RAM, and the improvements to the GPU will be minimal (if I'm not mistaken), so CPU and iGPU (and maybe slimmer form) may be the only things worth waiting for.

If next year's model will lag less due to hardware, I'm all for it, but I have a feeling it might be the software updates that will be behind all the polishing up.

 

[grahamnp]

LOW END?

You mean the LOWEST END.

IGP is a joke and it makes me laugh watching people trying to play even the most basic modern 3D games.

Intel has less experience than AMD and even AMD can't get half decent power in their IGP.

Look at Larabee Intels silver bullet. - it got canned just like every other less than stellar attempt at GPU.

Intel 4000 has finally managed basic GPU level performance.

It took them nearly 5 years.

You seem genuinely upset about this, did you get burned by a 'lowest end' IGP? I really don't see what your problem is. Intel never claimed that their integrated GPU would do any more than it is doing right now. The only thing you can realistically compare it to are the previous Intel IGPs, even the AMDs aren't a relevant comparison because the CPU performance is nowhere near Intel's because the GPU occupies half the CPU die. What exactly is your definition of decent power anyway?

 

[Asherpotter]

There will always be an improved component on the horizon, but that shouldn't be your primary reason to wait at least a year immediately after the latest one was released.

 

[skaertus]

As far as I am concerned, the great thing about Haswell will be dramatically increased battery life. Intel is announcing it will run 24 hours with a single charge and 10 days of connected standby.

Given this huge difference, and considering that Haswell will handle better retina resolutions and will have a faster release schedule (as announced by Intel), I think I'll wait for it.

 

[walex19]

As far as I am concerned, the great thing about Haswell will be dramatically increased battery life. Intel is announcing it will run 24 hours with a single charge and 10 days of connected standby.

Given this huge difference, and considering that Haswell will handle better retina resolutions and will have a faster release schedule (as announced by Intel), I think I'll wait for it.

yeap I decided to wait yesterday..actually canceled my rmbp order..will just throw an ssd into my 2010 mbp.

 

[lcseds]

Intel is announcing it will run 24 hours with a single charge and 10 days of connected standby.

Given this huge difference, and considering that Haswell will handle better retina resolutions and will have a faster release schedule (as announced by Intel), I think I'll wait for it.

No way in H%*&. Batteries are not at this level. Nor is any electronics. 24 hrs of run time is a dream for the near future. I would say at best battery usage will go from an advertised 7 hours to perhaps advertised 9 hours. No chip set is going to change power usage that dramatically.
Don't build Haswell up to such epic expectations. It is going to be better for sure though. I prefer to buy today as speculation of the "next" great thing usually leads to some disappointment. Every one was waiting for Ivy Bridge and it's enhancements. Now it's here but not good enough? 'nuff said.

 

[pandamonia]

You seem genuinely upset about this, did you get burned by a 'lowest end' IGP? I really don't see what your problem is. Intel never claimed that their integrated GPU would do any more than it is doing right now. The only thing you can realistically compare it to are the previous Intel IGPs, even the AMDs aren't a relevant comparison because the CPU performance is nowhere near Intel's because the GPU occupies half the CPU die. What exactly is your definition of decent power anyway?

Im genuinely upset today because it looks like my IGP BROKE!

That aside, IGP's are the devil. When there is talk about substituting Discrete GPU with a IGP i get my pitchfork out.

Haswell wont be a quantum leap in IGP. Nvidia's 650GT in the new MBP IS a quantum leap in terms of mobile graphics thanks to a very good Architecture and 28nm power consumption.

Intel's efforts in the IGP space are the WORST in the industry by far. Haswell wont change that.

 

[lannisters4life]

Im genuinely upset today because it looks like my IGP BROKE!

That aside, IGP's are the devil. When there is talk about substituting Discrete GPU with a IGP i get my pitchfork out.

Haswell wont be a quantum leap in IGP. Nvidia's 650GT in the new MBP IS a quantum leap in terms of mobile graphics thanks to a very good Architecture and 28nm power consumption.

Intel's efforts in the IGP space are the WORST in the industry by far. Haswell wont change that.

The MBAs use IGPs, no Nvidia chips, and they're great machines. Fast, run for hours. My nephew plays Portal 2 on his 11" MBA non-stop, he introduced the game to me. All I need to do is physically pick it up to be amazed at what they pack into them.