What will next year look like?

I've decided to hold off on buying a new laptop for myself this summer. I can make it another year with the one I've got.

I know intel will possibly have Broadwell mobile (or maybe just a Haswell refresh) but what about GPUs, hard drives, chipsets, screens, power supply, etc? I haven't kept up with any of that.

Do we know much about what next years laptops will look like?

 

Next year, new graphics architectures from Nvidia and AMD are expected, so most likely there will be graphics improvements. SSDs are starting to be developed for the m.2 or next generation form factor slot, and computers next year are more likely to ship with an m.2 slot. High-end notebook screens have started to exceed 1920x1080, and more will likely do so next year. Still, most mainstream systems will retain 1366x768 resolution. Thin & light systems using u- or y-series Haswell or Broadwell processors are likely to see considerably better battery life compared to earlier systems. Also, Microsoft will continue to push for more touchscreens in order to make systems usable with Windows 8. More laptop/tablet hybrids are also expected.

 

My next year's notebook(s) will have the following (or higher):

Win9x64 PRO.

1080p IPS touchscreen with self-refresh technology.

Iris Pro level igpu (not only powerful, onboard graphics - but much higher cpu performance too).

DDR4 RAM - 32GB would be ideal as the 'minimum' - at least in a 15" notebook or larger.

Multiple 1TB or larger (finally!!!) SSD's based on a the 'performance' end of NGFF (4GB/s bandwidth).

See:
AnandTech | NGFF SSDs: Putting an End to Proprietary Ultrabook SSD Form Factors


'AC' spec'd dual band and minimum 2 antennae but better; 3 antennae wireless card, such as an Intel 7260 'AC' version as a minimum. Note that these cards are NGFF too.

See:
http://ark.intel.com/products/75439/Intel-Dual-Band-Wireless-AC-7260


As for 'chipsets'? More of them being integrated right on the cpu package... lower power, higher performance and of course; cheaper to produce.


This will hopefully leave room for higher quality and bigger/better battery technologies too - which will benefit us even more from manufacturer's stopping to concentrate on making the thinnest and lightest (but effectively non-functional; real world) models available to us.


To make this a long term 'keeper', this would all be embodied in a ThinkPAD of course. Along with any Keyboard, TrackPoint, Touch Screen and any additional reliability/durability/manageability improvements Lenovo can offer us over their current lineup.


In other words; nothing unexpected, given the glimpses we've been allowed to see so far.

Just hope that they all come together in a single (and worthy) package though.


Hope this helps.

 

Make this thing able to flip, swirl or w.e so it have a tablet mode.

 

No, I specifically DON"T want it to have a tablet mode because then we get into the 'make it smaller/lighter/non-upgradeable' mentality that ultimately gives us a throwaway product at mediocre performance levels and high end prices.


...

 

Convince them people can handle heavy tablet! I mean it is not even heavy to make that spec in 15".

 

I find it interesting that we've already achieved the same degree of technological proliferation as seen on the Star Trek television series and movies. A substantial number of people own smartphones that can interact with the surrounding environment in the same ways as tricorders. The different form factors of laptops and tablets means computing devices can be (and frequently are) taken everywhere and used in virtually all aspects of life, as the crew does with Star Trek. The popularization of Siri and Google Now allow voice control of these systems. Also, there are so many electronic components in modern vehicles, they cannot be used without such electronics, similar to how the Enterprise breaks down or has serious problems when its automatic systems fail. It seems the only ways the Roddenberry franchise remains technologically ahead of real life are in propulsion, weaponry, and medicine.

 

djembe, although we can debate the technology side of your statements - I am stuck on 'already'...

Considering that the world saw the first episode in 1966 and the show was technically in production from about 1964 - 'already' is not a word that I would use to describe this phenomenon.

As I was just a kid then, I have grown (very) old waiting for this 'already' to happen!!!

Can't wait until our technology goes into warp drive...

 

You have a point, and the Star Trek I saw as a kid was Next Generation, so I haven't been waiting as long. Still, it's supposed to take place hundreds of years in the future and it clearly is not taking anywhere near that long to get to the level of technological proliferation portrayed as would be expected by most people when it started.

 

We all desperately awaiting Maxwell GPU followed by Volta.....

 

If you need a new notebook now, buy one NOWWWWW.

If you don't, wait until you do.


The spec list moves forward forever. You can't get anywhere by chasing a moving target.

 

Just incremental improvements like we have always seen. Everyone always talks about the next big thing in laptops, but I have yet to see it.

If you want or need a laptop now, go for it. If not, wait until you want or need it. It won't make much difference either way. But I recommend waiting if it isn't a priority.

 

To the two previous posters:

This isn't a 'should I buy now or wait' question...

It is a 'what is coming down the road' question...


Not only is it hard to separate our wants and needs from purchasing decisions - it is just as hard forecasting if the next gen tech will be indispensable to us for the next half decade or so - and then; deciding if we need the very first iteration of it to have a productive 'edge' in our workflows - or, if we just want it to say we're at the bleeding edge (mostly for bragging rights)...


My 'forecast' of next years platforms (if correct) puts it comfortably into 'productive edge' territory for me.



Side note:

I finally had a chance to play with a 4700HQ for a few minutes (Asus G750JW): Haswell is a huge leap over IB in terms of responsiveness and overall performance vs. SNB i7 2630QM and even IB i7 3630QM based platforms. While the % difference is small (6-46% over the older platforms) - at this level of performance; even the 5% difference is a whole cpu's worth of performance from not so long ago...

I couldn't try any of my normal tasks with the system - but I have no doubt that in the next year or so Haswell/Broadwell based platforms will be truly 'unleashed' with NGFF SSD's, DDR4 RAM and certified 'AC' level wireless solutions (on both clients and routers).


And (shock!!!) this is based on a HDD based entry level QC Haswell 'sneak a peak' too; the G750JW.


See:
PassMark - Intel Core i7-4700HQ @ 2.40GHz - Price performance comparison

See:
PassMark - Intel Core i7-3630QM @ 2.40GHz - Price performance comparison

See:
PassMark - Intel Core i7-2630QM @ 2.00GHz - Price performance comparison


Now, imagine how the 'entry level' i7 4750HQ Iris Pro mobile platform with a high end NGFF SSD will respond to certain workloads:

See:
ARK | Compare Intel® Products


With a memory bandwidth 3x that of the 'lowly' i7 4700HQ version I played with...


If you can wait; this is definitely something worth waiting for - if you intend on keeping your system(s) for 4-7 yrs as I normally do. And your current system(s) are keeping you competitively productive in your respective fields until the next gen is finally available for purchase.


I agree we can't chase a moving target: but we can pretty accurately predict a little in advance what the next big jump will be.


These are incremental improvements?

Sure; like showing up for work in an F1 instead of a 1979 Datsun - just a little more displacement, a little tighter suspension and a level of performance that makes driving that 'so-similar' Datsun akin to walking...

.

 

The bottleneck was at somewhere else.

I've no idea what it was, could be pretty much anything, but not CPU doing any "huge leap".


You are trying to hit a moving target when comparing the two like old Datsun vs F1. If you get excited about new gen so easily, it's almost guaranteed that you'll get excited again next year, and regret your last purchase at the same time.

 

Next year, in the future(!), there will be robots! Intelligent machines that will predict your every need, with programs that can be controlled by thought and gestures in the air. One waves the hand, and controls graphics on the holographic super-screens as if conducting music!

Otherwise, if that's not going to happen - we're going to get a very large batch of Haswell rerolls. Which will dominate the market, and maintain the need for "ultrabook" +35w dispersion packages to be the baseline for medium to full-sized notebooks. SSDs will take off as an option, but the amount of hdds in existence (as well as the persistence of the production lines for hdds) will make that 7.5mm-9mm x 2.5 inch package continue to be used.

Same for form-factors and contacts for graphics card modules. And modular design on motherboards will lag behind for the same reasons (and stop the possibility of Tegra4/5 laptops as flat as a carbon sheet with a keyboard led on top, attached to a transparent touch IPS screen, or for example an APU laptop with massive amounts of additional slots, etc).

Meanwhile, graphics performance on the lower end is going to stay unchanged (the new line of nvidia and ati cards will be higher end cards, and it doesn't seem like cut-down versions of that variant is going to actually yield better performance at the ultrabook-notebook level, in terms of performance/watt at the 30w or lower dispersion package level).

In other words, where we're going to see improvements is in the high-end laptops. In terms of better screens and larger storage. We may finally see some more LAMD controller type competitors turn up as well, to provide an actually well-made and laptop-safe ssd option. We're very unlikely to see larger glass-coated but extremely thin and light IPS screens for laptops, unfortunately (see transparent LG-screens). Simply because producing these screens for laptops is a very different prospect, mired in different license agreements and standards when compared to in-house made "non-transferrable" chipsets. And with the same problem that keeps hdd a "well considered option" in small/transportable laptops, that's going to stop these screens from being used on laptop designs. Maybe next year.

Good news is that 1920x1080 LCD screens are becoming laughably cheap. Same goes for cpu and graphics card setups - architecture improvements means that the actual production costs go down, and this will make the artificial price-points we see now result in a percent-wise much higher drop in prices much earlier and faster than with earlier releases. And this is going to make purchases of "budget tops" next year give you massively awesome purchases very quickly.

In the sense that you may very well get a very high quality plastic/composite design, with improved power-management, with improved chassis and keys, etc., with quite expensive components for a very small amount of money. We see the same tendency now, that laptops that would have cost about 1000€ a little while ago is dropping down to 600€ and staying there, even though no new models have actually been announced yet, and the model is in truth selling to or at above expectation from the supplier.

What hopefully will also become more common is exchangeable lithium polymer batteries. Or synthetic lithium batteries. They can be shaped in.. you know.. tetris piece shapes if you want, can have higher charges with no safety concerns. And they'll last you ten years of constant charge-discharge cycles. Over the lithium-ion batteries, or the batteries with an organic solution, that needs to be cylindrical, and which will lose it's charge after a year or so. And also will blow up in some half-dangerous fashion if mishandled, or suddenly the electronic control system fails.

We're also not going to see architecture changes on microchips that will floor the current systems, either on desktop or laptop. In fact, the increases over the next couple of years - barring the sudden emergence of an industry approved integrated programmable bus with multi-directional memory - will be incrementally smaller and smaller, until they drop off completely. The tendency we're seeing with Intel, of all things, going for more energy efficient chip-designs is an expression of the fact that chips with sequential performance is becoming impossible to justify with the current designs, even at the current level, both in terms of architecture and cost.

Memory will be the same as before. But one hopes that laptop manufacturers finally figures out what XMP and JDEC is. Rather than go the route of Apple with their laptop-releases, where the only options are factory-installed, that then cannot be changed after purchase.

 

Scifi writers forget that as more technological/automated a society becomes, scientific advancement and new technology are developed/discovered exponentially faster (there is no 'linear' development).
Besides, Star Trek TNG wasn't nearly advanced as it should have been (most of the technologies it uses are basically something that we had the ability to do practically decades ago, such as magl-lev transportation on a global scale, hydroponic/aquaponic/aeroponic food growth in fully automated vertical farms, touch screens, etc.)
With 150 different races working together, and the kind of mindset the Federation is supposed to employ, they should have been exploring other Galaxies at least by TNG - the universe in fact.
Essentially with Kirk's era ending in the movie 'Undiscovered country' (78 years before 'Generations' movie), we should have seen a lot more by the time Enterprise-D was shown.
Instead, it was 'dumbed down' for the sake of not making it 'too disconnected' from today (which was a bad move because its supposed to be a different setting - that still could have easily addressed things that happen today - limitations of TV).

Right now (in real life) our science is about 60 to 100 years ahead of anything we presently use, and are continuously making large jumps in an ever shortening time frames.
The methods of production we presently utilize are also outdated by decades, we do not even automate as much as we can (75% of the global workforce can be presently automated with what we have now, and in less than 10 years, the rest easily) and we don't even use superior synthetic materials that we can produce sustainably in abundance.
This is largely due to monetary based economics that focuses not on what's possible technologically and resource wise, but instead on 'how much does it cost' and 'cost efficiency' (which has nothing to do with technological capability, resources, let alone technical efficiency) - this is the primary reason we see minor/incremental increases once per 12/24 months in the computer industry. The industry in general profits from planned obsolescence (which in turn promotes cyclical consumption and results in enormous waste) and making technology in a manner with least technical efficiency. We don't make things that last (they are intentionally designed to break or fall apart), they are not made to reflect the best of what we can actually do, they cannot be upgraded (or rarely/seldom can be), and we rarely recycle (in a manner that we harvest the materials from previously used things and use them to make new things).

Consumer grade technology is not a reflection of our latest science or our real technical capabilities - simple as that.

 

As mentioned, I keep my systems for a long time (4-7+ yrs).

Also; I did state QC i7 SNB, IB and Haswell platforms - can't get much more 'similar' across generations than that? You did look at the PM 'scores', right?

If anything: SNB was with dual SSD's (Samsung 840 Pro 256GB primary - M4 512GB secondary vs. the Haswell based twin 750GB HDD's - didn't check if they were in RAID0 though...


You don't have to believe progress is being made.

I have seen it for myself and I do.

 

*nods* I hear coal is a real hot investment now.

 

@Mr.Koala,

I'm waiting for some specific features, and I'll have a better deal on notebooks (through work) next year.

edit:

@all,

Thanks for the responses. Lots of cool info.

 

I'll admit I didn't read through all the walls 'o text here, but my take on it by Q3 2014:

- Intel Broadwell +15% improvement in CPU performance, + 10% battery life, + 25% iGPU... all marginal
- nVidia Maxwell +100% over GTX 680m if not more, with single top end GPU
- Wi-Fi 802.11AC standard
- DDR4 1866 16GB (2x8GB) makes its debut and more or less "standard" RAM for Q3-Q4 2014 laptops
- 1920x1080 still standard however more predominant on 15"+ laptops as standard, plus lower power tech, I don't think touch will be any more predominant as it is today
- mSATA still predominant small form factor SSD, although 512GB to 1024GB readily available and affordable, 2.5" SSD running 512GB to 1024GB at lest than $0.50/GB. NGFF to make a mark as replacement tech
- USB 3.0 still standardized
- Windows 8.1 or other minor update will be predominant OS, Win 9 to be released Oct/Nov 2014 to satiate the customer a la Windows 7, making it what Windows 8 *SHOULD* have been
- Cheesecake, beer, NHL and NFL are still awesome as ever

 

That would certainly be nice. Is there info on next gen GPUs / Maxwell? 16GB of RAM would be very useful for me.

 

How big of a jump is DDR3 to DDR4 ram ? That's the only real things that's catching my eye but it's a shame that these big changes are happening a year after I need a laptop for work

I really don't want to hold out but we are safe to assume most of the laptops with most of these features are going to be 1K+ right ?

 

DDR3 to DDR4 is a pretty minor change on its own, although in keeping with the overall trend: slightly less power and slightly higher frequency.

As to price, there are certain things that drive prices up (typically higher-resolution and touch-capable screens), but a lot of the rest isn't going to affect prices much, if at all. And some prices are dropping. As an example, a 256GB SSD currently costs between $150-$250 at most places. When sub-20nm NAND is commercialized, the same capacity SSD is likely to cost significantly less, say between $100-$150.

 

Tiller, you are nuts dude. You act like a few percentage points improvement means the difference between life and death, and then say that you use your laptops for 4-7 years? I hate to break it to you, but even my AMD netbook I bought two years ago is more powerful than anything that was out 7 years ago.

The only benefit to waiting is saving some money at the present. You can sit around with what you have for another year to get something only marginally better than you could get now or you can buy something now that will be almost as good, but you have it for an entire year before.

 

Well, I tend to benefit more from new releases than most users, as I use Linux (and it's easy to recompile something to benefit from the latest instruction sets), I do a lot of large code compilation projects (sometimes for school, sometimes for work where the projects can be massive), and various other really intensive things where a few % saves me 30-60 minutes of time. Having 16+ GB of RAM would help a ton with VMs too.

But waiting is something of a necessity anyways, next summer I'll have a company subsidizing the cost, which will likely shave off hundreds.

It's really nice to be able to hear what benefits waiting will have.

 

@ComradeQuestion

It looks like I should be sorry about some assumptions I made.

The new hardware do make more sense for you than most people. Actually a standard answer for the "will Haswell give me huge boost in CPU performance" question is "no, unless your tool chain is mostly FLOSS".

 

lol...


I know that the system I use is the best for me, (of course) and this is why:


I don't have a single system I rely on (notebook or desktop): I have at least a half a dozen options to choose from at any given time (notebooks and desktops).

What this 'buys' me is the ability to 'let' a machine get really old - but still be useable for a different task than it was originally bought for. While still being a complete 'backup system' to do things I need to do (hardware and software-wise) - if everything else was unavailable at 'that' time.

Yeah; I wouldn't want a 7 year old platform to do a current days work (raw graphic image manipulations) - it would probably take a week.

But it does allow me to compare what a single/smaller image 'feels' like on the older platforms.

And that is the cases where the 'small % differences' are not so small anymore:

The small differences are whole cpu's worth of performance, like I've said before; especially when a platform is fully exploited/balanced like I tend to keep my systems at (latest O/S; latest software, max ram, fastest hdd/ssd). This is where you see DDR2 RAM wholly inadequate vs. DDR3 - and where you see differences in Quad Core i7 generations that are more than the individual performance increases might suggest.


Synthetic 'scores' are nice. But they do not (and in my experience; never have) mirror real world workloads in any way shape or form.


The types of workflows I depend on make me realize greater benefits than most (I guess) going to each newer platform. If time is not money; then yeah; ignore what I've been saying...


But spending $200 - $500 dollars more today on a $2K-$5K platform to get 5% better performance over the lifecycle of a new system is a no brainer to me.

30-60 minutes time savings?

I've spend that (and more) to save 5-10 minutes a day (but; multiply that by two dozen systems...).




Nuts? Maybe a little.

But my form of crazy is wanting to pay once (for the hardware and the system setup cost in 'time') for a maxed/balanced setup - and then run the system into the ground* if I can.

And this, I do continuously on my 'main' systems...


* (Nah; I usually give/donate my systems away when I'm done with them).

 

The only laptop components/upgrades that will be superior to the current models is the new maxwell GPUs. The performance increase will be similar to the 680m's jump over the 580m. Next year will probably be a haswell refresh in terms of CPUs. Given that the maxwell mobile gpus will probably be about 50 percent faster than the current 700m series keplers, it could be possible that higher end laptops will start releasing notebooks with resolutions higher than 1080p. The haswell refreshes will probably only be about 10 percent faster, and any other hardware will not be that much of an upgrade from their existing counterparts. Technology is constantly upgrading, and it makes no sense to wait an entire year for a system. Buy a laptop now, upgrade later, sell later.

 

I mostly agree with your forecast, but I think improvements will be more conservative, maybe up to 60% on gpu side and probably less than 10% speed improvement on the cpu side (though the battery life and igpu improvements sounds likely).

I don't think we'll see ddr 4 soddims next year, but I think we'll see ddr3-2400 in laptops, especially in apu based laptops.



Sent from my SGH-I717D using Tapatalk 4 Beta

 

Speaking of RAM, is the (G)DDR5 on next gen AMD APU still possible?

 

Quite possible.
In order to get the most our of hUMA and HSA, I would imagine GDDR5 on the chip itself would be required, while the lower end might have standard DDR3 capability which would impede performance.
There has been no official word from AMD that they dropped GDDR5 after all (at least, as far as I'm aware of).

 

On-die GDDR5? I really doubt that. Having a few GDDR5 memory chips on the motherboard would work though. Although they have been pushing this kind of setup for increased performance, most manufacturers don't seem to bite because of the space requirements and/or cost.

 

Why would AMD want for GDDR5 on the motherboard?
Its only logical that it would be an on-die part, especially if they expect hUMA and HSA to properly execute CPU commands on the GPU without delays.

 

A decent amount of gddr5 (even as little as 2GB ) would quadruple the size of the processor and massively increase the price as it also introduces more room for manufacturing defects.

Gddr5 on the motherboard makes the most sense as there are currently no gddr dimms, and soldered on the board is the least expensive method to add gddr memory.

Sent from my SGH-I717D using Tapatalk 4 Beta

 

This is not a certainty however.
Also, isn't Steamroller's/Kaveri die area larger anyway?
AMD could have found a way to integrate GDDR5 in some form so it can be an on-die part (for example, using it for hUMA and HSA only)... for one thing, the reported bandwidth of Kaveri chips is (I think) lower than what GDDR5 will allow on the PS4 for example (granted the PS4 is a heavily modified APU, but AMD could have done something similar with Kaveri after all).

 

I7 5700QM / 32GB DDR4 / GTX880 4GB

Me like!

 

Well, we do have an Intel roadmap for 2014, and that doesn't include Broadwell (or 14nm chips) at all.
Also... I was under the impression that DDR4 (for non-servers) won't be ready until 2015.

 

No Broadwell is a real shame. Do we have much information on the Haswell refresh for mobile? On desktop it'll finally bring 8 cores, but what about mobile? Think we might see an 8 core CPU?

 

What's not a certainty?

If adding lots of memory on die was a good method, we would have seen more implementations already.

On die memory would be extremely risky as any manufacturing defect would ruin the whole chip, cpu, gpu, etc.

On package memory might work, but again space is a severe limitation, the size of one sodimm for example is double the size of a processor's socket and probably over eight times the size of the die.

The only commercial processor with on package memory that I can think of is Intel's high end Haswell processor with iris hd that has 128MB on package eDRAM used as a cache. http://www.anandtech.com/show/6993/intel-iris-pro-5200-graphics-review-core-i74950hq-tested/3

Sent from my SGH-I717D using Tapatalk 4 Beta

 

Its not a certainty that GDDR5 will not be on-die.
Right now we simply don't know enough about it to make such claims.

As for on-die memory being extremely risky, etc... as you already stated, a high-end Intel Haswell chip has a similar implementation with it's 128 MB eDRAM for cache, and apparently they don't really have too many manufacturing problems with it.
GDDR5 is used on a wide-spread basis in the industry and has very large bandwidth compared to other memory interfaces.
Its possible AMD will do something similar in this regard on Kaveri/Steamroller for hUMA and HSA.
They did it for PS4 after all, so why wouldn't they be able to do it for Kaveri/Steamroller?

Besides, I don't work at AMD, and Sanjiro (with all due respect), if you don't work there either (specifically in a section that's informed of Kaveri/Steamroller), you cannot know 'how' would AMD go on about executing GDDR5 on-die if they decided to go for it (which is quite possible) because they've been extremely cryptic about it - we couldn't know anything on Intel Iris Pro iGP's and how they were able to gain such an 'advantage' until they were released after all.
It is because of this why making any 'not on-die' predictions about GDDR5 in Kaveri/Steamroller right now is moot.

 

I guess you didn't read my message or the link; the eDRAM is on package, not on die, big difference between them.

On package means it's multiple pieces of silicon together on the same processor package, so you can mix and match various things together.

On die means everything is one solid piece of silicon, so the same fab process would have to be used to make the apu and the memory as one giant die, any defects will pretty much make the whole chip useless unless you can fuse off the bad parts; all in all it's really not worth the huge additional cost and less flexibility for a minor reduction in latency over on package.

Also if you read the link I posted, 128MB of eDRAM is 1/4 of the die size of Haswell, so if you wanted 8GB of ram on die, you would need a die at least 16 times the size of Haswell...

Sent from my SGH-I717D using Tapatalk 4 Beta

 

If on-die is tricky, is on-package (G)DDR5 Kaveri possible?