GTX 980M 28NM vs GTX 1XXXM 20NM

So what do you think?

Should someone buy a 980M now, either single or SLI? With 28NM Architecture?

Or.

Should they wait for the 20NM Architecture.

The reason I ask is wouldn't 20NM be far more future proof than 28NM?

I mean if you are going to sink 2-3 thousand dollars into something wouldn't you want it to last for a while?

Please let me know what you think.

 

There's always something newer and better on the horizon. Since I don't care for 3-4k, I purchased a 980M GTX because it should be able to max out 1080p for the foreseeable future, without theoretical yield problems.

 

They don't even have 20nm desktop, yet. And that always comes first.

You may as well wait for Pascal if you want to wait until the end of next year for something that may or may not exist.

 

Good luck waiting for Pascal in late 2015 or early 2016...
Heck, if you are gonna wait almost 2 years, why not go for the cards coming after that again?

Sarcasm aside, if you want the best of what Maxwell can offer, buy a notebook next summer. Then we will most likely have a full GM204 but of course, it will be in 28nm as well.
Think 20-30% faster than GTX 980M

 

Doubtful there will be a die shrink for Maxwell next year as NVIDIA have left themselves plenty of room for performance growth out of the GM204 chip for notebooks (and for the desktops, they have the GM200 chip). Fully enabled GM204 next year will net a very decent boost over the 980M without the need for a die shrink. Then it's Pascal's time to shine the year after. Pretty sure that's what will happen.

 

Always best to wait if you can

 

2015: "always best to wait if you can"
2016: "always best to wait if you can"
2017: "always best to wait if you can"
2018: "always best to wait if you can"
2019: "always best to wait if you can"
2020: "always best to wait if you can"
etc.

 

I do not care about 3 or 4k either.

How long do you think you will be able to get 60+ FPS on single 980M?

I agree. I do not think I want to wait that long.

I see. What is this about Full GM204? Is it worth waiting for? Check this link that I found. Is there any truth to it?

So if there is a great increase next summer with Full GM204 why are you buying a P35x V3 now, when you can wait 6 months for upgraded maxwell?

Are you just tired of waiting and ready to pull the trigger?

Seems like you can wait forever ...

Everyone please check this link out and tell me what you think ...

NVIDIA completely skips 20nm, 16nm four SKUs are Coming

NVIDIA completely skips 20nm, 16nm four SKUs are Coming

SemiAccurate today has exposed a number of Next Generation NVIDIA Graphics card release plan for the future, the future will be straight across the 20nm and 16nm.

This year the mainstream NVIDIA flagship core Kepler architecture GK104 will be replaced by GM204 the second generation of Maxwell architecture. Now GM104 is the first generation of Maxwell which will be retiring soon.
NVIDIA completely skips 20nm, 16nm four SKUs are Coming

A big GK110 core successor will be GM200.
The GM204 graphics core will have at least four models, expected to be named as:

GeForce GTX 880 Ti
GeForce GTX 880
GeForce GTX 870
GeForce GTX 860

All four are expected to release third quarter of this 2014.

In all four models difference is in number of stream processors, frequency, memory is different, there may be up to 15-20 streaming multiprocessor units, namely 1920-2560 CUDAs. Efficiency Maxwell architecture, energy efficiency will be on a large scale and therefore GM204 may not exceed GK110.
NVIDIA to skip 20nm process, third Generation of Maxwell to use 16nm

The most striking is, NVIDIA 20nm process will be completely ignored. Before and after the 2015 first quarter interim, NVIDIA will launch the second wave of new products, or third-generation Maxwell, the core architecture or GM204, but the process will be used directly 16nm – TSMC plans early in 2015 production of this process, that NVIDIA will be the first time to follow up.
In other words, you can say 20nm GM104 A stepping, 16nm GM204 is a B stepping.

The16nm version might be a GeForce 900 series, of course, possible name change (to be honest) or add a suffix to existing models (if you remember 9800GTX +).
As for now, why they are going to abandon 20nm, there is no clear view, the process may be in one or more aspects of performance, power, area, cost, and may be other people on organization are not very satisfied with the upgrade.

...

Is there truth to this?

Are they going to skip 20NM and go straight to 16NM?

If so then it would be safe to upgrade to 980M because it will be a while before they truly do 16NM ... Right?

 

What is this "safe" you're worried about? No matter what you buy, when you buy, you'll only be on the bleeding edge of technology for, what, a week? before the next best thing gets announced.. and it's not like what you buy today will suddenly stop working when the next thing gets announced!

 

Why not wait a few months after that for Skylake? And then a few months after that for Pascal? You nailed it on the head - I'm tired of waiting. I'm planning on keeping the Gigabyte until Pascal launches (though I do seem to be slipping into yearly upgrades; 2012 Samsung -> 2013 Clevo -> 2014 Gigabyte)

 

It's not a silly question to ask but there is not something that will blow it out the water just about to hit, no.

 

By induction, you won't ever buy a laptop. QED

 

That happened to me, finally decided this is the year I must break the cycle and commit to a new laptop.

Sorry old lappy, we've had good times but it's time to move on, regardless of whatever is around the corner.

 

Let's all just wait until we retire. At that point it won't even matter.

 

The fact that you survived 5-6 years on that laptop shows commitment, love, almost a marriage

 

Haha so sad but true! IIRC it was north of $3k when I bought it, but certainly lasted a long time. Still handles demanding apps pretty well and many games at 1080p, but starting to show its age. And the battery exploded at some point so I can only use AC power.

 

Upgrading because you actually lack performance you desire is perfectly fine I think. I more want 980Ms because they'll be a lot cooler and I'll be able to use a P370EM heatsink for my CPU which will further cut down the heat, but I do in fact wish I had both better CPU and GPU performance right now. It's just, it's currently in the realm of "I don't need more performance, but I want it" instead of "I need more performance, this gaming experience is unacceptable". I more wish I could do things like hit 120fps constant in Grid 2 or 125 constant in BF4 on ultra, instead of not being able to hit 60fps in any game XD.

 

well looking forward to the upgrade. my current laptop is a core-i5 (11"-12") tablet-pc with a radeon 5950

but still undecided on the

gt72 (4x 128)
g751 (asus for warranty)
or local made (when they arrive) sager/clevo slim
g35x slim

sigh

 

i loathe this response with so much hate....so much hate. There are without a doubt good times to upgrade and bad times. For an example the 22/28nm die shrink was a golden time to build a computer, which is what i exactly did.

Just get the 980m then upgrade down the road in 2016/17 again when you get 10nm/16nm

I might be getting a 980m then hold out til 2016/17

 

For real. Too much kvetching over GPUs, not enough actually enjoying games.

 

I will kvetch over GPUs 24/7... while playing Dark Souls 2 inbetween periodic forum checks XD

 

Oh I am just saying to OP if you wait, you'll be waiting forever. I bought a Genesis Coupe (2012) and in 2013 they upgraded the engine to give ~50 more HP. Do I regret it? Hell no. I love my car, and it suited me at the time of purchase. That's the way things go, there's always something newer and better. Just make the most informed decision for what's available at the time.

 

Volume production of 16nm FinFET to begin in Q2/Q3 2012. A little earlier than when 20nm began, so we should see 16nm Pascal cards taking over for 28nm Maxwell in early 2016.

TSMC: Volume production of 16nm FinFET in 2H 2015

 

2012? And thus, with TSMC gaining access to the TARDIS, things started becoming a lot more wibbly wobbly timey wimey.

Sent from my Nexus 5

 

The good thing though is that 16nm FinFET is a lot easier to produce than 20nm SOC according to TSMC.

But we will see.

 

My basic philosophy on gaming laptops:

I have come to the conclusion that high-end gaming laptops are only good for about about four years at stock. Sometimes you can upgrade the GPU, but I'm not sure that is cost effective from a $/yr perspective over the whole life of the laptop.

After four years a game may run, but it might not run with all the effects the then current effects a four years old GPU can do just from a graphics perspective. After four years, if it isn't the GPU, it's the CPU showing its age.

That doesn't mean that no games are playable after four years. It is better to view gaming laptops as gaming consoles. Over time less and less games will come for your "console". The graphics can be looked at the same way. You may find you can run some/most games, but they may not look as good as on newer systems. It all creeps in little by little around the fourth year and accelerates as you near the fifth year.

As for the question of getting a 980m vs whatever comes next after I will say that a laptop with a card that is about as powerful as a desktop GTX 780 looks like a winner of a choice to make. If the 980m's replacement is a 20nm/16nm die-shrink and it can get here next fall, that seems like another good choice. Pascal too, that is also looking like a good choice for 2016.

That doesn't mean that a laptop with any of these GPUs won't adhere to my general rule about gaming laptops after four years.

Four years isn't bad. It wasn't long ago that four years was considered for "future-proofing" when it came to desktops. For many, four years is still often about the time new desktops builds get considered.

Both current and next gen GPUs look like obvious winners. CPUs are stagnant, so in a way you can look at it like the 4yr-calendar hasn't even started on the CPU side. If you need a new laptop now, get one now with no regret. If not, wait.

 

680m and dual 680m machines will still be getting on pretty well.

 

Right, and the 680m was released about mid-2012, so we've just passed the two year mark on them.

My little rule is a general rule about what is reasonable expectations for the longevity of higher end gaming laptops. It is there to help keep perspective on making purchases for them. The point is, there will be a limit to what you'll get for waiting or picking upgrades no matter what and the law of diminishing returns comes into play somewhere along the way when it comes to upgrades..

It is not a perfect rule. Like human life expectancy, gaming laptop expectancy is generally getting longer.

Graphics are plateauing, so GPUs are relevant longer for that. In addition, consoles play into it too, the 9-series is the first DX12 and even the mobile is near to the performance of the desktop high end. The first DX after the last gen consoles was DX10 and the top end 8800 GTX was about the power of a HD5770 ...which is basically what I have in my G73jh that is almost five years old now.

Intel mobile CPUs took a big jump with Sandy Bridge and another smaller jump with Ivy Bridge, but more or less stagnated since then on two iterations of Haswell that will probably continue with Broadwell. You could make the case that the four year calendar on Ivy Bridge/Haswell doesn't start until Broadwell launches since they're all so close together performance-wise. Of course, that probably means that all those CPUs hit the dustbin about the same time.

So ...to conclude, two things.

1: I've obviously thought about buying my next gaming laptop way too much.

2: A laptop with a 980m is a very good choice.


I know I'll have $2k by April and I'm probably going with a 980m Sager laptop in the spring.

 

If 20nm doesn't come, or if it comes late. Expect a full GM204 mobile core (i.e. 2048 CUDA units) or a 1664 unit if AMD is really noncompetitive next year. 20nm is expected to reduce power consumption by about 30%, likewise this can also be translated to higher performance so I expect both companies to go for maybe a 15%-15% performance/power ratio.

There is no such thing as future proof, get the solution that best suits your needs now and is expected to age gracefully, right now, Maxwell looks to be the most ideal.

For example, take the desktop AMD 290, it suited my needs 6 months ago due to excellent price-performance ratio, is it future proof? definitely not, it is already rendered obsolete by the svelte and cheaper GTX 970,
is it still useable? definitely yes, uses a lot of power but very useable
is it expected to age gracefully? my prior research indicates so, the PCB seems reliable due to the 12months worth of reports from miners, the 4Gb RAM staves off early bottlenecking, core is still competitive with the 2nd tier of the next generation etc etc
In conclusion, don't think future proof, you will always be waiting. Think, is this a cost effective and efficient solution to my needs.

 

Well, AMD is supposed to come out with HBM memory gpu's early next year.
I am thinking of personally waiting for that since its only 5 months away (and the current laptop - in my signature - which is 6 years old btw) serves me ok still.

Essentially, switching over to HBM (and stacked memory) represent the biggest change as far as consumer tech is concerned and moving away from 'traditional' layouts.

Get that and you'll be set for a long while at least on the GPU front.

I think HBM (if delivered on time) would be worth waiting for since its slated to come out in February next year... while Pascal and stacked memory are not slated to come out until late 2015 or sometime in 2016.

Who knows... its possible Nvidia would be spurred into releasing Pascal earlier on as a result - though we won't know for certain.

If you need new tech now... then you can get Maxwell and it will probably be just as good in terms of gaming for a long while.
But if you want newer technology which would be 'worth' waiting for... then waiting for HBM/stacked memory would probably be the way to go.