Pascal: What do we know? Discussion, Latest News & Updates: 1000M Series GPU's

Just my own predictions. They've painted themselves into a corner, marketing-wise. Any future M GPU will be regarded as inferior to the '980 for mobile' by the general masses who would naturally expect a full-blown desktop GPU to launch soon after, as NVIDIA has now set the precedent. It would have been easy to brand this thing as a 985M/990M etc. but they deliberately chose 980. It signals to me that mobile will no longer play catch up to desktop. I honestly believe they are trying to unify their desktop and laptop GPU lineup for stronger branding. Pascal will enable this vision to become a reality. The big die chips (Gxx00) will always be reserved for desktops in the form of the x80 Ti class products and above. The stuff aimed for the smaller, thinner, lighter laptops that wouldn't make sense on desktop could occupy the x40, x30, x20 and x10 class range, with desktops starting from the x50 class. Everything in between could then be shared between desktop and laptop. Having desktop class GPUs in laptops sounds a lot more advanced than M class GPUs and would cut down on a lot of the confusion that surrounds the mobile lineup (what is xxxM equivalent to in desktop terms? etc.). Even though we enthusiasts know to look past the marketing and branding. We're just a minor section of the community though. It's all about that marketing.

 

The "general masses" you're referring to make up less than 5% of their target market, lol.

95% of consumers don't even give a crap, and quite frankly, probably don't even know what MXM is, let alone all this class talk of GPU's. The big money is not with us enthusiasts. We are the minority. You have to remember that NVIDIA (like all businesses) practices under one goal: to maximize profits.

It would be in their best interest to maintain the traditional method, while adding a little kick for us enthusiasts from time to time.

 

I just don't see them ditching the "M" moniker completely. No way. There is too much variation in the mobile market vs desktop. High end they may just ditch the M altogether from now on, matching a high end desktop GPU. But below, where you have 15-50W TDP GPU's there are no desktop equivalents.

 

By general masses I was referring to the non-enthusiasts.

They look at numbers and letters on a spec sheet. Seeing M causes confusion. Seeing the same GPUs as desktop reduces confusion. Creating a stronger brand image, less fragmentation.

 

Pipe and dream. I don't think it causes confusion. Just say it's "M" for mobile and people seem to nod their heads in understanding.

 

I think they just wanted to shut us enthusiasts up for a while. Pretty sure it worked.

They should keep things as they have always been, but add a little extra something for us enthusiasts (annually).

 

FTFY

 

We shall see, we shall see. I'm calling it. This is my prediction.

Everyone thought that the laptop market was heading towards soldered only and overclock blocking. It's gone the complete opposite direction.

Anything can happen.

 

No. Only Clevo is swimming against the current.

 

Clevo is going to dominate the market (performance wise) and others may begin to lose sales. You can speculate as to what may happen next... But it's good for us.

Clevo be like... "Screw eGPU's, we'll just stuff a desktop card in a laptop!"

 

Fixed again

 

NVIDIA reveals first generation Pascal chips will have 16GB HBM2 memory. The 32GB chips will be manufactured later on, being available in late 2016 and early 2017 on the first and only refresh of Pascal. - Source

NVIDIA also confirms Volta will launch by 1H 2018 and will be up to 12 times faster than traditional PCIe 3.0 with NVLink technology. Don't worry, Pascal will also see NVLink support but not at this scale. - Source

It seems Pascal will be approximately 3-4 times faster than Maxwell, and then Volta will be 3-4 times faster than Pascal. Exciting times ahead!

 

i cant afford these kind of things. T.T though if i do get another laptop its MXM boards all the way for me.

i just hope pascal keeps standing up to gaming even after volta shows up. otherwise i might have to go desktop to keep my wallet from killing me in my sleep.

but interesting indeed.

Sent from my SGH-M919V using Tapatalk

 

I think Volta will have a new socket with Pascal transitioning us to it. Pascal may be the last MXM generation.

 

yea, your probably right. Though i ferverently hope that OEM's recognize the collective boot to the arse the enthusiasts are trying to send them and keep MXM in some form or fashion.
thiugh from what i have been hearing they dont plan on MXM disappearing any time soon.

Though i suppose i could go back to all soldered. Not something i look forward to.

Also speaking of NVlink im curious what exactly it is, i have been hearing its a replacement to SLI, and sometimes im hearing its gonna replace PCI-E on the board.

Sent from my SGH-M919V using Tapatalk

 

It's a replacement for PCIe, in its broadest sense.

NVLink: http://www.nvidia.com/object/nvlink.html

NVLink and Pascal: http://devblogs.nvidia.com/parallelforall/nvlink-pascal-stacked-memory-feeding-appetite-big-data/

Basically, it eliminates the memory bottleneck you may have heard of by enabling the GPU and CPU to communicate at an exponentially accelerated rate. And this is where DX12 steps in... Pascal will utilize DX12 to its fullest extent.

 

So where does this leave AMD. Is this proprietary to nvidia?

Sent from my SPH-L720 using Tapatalk

 

Not sure what AMD's response is going to be. I suppose we'll have to wait and see what they do with Arctic Islands.

AMD specific hardware wouldn't have an NV-Link slot on them. They must be planning to have some sort of bridge connector between PCIe and NV-Link (or CPU?). However, with any form of adapter, there is possible loss of performance. Can't really give you a better answer at this time.

Maybe they'll use a PCIe based CPU controller? Not sure. They'll fall way behind in the gaming world if they don't do something.

 

So..and im throwing this into the dark here, with what im understanding is they are going to intergrate NVlink into the mobo, and then they are going to stop using sli cables?

Sent from my SGH-M919V using Tapatalk

 

That's what it looks like Intel and NVIDIA are planning to do, yes.

 

Huh, interesting. i wonder what they are going to do to keep connectivity. though i cant see it being too hard to develop something.

not just for laptops but desktops as well. i highly doubt they are going to come out with nvlink capable models and non, as well as backwards compatability.

Sent from my SGH-M919V using Tapatalk

 

I am hoping I can stuff some Pascal stuffing into my M18x R2.

Sent from my SPH-L720 using Tapatalk

 

Im hoping mxm stays or see's an evolution. i dont want to go back to all soldered.

Sent from my SGH-M919V using Tapatalk

 

It's possible to have NV-Link and MXM, but I think MXM will reach its limitations soon.

Pascal probably won't work on older systems, anyway. Motherboard won't support NV-Link.*

*This is pure speculation. I'm sure we'll know more by CES. Not good business sense to immediately jump ship (from MXM), though.

 

This is why I haven't pulled the trigger on 980Ms for my big AW. 680M SLI works surprisingly well in the games I play so I am not really feeling any pressure to upgrade just yet.

Sent from my SPH-L720 using Tapatalk

 

Waiting won't hurt. They'll only get cheaper.

 

Wow, 680mSLI is slightly more powerful in "3DMark 11 - Performance GPU 1280x720" than my 970m o0
Even though its SLI, Im still impressed.

 

Yep, grandpa can still kick ass.

Sent from my SPH-L720 using Tapatalk

 

Im waiting for pascal to come out before i upgrade from my asus with the 765m.



Sent from my SGH-M919V using Tapatalk

 

NVIDIA holds secret meeting with members of certain press outlets in preparation for the launch of their upcoming products. - Source

 

Dual grandpas

So much for secret

 

can't wait

 

loving those "highly detailed specs"

Sent from my Nexus 5 using Tapatalk

 

Titan XYZ confirmed

@n=1

 

I read that news for at least three times, and I didn't notice that at all....


Boring boring Chelsea

 

If it's a 990(2x 980Ti) I may replace my Titan X with it

 

There should be a GTX 990 released this October or November. It has been rumored since the 980Ti came out.

 

So what will be different between 980 Ti, 990, and Titan X?

 

They should call this one Titan XX, so after the transition to HBM2 they'll be able to make a triple GPU card with the PCB space savings, and we'll finally have a true Titan XXX.

 

More cores, many GB per second, a lot of watts, many memorys much wow and expensive.

 

990 should be 2 x 980's (Ti's?) on the one PCB.

 

J.Dre, a 990 desktop GPU, not a 990m mobile GPU, right?

That is, until they fit the 990 desktop GPU into a laptop

 

Titan X would be the 3072 core 12GB vram full GM200 chip based monster

980Ti would be the 2816 core 6GB vram reduced GM200 chip based card

990 would be two 980TI's on one PCB(like the GTX 690, 590, 295, etc.) with 5,632 cores(2x 2816) and 12GB(6GB effective [2x6GB]) vram monster, unless they base it off of titan x with full vram and cores then it'll be like $1,999 or base it off of 980Ti for $1,499.99.

 

Call it Titan X^2


Boring boring Chelsea

 

Call it XXX - pr0n for the geeks.

 

From what I understand, the GTX 990 will more likely be GM204 (x2) with a higher clock speed because GM200 (x2) would be difficult to fit. However, in light of the recent developments of "980 in laptops," I have no doubt in their ability to make GM200 (x2) happen.

I'm hoping for 980TI SLI performance in a GTX 990.

 

2xGM200

http://wccftech.com/nvidia-launching-dual-gpu-flagship-graphics/

 

Haha:



Graphics Card:Coming Soon
GPU: GM200 x 2
CUDA Cores: Twice As Many
Core / Boost Clock: Pretty High
TDP: A Lot Of Watts
Memory: Many GBs
Memory Bandwidth: Many GBs per second
Price: Expensive


I bet it will cost less than $2000 because then it would just be in competition with Titan X SLI. $1799 is my guess.

 

They're just speculating. Could be GM204.

Probably not worth the cost either way.

 

No it won't be GM204. AMD will be releasing Fury X2 later this year. If they did a 990 based off of GM204, the Fury X2 would destroy the 990.

nVidia had made the Titan Z in a 375W TDP design using two GK110s which used more power than GM200. They can make a 990 based off of two GM200s with the same 375w TDP.

We don't want an R9 295X2 vs Titan Z situation again but worse.