nVidia 2015 mobile speculation thread

Would be dumb to release a "Pascal-based" anything at this point. Just losing sales later on. Save that hype and let it build for the full launch.

If there's a 990M this fall (which I see as unlikely) it'll probably be a mistake like the 880M because they'll be experimenting in new territory. NVIDIA killed it with the 680M and the 780M, and then the 880M comes along (out of nowhere) and has so many issues.

 

Just no. You are fueling very unlikely rumors with even more unlikely rumors...

 

Wasn't my intention, just pointing out that this mythical "990M" might be that Chinese CEO's projection of what a mobile Pascal GPU might be like. Everything certainly seems to line up pretty well. And no I don't believe we'll see any Pascal GPUs until at least at year from now.

 

The couple of things I find weird is:

They have both bios and vbios issues with this card according to the guys that have it. If it was another GM204 would that happen?
The price point for the card is $1200-$1300 according to one reseller.
Why not call it GTX 980MX if it just got a little more cores? GTX 990M or x90 name are usually for dual GPUs (GTX 690, GTX 590)
Dual 980M on a MXM (really really doubt its even possible) explains 16GB GDDR5 (8GB x 2)

Something in the back of my head wonder if its a mobile Titan due to the price and compability problems.
Its certainly interesting that the Hasee CEO said 2x980M thats for sure. Because the best performance with a full GM204 I can come up with is +35-40% above 980M.

If only we knew. :/
So far no drivers or software have listed the GPU, which usually happens a couple of months before launch. They add the GPU to the driver tree.

 

So, my estimate was a little high.

I did suggest that it might be called something like "Titan M".

Or else this is an elaborate prank or a hell of a scam.

 

You`re not the only one. I was thinking GM200 with around 2500 cores earlier. Right in the middle of GTX 980 (2048 cores) and Titan X (3072 cores).

 

How could GM200 possibly fit inside the power and thermal envelope of notebooks whilst still providing significantly more performance than a 2048 core GM204?

I mean, the 880M was on the borderline of being too much for notebooks.

 

I think whomever translated this CEO's presentation made a mistake. The 990M sounds too good to be true.

 

6990M wasn't dual GPU. Neither is M295X. Mobility 4870 X2 was two separate MXM cards (i.e. 4870 CrossFire), not two dies on the same MXM board. Naming doesn't really mean anything.

 

No, I think the translation is accurate. I think the information that was fed to the presenter is wrong. In my limited experience the Chinese tech press doesn't do much fact-checking.

 

People are still running with this impossible rumor? Wow. If it isn't Pascal, it isn't possible. It's that simple.

Only other option is that Nvidia expects ever notebook maker to build a specialized, ultra-thick machine too accomplish the feat.

 

And we know it's not Pascal because HBM 2.0 is not ready (plus AMD has priority access) and neither is 16nm FinFET.

 

will someone just show me the original press or w.e.? that will solve a lot of mysteries as I don't need translations

 

speaking of cores, i noticed that the 880m and 980m have the same amount of cores. is there a leap in gpu accelerated video editing using cuda cores? i read somewhere that the number of cores matter more than the architecture and clock speeds. if that's the case, how much of a performance of having more cores in this ''990m'' would be seen in 3d modeling and video editing, not gaming?

 

Easy

GTX 680 (1536 cores): 195W >>>>>>>>>>> GTX 880M 125W (1536 cores)
GTX 980 (2048 cores): 165W

You see there is another 30W (165W > 195W) they could use on more cores than 2048 cores to build a GTX 990M with the same TDP as 880M.

I have no experience with video editing, sorry.
Perhaps someone else here could answer that question

 

Now that Skylake and Windows 10 is on the Threshold... when comes the new GPUs? I even miss the rumors... no info, not even on the naming (1000 GTX or what?)

 

GTX 990M is probably here somewhere around September.

Pascal might be here in early Q2 2016. But mobile could very much be launched in some computer fair like computex in June 2016 or something. Who knows

 

I still believe it'll be called 1080M. I call it!

 

I love my laptop. I will only upgrade my laptop if this Pascal mobile GPU can run Witcher 3 max setttigs 60fps 1080p on one card only.

 

It's not Pascal. Sorry to disappoint. Pascal is most definitely 2016.

 

Yeah I know it will be 2016. Got some time to save cash hehehe

 

They'll only release (or name) a GTX 990M if there's a GTX 990. Keep an eye out for that.

NVIDIA skipped the 800 desktop series, so that mobile and desktop naming schemes would match up. #ThugLife It would all have been for nothing if they screw it up again. That was the logic behind the original idea we had come up with for the next card, the "980MX."

 

I'm telling ya - 1080M. It'll happen

What other year hasn't it happened?

 

This year? Doubt it. There won't be a 1080M this year. They always launch the x50M's, x60M's, etc. first. We would have already seen those or something about them.

 

900M 2014
800M 2014
700M 2013
600M 2012
500M 2011
400M 2010
300M 2010
200M 2009
100M 2008

It'd certainly make history if that is to be the case!

I suppose that maybe, just maybe for marketing purposes, I could see them reserve the '1000M' nomenclature for a big upgrade like Pascal.

The OEMs won't like it though!

 

The 880M wasn't considered a new card, though. It's just a re-branded 780M with more VRAM and higher clock.

680M 2012
780M 2013
980M 2014

They also skipped 800 series in desktop, so this doesn't really count.

They'll probably release another 980M but with more cores and a higher clock this year. Everyone seems to think Pascal isn't ready, and I agree.

 

960M and 950M say hi!

Exactly! OEMs love to rebrand stuff! I can't imagine they would like to go two years using the 900M nomenclature.

Fact is, only us enthusiasts know the difference between a 780M and a 880M. Average Joe's gonna always think that bigger number automatically is better (which is how it should be of course!).

 

Don't get what you're trying to say here. There's a 980M that corresponds with my point.

I was suggesting if you think there will be a 1080M this fall, they would have already begun launching the 1050M, 1060M, etc. Or at least have released some information about them. And we definitely would have seen them in the GPU-Z database.

We always see newer cards in drivers or GPU-Z software before they launch. Nothing yet.

 

980M and 970M came before 960M and 950M. And 965M for that matter. So the midrange parts don't always come first. That's my point!

I'm just thinking it'd be unprecedented in the mobile world for there not to be a 'new' generation of GPUs launched this year. Whether there's any new chips or not, they usually like to come out with something 'new' to pacify the OEMs.

 

Again, that was a special case because the 960M is an 860M. They didn't need a 960M at the time of the 980M/970M launch.

Usually, those lower-end cards come first. The whole release schedule was screwed up because of Intel and AMD. Hopefully the schedule is put back in track for 2016. I miss the good ol' days.

 

The PASCAL GPU should be upgrade friendly.

 

It doesn't matter if they're the same, it's still the marketing class of GPU. Like Cakefish says, Joe Shmo doesn't know the difference. It's new and shiny and must be better. Nvidia update an entire line from low end to mid range to top end. There's always a rebrand somewhere in the lineup.

We will have gone two years without an update from Nvidia if this is the case. But you know what? It doesn't matter. There is no competition. Now we probably will see 2-3 years for a single top end GPU to be king, because there's no competitive reason to release anything new. When Pascal hits, Nvidia can and probably will sit for 2-3 years and just rake in the profits.

 

Screw this, I should find a new hobby

 

Me too. I really used to enjoy PC technology. Figuring stuff out, all the new advancements and options. All the *quality* games and large number of interesting game ideas that came from companies that were about the games more than turning the biggest profit they could possibly make. The 1990's and early 2000's were the best. Now everything is being filtered down to the lowest common denominator. OS's and hardware are now designed around what the company wants and not what the users want. Everything is so homogenized and integrated that there's nothing there but a few chips on a PCB. There's no risks, no developments, nothing exciting. Just slow progress. I remember 3DFX fondly, and the massive leap in gaming technology that it brought. I remember Windows 95 how significant it was over Win 3.1 and DOS. I remember the leaps and bounds from having 640KB of RAM and 1.44MB floppy disks and 256 color video cards, and the rapid escalation to 4MB of RAM, 80MB hard drives, and video cards with 8MB of dedicated RAM with 32-bit color! Ooh!

Now it's nothing but "The Man" squishing us with his thumb and pointing to us telling us we'll like because he says so...

 

I'm with you. New hobby incoming. Someone offered me $2500 for my laptop, and I may accept.

@HT: I know the average consumer isn't aware, but I was talking of the exact opposite - production side.

 

May I suggest becoming an audiophile... Expensive to do right, but well worth it! Also picking up woodworking is fun, too! Finally, if you like being frustrated and trying to figure things out (especially why things aren't working), politics and the modern theories underlying each area legislated can do it for you! Hours of reading, hitting your head up against a wall, yelling at nothing wondering why someone would do something like that, and finally a deep depression that comes when you realize there is little you can do about it other than write blogs, articles, or books, or run for office. (Bernie 2016!!!)

 

I had a long rant here but I'll cut it down to something manageable: it's our own damned fault. Gamers demand more: more eye candy, more play modes, more everything. Games take time and money and people to develop. The more that we demand from games, the more time and money and people need to be allocated to create them. This cycle has been repeated for so long that the only studios which can afford to make triple-A titles are the ones backed by large corporations which are, by definition, greedy. That's the price we pay for "more".

 

I only agree to a point. The problem is everything has been filtered down to graphics, and enthralling gameplay be damned. There are so many same as games out there, very little new comes out that takes any risks or tries something out of the ordinary. 20 CoD games just doesn't cut it. Don't get me wrong I do enjoy being part of the interactive movies that single player campaigns have become, but there's little there to give us more. Multiplayer is the main focus now so it's all about graphics and pew pew pew. Battlefield games are completely locked out from an modifications. I can only imagine what the talented enthusiast modders could make with BF4 if a dev kit were released.

Look at the classic games from the 1990's and early 2000's that people still play and find fun like Deus Ex, System Shock 2, Half-Life 1 and 2, Civilization, Descent Freespace, C&C (early ones like Generals and Red Alert), Neverwinter Nights, Warhammer, KOTOR, NOLF, Between Good and Evil, Operation Flashpoint, not to mention great flight sims like Jane's, Falcon, and Enemy Engaged series, among many others. There are still games in the last ten years that stand out but they are few and far between. Games like Witcher, Portal, Elder Scrolls (but just builds on games from the 90's), Borderlands, Bulletstorm, Spec Ops The Line, Bioshock (although based on System Shock) are pretty good, but many of those didn't fare well and will never see an expansion or sequel.

 

Ahh, as the sun sets on 353.49 a new Beta driver climbs out of the pit. Behold 353.51, be afraid, be very afraid:

http://www.mediafire.com/download/r...desktop-win8-win7-winvista-64bit-NDA-beta.exe

 

I have it on good authority that the new GPU will be 79-89% faster depending on driver. I have promised not to release sources. Even if I was asked by the mods of the site... Admins too...... Pascal numbers i can not say one word about except to say "impressive"

 

That's Pascal though right? 2016?

Edit: wait you're saying they can get up to 89% faster using Maxwell? That's much faster than a desktop 980. And desktop 980 needs 165W.

 

hehehe, you have a source that knows performance about GTX 990M and upcoming Pascal GPUs?
Are you trolling or sincere now?

 

Can anyone explain how a '990M' could be 79-89% more powerful than a 980M and yet consume less power than a desktop 980? New fab node? That's all I can think of.

 

I'm a mod. You tell me. Note I mentioned drivers...

 

GM200 perhaps?

Try to enter this in google search:
(2432*1100)/(1536*1100) =
(2560*1100)/(1536*1100) =

 

The new chip will consume too much power to have SLI on a single 330W adapter... Its very unlikely that there will be an SLI variant.

 

So there are hopes it is a GM200?

Full GM204 scenario calculated:
(2048*1100)/(1536*1100) = 1.33 = +33%

I find it extremely unlikely that drivers make up for the extra 30-40%. But perhaps your source was very positive about the drivers. Interesting info Ethrem

 

So in other words, this thing is only going to be compatible with the most uber of gaming notebooks with 330W PSUs? P6xxSG not going to be invited to the party?

Will there be a 985M (GM204 fully unlocked) to fill in the gap between the 990M and 980M? One that would be compatible with existing 980M notebooks?

 

Sounds like my batman needs a new PSU

Gesendet von meinem SM-P600 mit Tapatalk

 

Its being tested. I'm under NDA. I have been told what I can and can't say. It is a very high power chip and that's all I can say... Wish they'd send me one for the NDA!