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

980ti was about 15-20% faster than 980, if I remember right.

This means that there is about no chance that a 1080m would beat a 980desktop for laptops.

 

I still have one of them 860m, and a Clevo P870 with a desktop 980 seems a much better idea to me. If 1080 is only 62% than a goodly oc'd 980ti, then I can safely assume that 1080m will be at most as fast as desktop 980 in a laptop.

 

By this do you mean the 1080M will likely ship in the same kinds of anorexic notebooks that carry 970M now? Or that 1070M/1060M will be tremendously faster than 970M/960M?

 

?? The 1080M will be as fast or faster.

 

for me to be interested to upgrade to a 1080M it would have to be about 35-40% faster than the desktop 980 otherwise ill enjoy my trusty 980M for one more gen

 

You're dreaming.

 

Just that the upgrading from 970m to 1080m will be more compelling than 980m to 1080m.
For 980m it seems like an easy skip.

Yes, there are ppl with non anorexic laptops that carry 970m .. There are even ppl using a P700ZM with 965m because all they want is desktop CPU on the go (not me though)

If 1080m is between 980m & 980 for notebooks and running hotter than 980m, Ill wonder why they'll bother releasing it. But then again there are enough ppl who buy without research ... Even I bought my first Clevo (P651SE) without knowing what bga meant.

 

The 980 was very much a niche market and shouldn't be considered in the performance expectations from a high end but still mainstream part. The 980 was a special part that took a lot of power and a lot of cooling, not to mention proper binning of the chips. NVidia isn't going to come right out of the gate with a 980 Ti performance level chip for mobile when they will be milking Pascal.

 

Reference GTX 980 Ti only has a Firestrike Extreme graphics score of 8026, this stock GTX 1080 scored 10.1k.

GTX 1080M will be faster than a GTX 980. Don't be silly.

 

That all depends on what AMD brings to the table. Polaris will be out first for mobile... Considering AMD is definitely going to milk Polaris, nVidia has no real incentive to push things for mobile, especially now that Volta is going to be pushed back.

 

1080m must beat desktop gtx 980 because there is mxm modules for desktop gtx 980 for clevo and msi and msi promised that it will have desktop mxm modules for next gen gpus for gt80. So there will desktop mxm gtx 1080 for sure in clevo and msi even alienware (if they want to be competitive again) by this years end on Q1 2017.

to put it simply gtx 980m scores 10k at stock in firestrike.

gtx 980 mxm scores 13.5k at stock in FS.

So 1080m must be in the range of 13k to 15k at stock for sure.

Prevously 980m was around 90% of gtx 970 and is equivalent or overtook desktop gtx 780 at stock

so this time around will be the same or better, will run cooler and consume less power and nvidia promotes closing/reducing the gap between notebook and desktop gpus..

 

I think vega is too close for a polaris re brand. I think polaris 10 & 11 will be the only polaris chips we will see until early next year when vega arrives with HBM2. Following that Nvidia will release GP100 with HBM2.,,,,Just my feeling.
I stand by my prediction here
http://forum.notebookreview.com/thr...00m-series-gpus.763032/page-207#post-10250837

Edit: I cant get the post to link properly. Here is what i said.

 

I believe the 1080M will be around 50% better than the 980M, stock for stock, which puts it just above the regular 980 (20%~ give or take).

It's a fair bump but not enough, IMO. I also believe they'll be intentionally holding back Pascal to milk it for at least one, probably two iterations. Hell, would be willing to bet now that they'll delay Volta to 2019 to milk Pascal for a third time like they did with Maxwell (aka. Kepler 28nm).

 

But you're talking about desktop parts. AMD won't be stuffing Vega in laptops and nVidia won't be stuffing GP100 in laptops. They're both going to leave themselves plenty of milking room for mobile, it just depends on how much.

 

Ah, is that so you can get better CPU cooling & therefore maintain higher clocks when gaming, which I imagine could be more useful when your CPU is closer to becoming the bottleneck (especially when referring to Pascal)? I'm still jealous you can upgrade to all these latest MXM cards with your M15x from 2010 or was it even 2009!

 

2009 it was and yes, jealous is indeed the right word for that just incredible! clevo should just leave those vendor specific mxm pins blank and voila wed have ourselfes infinite compatibility

Sent from my Huawei Mate 8 NXT-AL10 using Tapatalk

 

Would luv to have one of those and pimping it up like that!

 

That is indeed my plan. 980M can already play most things at 3k with DSR fairly well. A small bump in performance at much lower power consumption would be great for me.

920xm doesn't bottleneck too badly for games yet. I can still reach around 80 -100fps in most games including modern fps games when the bottleneck is shifted over to the cpu. My reason for connecting the heatsinks is to be able to play games without a cooling pad (the 980M is HOT) while perhaps even allowing me to clock the 920xm to 4ghz without further mods.

M15x can be had second hand pretty cheaply. Pimp away...make sure you got deep pockets though if you want to bring a badly specced one up to date. It is a relic from an age that might never return (aside from Clevo machines). These days a machine even comes with soldered system ram!

 

980m and hot
I keep thinking u talk about a gpu when u mention 920 xm

 

It probably is better from a business point of view to just solder them all in one piece. They can milk the most money, and people don't care either way.

Too bad for them that there are bad guys like us who don't accept something like this.



I wonder why they keep doing this for real. I mean, car industry makes much more money from parts and repairs than from new cars sold. I think they would make more if keeping everything socketed.

 

I'm pretty sure I saw this linked in this thread earlier, but it's nearly time for this event so thought I'd link it, and also for those in the UK then this happening at 2am tomorrow (or tonight whichever way you look at it!): http://www.geforce.co.uk/whats-new/articles/nvidia-special-event

Annoucement time! You can view the event at the link above, but I'm not gonna stay up for it.

 

Yes, NVIDIA is announcing Pascal today. Knew it wouldn't be late 2016, ya'll are so pessimistic.

http://www.nvidia.com/object/gpu-architecture.html

EDIT: I am still hoping for a Computex launch at the end of this month!

 

No numbers, no care.

 

Just talked to the sales rep at Sager and started the RMA for my new laptop, with pascal being so close it makes the most sense since I've only had this for a week.

 

How much is the restocking fee?

We still don't know if Pascal mobile is close to release, or how long it will be before laptop's with Pascal single GPU and then Pascal SLI will be shipping in quantity.

Will they wait for Kabylake CPU's before releasing Pascal models?

Would you want a Skylake CPU version of a Pascal laptop, or would you wait for Kabylake? Or wait for new mb chipsets with Cannondale CPU laptops with Pascal?

Whatever you get, in 6 months it's going to be "old news", but still more than capable for gaming

Whatever comes out first with Pascal SLI is going to be expensive, much more than today's 980m SLI.

Possibly priced in the $4500 range like the GT80S 980 SLI.

And, given 980ti SLI equivalent 4k performance it's still not going to be fast enough for AAA 60+ FPS gaming.

It's just going to be a 150 FPS 1080p gamer instead of a 100 FPS gamer like the 980m SLI is today...

So, maybe 980m SLI is still going to be the sweet spot price / performance wise for a while yet.

Maybe that's why Sager is willing to take back your 1 week old fully functional laptop without a stocking fee. It's going to hold its current value for the foreseeable future.

We won't know for sure until 1080 mobile SLI is shipping

 

No restock fee, full refund guarantee from sager directly

 

Well I will start the RMA , and wait to ship it back out tomorrow. If they mention mobile , it's going back. If nothing for mobile, I'll keep it for good

 

Provided that most professional applications don't support SLI, I think that unless 1080m beats 980desktop, it is a no go for professional users. And I don't expect it to beat so far, if 1080 desktop barely beats 980ti, the m version is probably not going to be enough fast for professional usage.

 

Professionals shouldn't be using a GTX card, but a Quadro. Last I checked nVidia intentionally neuters GTX series cards so they can sell quadros.

 

What exactly are neutered on the last 3 generatio of GTX mobile cards? Maybe ECC support? Which is not needed depending on the application.

 

Should clear things up:



@HaloGod2012,
All said and done, I think you made the wise choice. Even if mobile doesn't come out until June or something, that's still not long enough away to justify spending top dollar on last-gen hardware, IMO. And even if the 1080M is only 50% better than the 980M, 1080M SLI will destroy 980M SLI, enough so to justify the wait.

One could argue there hasn't been a new generation of hardware since the 680M, but meh. Anyway, good choice.

 

Go look at how a 780M performs on pro applications vs the equivalent quadro K5100M. The difference is cringe worthy. AMD doesn't pull this BS with the Radeon/FirePro, both do very well in pro work, FirePro performing better. The nVidia neutering is just god awful, and it didn't need to be done. Just more greed.

 

Maxwell?

 

I didnt mean driver side things. I know how badly certain apps favors quadro for no apparent reason other than drivers.

You can upgrade a geforce to quadro sometimes by changing resisitors.

 

I think he means 680M, 780M, 880M, 980M were all 28nm

There has been no documented way to do this on mobile nvidia GTX cards, and quite frankly, it shouldn't require a hard mod. nVidia artificially gimps their GTX performance so that you have to buy a quadro for pro work. Really a dick move. This was the reason why I went 7970M instead of 680M in college. I was able to use my 7970M and NX/Solidworks, where my friend with the 680M cried about the piss poor performance in both. nVidia will not get a dime from me until they stop artificially gimping their cards.

 

Ohhhhh right. OK

 

Sorry I couldn't resist it.


If you feel offended i will take it off.

 

Oh yeah.

M3000M is barely double GTX860M performance,

M5000M is 68% from a GTX 980Desktop found in a P870.

I am sure I should be using Quadro for making games, but I will say just no.. Additional cost looks like +1500$ for less performance.

Unless doing very specific workload that really needs quadro, most professionals will still be using GTX cards too. For example, Professional photographs still use GTX, and so do graphic artists. Only certain 3D workloads come better on Quadro.

 

I didn't pay for any of those cards, so technically, not a dime from me. They were all system replacements or leftovers from people I performed upgrades on. They will be all replaced with Polaris cards if they do arrive on MXM

 

My sparse experience with AMD was not that much better than nvidia. They have their own problems with drivers, and tend to crash, or to have bad results.

nvidia at least worked after many tries of bug fixes and many tries of configurations.

 

Nvidia uses physical resistors to mark the hwid so they cant be altered driver side. Here is an example on desktop: http://www.eevblog.com/forum/chat/hacking-nvidia-cards-into-their-professional-counterparts/

I believe the exact same is done on mobile side as well.

 

Neither companies are in it to make you happy....

AMD would turn around and do the same thing as nvidia if they had the market share.

 

Thread Synopsis: everyone expects unrealistic performance and architecture that breaks the rules of physics.

Zero defect drivers and utility software. Updates to those drivers to be weekly, always provide a performance increase, never have any bugs, and never fix any since there were none to begin with.

Nvidia to make minimal profits. Practically give their hardware away and cater to even the most niche use case, who cares how much it costs in the production pipeline or what the market is driving towards.

Sent from a 128th Legion Stormtrooper 6P

 

Haha, but I still think they milks too much!

Newsflash: NVidia sues popular Swiss Chocolate Company for copyright infringement:

 

But they could be friendlier and give us more.

If you told someone that we would have the performance of GTX980 8 years ago, they would never had believed.

I am sure it is possible with current technology, and they have it in the laboratory, but they just want to make as much money as possible.

And I don't want drivers to be perfect, personally, I want them to work and not crush my work.

 

I do not entirely understand the joke

 

You cant scale things like that. Just because one of the ES chips is working doesnt mean you are getting actual decent yield viable for commercial launches. It usually takes awhile for the first chip to go through revisions before it is "launchable".

 

I personally don't even bother with the first releases on a new node, as they always tend to have teething issues. The second iteration is always a bigger improvement in terms of reliability and refinement.

 

I'm hoping they do not release mobile cards for a while so I do not have to return my new sager. I actually love my current sager and the 980m sli is more than enough for my resolution and games. Plus the system runs perfect, no issues. BUT....I couldn't stand it knowing a 1080m sli option was coming very soon. This is the first time in my pc gaming history that I want something to be delayed

 

Desktop is going to be today most likely. Paper launch and actual launch in end of May? Not sure.

Mobile is gonna wait for awhile. I wouldnt worry that much about a GP104 based 1080m any time soon.