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

Agreed. I've always thought NVIDIA would milk Pascal.

 

Moo.

 

_______________________________________________

If I am reading these link infos right, it would seem that mobile Pascal will also be released for Gaming Laptops in June or July?
If this is indeed the case, then it would seem the best thing to do, is to wait for a new new Gaming system until that time, and just skip the 980m/980mx altogether, to take advantage of much better power efficiencies, gaming performance, etc.

 

_________________________________________

Looks like I spoke too soon here, by thinking a June/July release date for mobile Pascal.
If you are right, then it's going to be more than likely late September/early October for the release, which is tough to have to wait for....

 

Nobody knows when it's going to be released, but the summer makes sense.

 

I'm pretty sure Volta is 2018 and not 2017. The timing is way too close, especially considering how big Pascal is.

 

Yeah volta is definitely 2018 for consumers.

 

Volta is definitely 2018 or later. NVIDIA's roadmap says it. They'll have at least 2 generations of Pascal, if not more. I'm betting they'll delay Volta to 2019 just to re-brand Pascal because I doubt AMD is going to be able to compete with Pascal 2-3 years from now.

Just look at how much performance they squeezed out of 28nm... Damn impressive, IMO.

 

kinda off subject, but will the pascal titan outperform two 980 tis in sli?

 

We dont know. Depending on how nvidia configures titan pascal? This is far too early to ask that question. We dont even know the base core config for GP100 yet.

 

Synthetically, no. I doubt that. It will be dialed back for re-branding purposes. However, considering how poorly made NVIDIA drivers are lately, maybe...in some games? There's no way of knowing at this point. I'd say a fair estimate of all "first generation" Pascal cards will be 50%-70% over their predecessors (e.g. 1080 is 50%-70% better than the 980).

One thing I can almost guarantee you: 4k gaming with a single GPU (1080 or better) will be possible at 60+ FPS.

 

Can't wait for this so I can fully utilize my 4k display. But it's probably not gonna happen before Q4 of this year for mobile cards. And then the card itself will cost ~$900 from the resellers >< /sigh

 

then sell off your current card and ull limit the overall cost

Sent from my Huawei Mate 8 NXT-AL10 using Tapatalk

 

Don't think there's a market for dGPUs in my country And I've never sold on ebay before so no one will buy from me with no ratings probably. Would be nice if we could sell them back to resellers somehow so they could recycle them.

 

980M will be good for a while. Sell the whole laptop when it's time to upgrade next year. Wait for 2017 when they released the "REAL" Pascal. Like they always do, the "REAL" Maxwell 980M will be released soon.

It would not surprise me if the 980MX has actual full DX12 support while 980M will not, whether it be hardware or software... lulz.

 

Although I always think the first iteration of a new architecture is the one to get, because the second iteration is normally just a frequency bump or maybe only a few more cores added - either way if you get the architecture when it comes out new you have nearly the same performance & with overclocking you can match it (& surpass it) yet you end up having the laptop sooner while that GPU tech is more relevant - you probably get to enjoy the laptop for longer. For example the 680M was a great GPU to buy, and wouldn't have made sense to wait for the 780M or 880M, as performance difference is not great between them - especially when overvolted & overclocked.

(If I had a 980M though I'd probably wait for Volta if I could).

 

I don't see mobile Volta coming before 2018. Selling the whole laptop would suck though, as the whole point was to get an upgradable machine. And I really need Pascal for the 4k performance. Maybe I can offer the 980M on the NBR marketplace though...

And yeah, I always have the sneaky suspicion that the first iteration of anything is not the one to get. Like with this whole laptop. Had I waited for the DM version (essentially v2) I would have been a lot happier, but circumstances like my work and location back then pushed me to get it when I did.

 

Ah, well if you want 4K soon/now then it makes sense to get the Pascal rather than waiting for Volta. I'd get the first iteration of the tech though, so you get it a year earlier, and won't differ much in performance in comparison to second iteration.

 

You can always sell your current rig and upgrade. For those with 980M SLI, it probably won't be worth upgrading until 2017. We'll have HBM2 and all the goodies by then for sure. We still don't know if the first iteration of mobile Pascal will have HBM. It probably will, though.

If you want to sell, sell now.

 

It all depends upon the level of satisfaction one want to achieve. For a person like me a setup with 980m sli should last beyond 2020. But I have seen people with 780m upgrading to 880m when it was released.

 

Haha, I bet they weren't happy with paying for that downgrade! Modified vBIOS & 780M used to kick 880M modified vBIOS everytime. Talking of GPUs, I wish Alienware still made 'em like they used to, can't believe the M15x can run a 980M without issue! I would have upgraded my laptop already to a 970M / (980M) if it was compatible - they should just keep it simple and manufacturer quality upgradeable stuff, quality that keeps on giving!

 

Yeah totally agree. Sometimes logic is not the top priority when it comes our decisions.
Regarding Alienware, you know sometimes i wish that dell had just discontinued the Alienware instead of lamafying it. It feels so bad to see these generic machines in best buys with Alienware name slapped on these.
Having never owned an AW machine, when i bough my current Asus laptop (almost 6 years ago), I was tight on budget but promised myself that my next system will be an AW. But sadly now it has lost all its charm and i regret now why i didnt get it back then (not that i have any complaints with the Asus which has served me with loyalty).

 

maybe not newer games. two 980tis in sli can barely go above 30fps on rise of the tomb raider in 4k. then think about other games like the witcher 3 and a few other demanding titals.

 

Yeah, and new games are only gonna get more demanding on the hardware. Older games and some current ones could be possible in 4K though.

 

I posted this in the GeForce MX thread, and I suppose I ought to put this link here too:

https://semiaccurate.com/2016/02/01/news-of-nvidias-pascal-tapeout-and-silicon-is-important/

TL;DR: All the sites saying that Pascal will arrive by June are pulling information from their rear ends, and Nvidia was lying about Pascal silicon already being in their hands.

Even more TL;DR: People looking forward to Pascal are in for a looooooooooooooooooooooooooooooooooooooooong wait.

 

brilliant article, thx for the link

Sent from my Huawei Mate 8 NXT-AL10 using Tapatalk

 

Yet some people point at AMD with disappointment LOL

 

Well if pascal is going to be a pipe dream I hope amd will fill the gap with something nice.

Sent from my SGH-M919V using Tapatalk

 

But, but but...

 

Interesting read. Seems more like a rant with poor grammar. It's funny that this writer criticizes everyone else about not having solid evidence when he himself doesn't. Maxwell was doubted as well, just the same. The loop of speculation continues, and we are stuck in it.

P.S. My "prediction" of a summer release is not based on any article(s). It's based on facts. Enough to notice a pattern. I'll stick with it for now.

 

I neither support nor disagree with the content in the article, but the author seems to know his facts. Rant or not, his break-down of the cargo manifest at the very least, sounds legitimate. Once again, we'll all be happier if Pascal releases on time (i.e. March-April), as have the past three generations of Nvidia GPU architectures.

 

Jen-Hsun Huang said, "Pascal offers 2x performance per watt than that of Maxwell." at the GPU Technology Conference in 2015. Source 1. Source 2. I think it's time we take a few steps back and recap. The information I base my "predictions" on is information released by both relevant and reliable sources. Something that can be proven. The articles often referenced are all speculative, written by journalists or writers looking to make a name for themselves in the world.

Considering actual facts, not opinions and speculation, we know Pascal will be released 1H 2016. And when a "delay" occurs, these journalists and writers blow it way out of proportion. A delay can be anything from one day to a week, or maybe a month. It doesn't mean the entire product line is delayed. It doesn't even mean production is off track, because things change every day. One delay may be countered by another furtherance later on.

Sometime soon, I may update the main post with relevant information and not speculation. Tired of this "back and forth."

 

If pascal was that much far off, we would have started to see their plan B. For a company who's survival is based on enticing customers to upgrade, it will be suicidal to just sit there and do nothing for a long period of time, specially as we know maxwell has still quite a bit of potential left in it and should be easy to re brand to keep their customers happy.
I know this goes against our concept that Nvidia is an evil company and Jen-Hsun just sits in a corner thinking of ways to screw customers while twidling his thumbs. but i feel this is the reality.

 

Oh yeah? Where's the hard evidence for this claim? Where are the test puppies? Any real-world results? I'm not saying that it wont be, I'm saying that it wont be soon.

Plan B is on a roll, it has MX suffix.

 

Well, I don't see how Pascal can come within next 4-5 months, considering Nvidia could not even showcase any (working or not) prototype at CES. On the contrary, AMD has demoed a working silicon in a gaming scenario, but that demoed card is still 5-6 months away from release. IMHO, there is no way in hell Pascal can come before the already demoed card, so the end of 2016 for the first desktop batch and I-II quarter of 2017 for mobile release seem highly likely.

 

Unless:
<conspiracy theory> Nvidia is actually hiding their completed Pascal chips and will unleash them in March, as have the past three generations.</conspiracy theory>

Probably not going to happen, but as I've always said - the earlier, the better for all of us.

 

Does nvidia really need that much of a plan B to deal with polaris 10 and polaris 11? Lets review their performances, polaris 10 is a 50w mobile chip that can beat a 950 ti. The new 965m is probably going to be a 60w? chip with similar performances. Its not like AMD is cozy with OEMs so I doubt nvidia have too much to worry about. Going for higher up, polaris 11 is a 970+ level of chip on at best under 100w. That gives nvidia an array of options such as 980mx/980 desktop and etc. Once you go up to 100w on laptops, people start to care less about power consumption because the battery time is going to be bad anyways. The best case probability for amd is that 970 performance in a 70w package. Which would theoretical be great for ultra thin gaming laptops but we will see if OEM bites.

 

Who said anything about what he said being "hard evidence?" I know I didn't.

My post was making the point that I'd take the word of a CEO of a multi-billion dollar company over some journalist on a website with the name, Semiaccurate.com. You can believe whatever you want, and deny all you want. I don't have control over you nor do I want it. Just know that if you decide to go against what the CEO of NVIDIA said, you're probably going to lose.

In layman's terms, the fact that he said it is evident enough of NVIDIA's plans. He has to keep his word or his company will lose money. So, odds are, he'll stick to his word and do what he says. That's what running a business is: profiteering & keeping shareholders happy.

 

Had you seen the fermi disaster with the same CEO?

I will trust semiaccurate over nvidia's official PR anytime of the day. Seriously, just look at the silicon from the demo, they werent pascal chips.

 

It's not NVIDIA's public relations team. It's the CEO of NVIDIA reporting information his R & D department told him.

 

Which is doing a public demo, hence PR?

Anyways, look up the fermi release disaster with the same CEO doing the demo. Charle was pretty on point with that one.

 

I don't know of that. But it's not relevant. I doubt they'd repeat the same mistake again, considering how long ago it was.

 

No, that demonstration showed nvidia had lied before in this type of demo before. So I am less inclined to trust them when there are solid proofs saying those dies showed at CES was maxwell.

And Fermi was still pretty relevant. It was apparently the first time nvidia rushed out a big die on a brand new process without any smaller dies before it. As you said, they wont repeat the same mistake.

 

@J.Dre - I can tell two things - the nGreedia guy is the last one I would take a word from. Second - you have never build anything in your life. If you did, you'll know that things almost NEVER go as planned. I'm pretty sure there are complications with the process and/or Apple is hogging the fabs, which is probably the reason to use gnusmaS as well, even though they were set to be all TSMC. The world is a bit broader than your POV.

 

Nah, TSMC just had a tendency of screwing up in general.

 

Typical... Can't defend your point, so you personally attack me. I won't engage in such pettiness.

Believe what you will. Only time will tell. And yes, of course I know things don't go as planned.

 

Its not about AMD vs Nvidia its about Nvidia vs Nvidia. If they want to do sustainable business, they have to keep offering customers (us) incentive. 900 series has been around for far too long. If pascal is not ready, they better come up with something else which is tempting enough. This lame 9X0MX with only 7-10% improvement will not be tempting enough. And in the end it will be Nvidia business that will suffer. I dont think they are oblivious to this.

Companies dont make their policies exclusively on the competitors business. Its about what the demand is and what they can offer. We can see this clearly in AMD policy. They are in survival mode. Hence they are targeting mid level customers where the biggest market is present. I dont think that AMD will even try to compete with Nvidia at top tier. It will be unwise of them to do so unless they can reach there with very small R&D and minimum expense.

 

pascal delayed? fine by me, extra time being top dog

Sent from my Huawei Mate 8 NXT-AL10 using Tapatalk

 

Right after the 980 in the U716

 

I kind of agree. Other than 4k, most users won't have a need or a reason to upgrade, and 4K is over rated.