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

That image is probably a render. Did anyone notice that the PCIe connectors themselves are missing from the card?

 

Could be a new connector on the back, like the old DELL and some Toshiba GPUs.

Yep, they didn't even bother to rebrand their stuff for this year. So, if anyone wondered how nGreedia is going to one-up the 7970m's rebranding, they just did that - no rebranding at all! Genius!

 

Although I find rebranding when there are no significant performance improvements worse than not rebranding at all - 7970M>8970M>M290x - if I recall correctly the 8970M and M290x are identical in every way and the 8970M has only something like a 50Mhz bump to the core over the 7970M - that's worse than no rebranding in my eyes - deception!

 

They kept the naming scheme current, but I guess you are right since there's nothing to go to. Desktops are 9xx as well at the moment. They could've give some free lunch with a 970/980MX while bumping the clocks a bit. It's not like these cores can't do that. But then again, they could've easily charge new GPUs money, which is not exactly free lunch...

 

Well, the only computers I've ever seen 'GTX MX' GPUs in are iMacs, so I wouldn't worry too much. Plus, Apple has gone over to Camp Red this time. Surprising move, to be honest. The R7 M370X in the rMBP is slow compared to the 860/960M.

 

The lower end mobile Maxwell GPU's are going in some laptops abd using the MX normenclature. That's why I linked the article a few pages back.

 

Fair enough, but what I meant is that we needn't be too worried about rebrands penetrating any higher than the low-end market.

Many laptop manufacturers are preferring to go Iris Pro instead of low-end discrete, primarily because of power/TDP concerns, at any rate. And as already mentioned here, Iris Pro has improved dramatically - it's on par with a 940M. The HD3000 back in 2011 was significantly worse than even a GT 525M, and this really is commendable, on the part of Intel. You can jack up settings in KSP to maximum and still get 30-40 FPS - that's really good.

I get a feeling 1060M will be a pared-down Pascal with GDDR5X, with 1070/1080M being the flagships with HBM2. Which means my next purchase will be a P640RE (or P640QE?) with a 1070M in it.

 

I don't see many Iris Pro laptops at all. Not only that most laptops with a CPU with Iris Pro cost significantly more than one with a ULV CPU and dedicated low-mid range GPU. The 940m is a good 20-25% faster than the Broadwell Iris Pro 5200 in mobile CPU's . The new Skylake Iris Pro 6200 is quite a bit faster, but yet to see it in the wild to really know.

 

if someone can get pascal to work on the R1s then i am set....on buying one

 

Man after waiting this long it's like... guess I have to wait till Pascel now but FFS I'm gonna wait for like another damn year aren't I. Getting real impatient now. >:|

 

We don't know exactly, but it looks like we may only see desktop Pascal by Q2.

Just FYI: We thought the same when they released the 700 series cards. They launched desktop at Computex. A week or two later, without warning, mobile launched. They launched the GTX Titan in February, the 780 May 23rd, and the 780M in June.

 

Has any leaks come up about what Nvidia showcase in CES-2016. From what I have been reading, they will focus on bs like autonomous driving. GPU will take back seat. Does anybody any more info?

 

CES would be too early for pascal, i think well only see that sometime later during 2016

 

Hmm. Maxwell is technically going to be 2 years old soon (since GM107 was released in March 2014), and any longer than 2 years means that Pascal has been delayed. Hope not, since the 14 nm process has been in production for a while.

 

They won't wait 2 years to launch a new product. They'll launch something by Q2 2016, probably before.

 

The proof is in the pudding AND i am getting hungry!

Sent from my SM-A500FU using Tapatalk

 

Have some cake instead.

 

Or Christmas cookies.

 

MXM Pascal design confirmed or proof that it at least exists. Looks like the same old 3.0b we're familiar with.
http://wccftech.com/nvidia-pascal-gpu-drive-px-2

 

Interesting, still a 3+1 VRM design. So power consumption shouldnt be too much higher then 980m.

 

yehaw!

Sent from my Nexus 5 using Tapatalk

 

Odd timing on the Lenovo Y900 17" 980m Gaming laptop release date:

June 2016, for $1999....

Won't we have Pascal laptops shipping by then?

What does Lenovo know that we don't know...or did someone forget to tell Lenovo about Pascal shipping dates?

Lenovo's latest laptop is a 10-pound powerhouse with a mechanical keyboard
Overclockable i7? Check. GTX 980M? Check. 64 GB of RAM? Check. Permanent damage to your spine? Check.
http://www.pcworld.com/article/3018...nd-powerhouse-with-a-mechanical-keyboard.html

Lenovo's first action game coming this year
http://www.pcworld.com/article/3019062/ces-lenovos-first-action-game-coming-this-year.html

 

ok so this lenovo is just one mm thinner than my dark knight, weighs around 1.3 kg MORE, does NOT sport a desktop cpu (im rather guessing the 6820HK due to overclocking, otherwise lenovo wouldve already advertized a desktop chip) and also does not SLI, but "only" a single 980M... whats wrong with this picture?

 

That you completely missed the point?

The point is, why is someone, anyone, shipping a $2000 laptop with a 980m in June of 2016?

Won't there be Pascal mobile GPU laptops shipping by then??

If not, then we have a longer wait ahead of us than we thought...

 

oh that went without saying, thats why i didnt enlarge on your point i just found the specs more interesting in comparison

Sent from my Nexus 5 using Tapatalk

 

The interesting part that relates to this thread, is when will Pascal 1000m GPU's arrive?

Lenovo shipping a 980m GPU in June 2016 suggests either Pascal isn't coming soon, or Lenovo made a major miscalculation.

We don't need to sell Clevo laptops in every thread, all the time, do we?

 

That Lenovo really is a joke, as @jaybee83 said. The P750DM easily overpowers it. Honestly, after using a Clevo, every other laptop doesn't seem as cost-effective, or has some issue. My greatest nit to pick are with Macs. More expensive than a P870DM while being less powerful than my W230SS.

 

Pascal should be much more efficient. A single Pascal core should perform equal to two Maxwell cores, likely using equivalent to or less power.

They're going to have a lot of fun re-branding Pascal for 2-3 years. You all better get ready for it. I wouldn't be surprised to see 1664~ Pascal cores in the 1080 (still outperforming the current 980 by a good 60%), and then seeing the "1180?" with 2048 cores in 2017, effectively doubling the 980's performance.

 

All thing considered, 50% ahead of 980m seems normal. It looks there is no HBM2 on GP104. As expected.

 

Its funny because most people dont realize that Nvidia showed GP104 today at CES
The die size is identical of GK104 and the power circuitry too.

Also, MXM cards YAY!
There is hope we all dont burn in solder with the upcoming notebooks





GTX 980M

 

Figured they'd present at CES. Been saying it all along. Now let's see if Computex is the launch.

 

AnandTech speculated it may be GP106:

More interestingly they also mention availability:

http://www.anandtech.com/show/9903/nvidia-announces-drive-px-2-pascal-power-for-selfdriving-cars

So that MXM Pascal GPU (either GP104 or GP106) should progress in its lifecycle beyond Nvidia-internal engineering samples in Q2 2016.

Still may be just limited availability, though hard to tell if Q4 2016 is because of that Pascal GPU or other PX2 components (e.g. there is also Pascal Tegra there).

 

yummy, MXM standard form factor, identical component layout = power / form factor / heatsink compatibility given baby! now its just a question of the bios / vbios im hoping for @Prema to the rescue if it aint PnP

 

Looks like we'll be stuck on GDDR5 still.

Probably won't see GP104 in laptops (MXM) until September or October though.

 

thats fine by me, means my 980M will be king of the hill for a while longer nope, NOT counting that 980 mobile desktop

btw, who says that GDDR5X will look any differently than GDDR5? just saying...

 

What was your 980M's highest clock?

Yeah that's true, we'll have to wait and see. I may skip desktop 980s and go directly to Pascal if I can wait long enough

 

Its definitely not GP106. Have you seen the 06 chips? They are much smaller than that.
Here is GM206



Interesting that Q4 availability. Hopefully since its car hardware that is more tricky to develop for or they dont prioritize as much as GPUs

 

as already mentioned in the other thread, i can push it to +290/+500 before it blacks out on me im guessing the VRMs just cant take any more, the PSU shouldnt be a problem

@Cloudfire: either that OR they actually DO prioritize car hardware, seeing as its already a major revenue stream for nvidia and the actual mxm gpus for consumers got pushed back cuz of that, yay.....

 

I'm sure the car products are manufactured elsewhere. Consumer products and car products are probably not even manufactured and assembled in the same country. Shouldn't affect the launch or release of consumer products. That would be extremely inefficient and poor business practice. If that were the case, and one were to be delayed, it would interfere with the entire company and all of its products.

My guess is that they are waiting for Intel and AMD, as usual.

 

GP106 wont be using the exact same power configuration as GM204. Unless the yield is so horrendously bad. I wonder what process its these chips made on though. TSMC 16nm LPM??

 

Yeah, I think NVidia will be missing out if they don't launch Pascal early'ish this year. If they release well in advance of AMD hardware, then people in the market for a new notebook will be buying NVidia rather than waiting for AMD, plus that 980M has been on the market for what feels like ages! (And 980 for notebooks is an expensive curiosity, so I'm not really counting that as a significant release).

 

I really, really hope AMD takes a seat at the table again for 2016. NVIDIA needs some real competition. Just look at what's happened since AMD has started re-branding year after year. The entire market is declining (quality wise) and re-branding is becoming a norm. Also, driver support is laughable at best.

Competition breeds innovation. Stop the re-branding! And Intel, you're just as much to blame, if not more so.

 

Until AMD brings out new mobile graphic solution, will still have to live with Maxwell for sure .
========
I'm still waiting for Pascal to pull my trigger.

 

AMD Polaris (with 2 to 2.5 times more performance per watt compared to Fiji on desktop which is already close to Maxwell, and a new architecture on 14nm) mid-range solutions for laptops coming out in 2 months time... followed by high-end ones in summer.

AMD didn't have an answer for Maxwell on the mobile I would imagine... actually, they could have used Fury Nano easily as a template. Drop it down to say 100W to 125W and get HBM first out to laptops, but I don't think they were thinking in this direction because desktop is a more critical area for them.

At best, I don't think Pascal will be all that much. I think we're looking at pretty much similar/equal performance per watt compared to Polaris... but the biggest differences will be in features, software support (DX12, 4K, VR, etc.), and of course whether or not we see HBM on laptops this year.
AMD was the one that brought out HBM1 for desktopts first... and with indications pointing towards Pascal using GDDR5 and/or GDDR5X... we have 0 clue if AMD will be yet again first to the table with Polaris and HBM2 on laptops (might be, since HBM was mentioned to be used across this architecture and that Samsumg might be mass producing both - with AMD getting priority access possibnly... but nothing specifically was said about HBM being used on laptops just yet - here's hoping, but I wouldn't hold my breadth).

 

GF 14nm is ready in 2 month? Thats interesting if it is true. The prediction of Arctic Island is still around Q2 2016 and Zen is at Q3 2016.


There is a correction on that WCCF article that reported it.

http://wccftech.com/amd-polaris/

"[UPDATE 2 : 2016 January 5, 11:59 AM ET] PCPer’s Ryan Shrout issued a clarification stating that the Polaris GPU AMD demoed was up and running after just two months from when initial production kicked off, rather than being brought up to market after two months."

If that is actually true and I understand it properly. It means Polaris taped out around early november. So we are on the same schedule as predicted and more or less around the same time as nvidia.

 

if AMD comes out with an MXM variant and it works with alienware then i think we may be onto something.

 

Wouldn't get your hopes up about being able to upgrade, even if Pascal and AMD are MXM. Maxwell barely works in Alienwares.

May be possible but extremely tedious and difficult.