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

Well I played games alot in my younger days. Took a total break in my young and mid 20's. Now I'm 32, settled and very much into games again. Funny how that works. I do think I'm ultra fussy about what games I play now though. There's very little that interests me.
Somewhere along the line hardware started to peak my interest more.

 

Same old stuff we know, but ... enough with this old man crap!



NVIDIA Pascal GPU’s Double Precision Performance Rated at Over 4 TFLOPs, 16nm FinFET Architecture Confirmed

Volta GPU Peaks at Over 7 TFLOPs, 1.2 TB/s HBM2.

 

so funny, i couldve written that post myself, EXACTLY like that, including the growing interest in hardware, picky about games and the age of 32

Sent from my Nexus 5 using Tapatalk

 

no it's normal, im 30 and am disinterested in the majority of the stuff that is released today, it's made for the younger gen with more free time on their hands

 

fallout 4 though, pry that from my cold dead hands

 

Heh I have a 6 year old, I feel guilty when I play games without him (ie on the laptop) so I don't really get to game until night time or if the wife takes him out with her.

 

Lol I feel the same way when I read that post! I am only 29 but also am VERY picky with what I play and therefore game rarely. I do still LOVE to mod though!

@Getawayfrommelucas I understand your sentiment too! I only game when my little one is tucked up in bed

Feel like an old codger!

 

I understand. My son is now 24 and we play online together since he now lives in another state.

Right now our game of choice is Mechwarrior Online. When we do a private match he pretty much destroys me. I love that kid.

Sent from my SM-G928V using Tapatalk

 

So when is the anticipated release date again? I have lost track.

Sent from my SM-G928V using Tapatalk

 

2016 +/- 6 months

Sent from my Nexus 5 using Tapatalk

 

I can wait. My obsolete P170SM-A is working well. I have some AW toys to play with as well.

Sent from my overpriced Galaxy S6 Edge +

 

Interesting bit from that article about HBM2 power consumption:

So I guess that would be more indicative for mobile Pascal GPUs getting GDDR5X instead of HBM2?

Though from that graph it looks like first HBM2 column is still smaller than last GDDR5 column.

 

So we should start putting thermal paste between GPU memory and heatsink

 

Lower-end mobile GPU's will most likely have GDDR5X. I have no doubt we'll see it sometime next year. How much memory mobile chips have also affects how much power they draw. These desktop chips (which I believe they're referring to) have up to 48GB's of VRAM. Mobile chips will likely be limited to 8GB or 16GB. So, power consumption should not reach those insane levels.

Not sure if we'll see HBM in mobile chips next year. It's possible. No idea at this point. NVIDIA will tell us more at CES 2016.

 

Devs trying to get Vulkan to use all that VRAM:

 

This is my issue too. Although I used to be heavily into flight sims and racing sims, but those are just time sucks now that I have kids. I'm fine with a round or two of Battlefield 4 or occasionally a single player campaign that I can get in and out of after 20-30 minutes at a time. I just don't have the time, patience, stamina, or interest, or all of the above. I really want to get back into combat flight simulation though and those DCS sims are sitting restless on my hard drive. They need to be flown and I just don't have the time to do it.

 

Haha, size of textures!

 

With double the bandwidth of GDDR5 I cant imagine even the next flagship mobile gpu being bottlenecked by GDDR5X.

Especially if they make pascal more memory bandwidth efficient vs maxwell like maxwell vs kepler.

Sent from my SM-A500FU using Tapatalk

 

It will most likely be "good enough" for Pascal. As we all know, NVIDIA loves to dial back performance to refresh for some more $$$.

Seems all NVIDIA does these days is "good enough."

 

HBM vs GDDR5x is like oil on butter on bacon on grease for any Pascal mobile graphic card we get. Its really not needed imo

Right, so it surpasses GDDR5 in power consumption. There goes the power reduction argument which was one of the key reasons why it was made in the first place

 

I severely doubt we will see the M pascals before summer. Pascal M are probably will be launched in Q3-Q4 2016.

I think the fact nvidia launched the desktop 980 shows the fact they wont have a mobile card ready before Q2.

AMD is probably gonna somehow do a paper launch again and completely ruin their launch.

 

Q2 2016, Pascal mobile will be available.

 

Pascal chips listed in the recent drivers (hidden in DLL).
Looks like we get GP102 chip as well now.

My guess:
GP104/GP106/GP107 launch in Q1 2016. All three with GDDR5X.
GP102 and GP100 after summer sometime both with HBM2

http://www.3dcenter.org/news/reihen...cht-gp100-gp102-gp104-gp106-gp107-gp10b-gv100

 

Oh, GP102?

Does this mean ngreedia is done competing with themsleves? (GTX TITAN BLACK/GTX TITAN/780 Ti <--> 980 Ti/TITAN X)

 

If Nvidia wasn't going to have a new mobile chip before the end of next year, they would've stretched Maxwell out further, instead of just dropping the GTX 980.

They would've started with a full GTX 970, then something in the middle, then the full GTX 980 in Q1-Q2 of next year.

The fact that Nvidia just threw the full GM204 out there proves that they're confident in Pascal dropping early in the year.

 

Your guess is as good as mine.
It could be that GP100 is for Titan cards with good computation (more FP32 cores) while GP102 is pure gaming brand with more FP32 cores disabled.
Perhaps its just to fill in the blanks between GP100 and GP104 which could mean a mobile GP102 chip down the line.
Maybe its another chip with HBM2 since GP104 will launch when HBM2 isnt available

Not to mention there may not be a dual Maxwell chip for consumers after all. The only rumor we got is the vague article from wccftech which is notorious for writing stuff that never happens.
Fury X2 rumors however are all over the place.

So the question is; why no dual Maxwell chip to take up the fight against AMD? Are they preparing something else? Nvidia have been ahead of AMD with GPU releases for years now. If they had a dual Maxwell it should have been out by now.

Specific Pascal codenames in drivers seems to suggest they are getting ready imo

 

It would make a horrible mess on nomenclature side. No one would buy a 970 when there is a 980M. Honestly, full GTX 970 would perform WORSE then a 980M OC for sure. The chopped bus would put a 970 way behind a 980m at the same clock. Nvidia wouldnt have bothered with nearly as much hype/advertising for GTX 980 if they are testing out the water for a desktop card. I feel like they are trying to play the titan(Premium Product) card with the mobile market now. If that is so, nvidia would like to milk it as much as possible.

Q2 launch for Pascal cards are probably likely, which means Q3 launch for M pascal depending on the yield. Probably we will see a GP 104 soon. Q4 2016 or Q1 2017 for GP 100 is my guess, as nvidia is focusing on Quadro/Tesla to ship out. Most of the high end tesla are still on GKx10 so I doubt the consumer will get the high end chip anytime soon.

 

Another thing to think about:

Maxwell OpenCL capabilities was added to the driver in early February. GM107 launched the same month.
Pascal OpenCL capabilities was added to a driver in early November.

 

anyone have a guess on how much faster than pascal 1080m would be over a desktop 980?

 

60-80% w/ GDDR5X

 

No. Only presentations from NVidia which are always wrong. Ridiculously exaggerated.

For both NVidia and AMD the computational/processing power increases don't equate to increase proportionally gaming performance. Way I see it, 60-80% increase in power really means increasing 60 FPS to 80 FPS.

 

And this is under the assumption of 1080m=1080.

GTX 1080 probably have a 30-40% performance increase over 980.

 

Knew they would release them earlier. Everyone gave me crap about it, lol.

The 1080 will be far more than "30%-40%" over the 980. It'll probably be close to 50% over the 980Ti.

The 1080M will perform better than the GTX 980. There's no doubt about that.

 

Wccftech speculates 60-80% improvement: http://wccftech.com/nvidia-pascal-volta-architecture-gp100-4k-60fps-graphic-card/

 

any questions and speculations regarding performance numbers are just one thing at this point in time: fun, but in the end worthless thats why i keep cringing anytime a user asks a question like @Phase just did... difficult to hold back a rant

Sent from my Nexus 5 using Tapatalk

 

I just want to know if it would be worth holding off getting a 980 desktop gpu laptop, if the 1080m will blow it away

 

The 880M is still good for gaming, makes sense to wait for the Pascal in my eyes.

 

I agree. If you have 880m I'd hold off for Pascal.

 

Thats for GP100 though, GP 104 will be 1080.

Edit: and there is the rumor of a GTX 960 TI running on essentially a 970m config releasing in jan 2016 to counter 380x. If verified, would further suggest Pascal is further away. Maybe AMD would launch first again. AMD launched first on both 28nm and 40nm. Also AMD does have more experience with HBM and is probably using GF's 14nm process over 16nm FinFET from TSMC.

 

I think so too. I am hoping that Pascal will work with my M82x R2 and the P170SM-A. If not, it will be a good excuse to get a new laptop.

 

I'm not sure about the M18xR2, Maxwell isn't particularly compatible with it - throttling issues (and other quirks I understand), so you might expect Pascal to be the same or even less compatible. I haven't decided yet what I'm going to do when Pascal comes round - if there was some stroke of luck that it's compatible with my R3 then I'd probably buy a 1070M, or failing that build a desktop if my 670MX doesn't cut it anymore (does at the moment though!).

 

I am hoping. The M18x R2 is such a cool looking machine I want to keep using it. The 680M's work well and I have thought about going to 980M's, but with Pascal so close I will wait it out. The only game I have trouble with so far is ARK, which does not support SLI. ARK runs crappy on just about everything but the bleeding edge anyways.

 

As always, I am most interested in the mid tier performance.

If the 1060m > 970m that would make laptops like the new XPS 15 pretty rad next year.

 

God I hope the Pascal chips does not follow the GT/GTX 10x0M names.

I have hopes they will be called
Eagle
Beast
Slayer
Destroyer
World Eater

etc

 

I'd say there's a good possibility the 1060M will perform almost identical to the 970M. It's about 50% better, if I am not mistaken (depending on the game).

It would be interesting if they launched a new naming scheme. As long as it performs as we all hope, the name could be "BeatingAMD180M" for all I care.

 

My guesstimate;

1060M = 980M
1070M = 980 for notebook
1080M = 980M SLI

Sent from my SM-G925F using Tapatalk

 

That's what I'm hoping for, but I'm expecting a little bit less. Here's to hoping they surprise me.

 

1080m sli = 980m quad ?

 

980M quad-SLI would scale horribly, so no.

1080M should be like the 980Ti (performance wise) or a bit better.

 

1080 Quad = World Domination