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

95C target, vbios programmed much lower to drop out of boost and an encrypted vbios idea has been passed around the table.

 

vbios encryption! bastards...

 

Nothing RSA, just basically a way to stop dumb people from frying their cards. They haven't had to worry about that since Fermi........

 

ok, then we just need to come up with a lil bot net with a bunch of NBR enthusiast raw power machines, do some brute force number crunching, hack that ecnryption and show them whos boss....

 

Do you think anyone here would be able to get around the encryption, perhaps maybe Prema?

 

I'm pretty sure that Prema and svl7 already cracked it ages ago when they cracked the nVidia flash tool... But if they actually put the current spin into a laptop right now..................... They've moved some transistors around to even the heat but they're waiting for the tape out to actually come back...... Its holiday season right now... There are not only mail delays but bigger customers than nVidia. Just like the 880M, it would have made an amazing desktop chip but they took those amazing 680 cores and undervolted them so they could push the clocks. If they shove Pascal as it is now into a laptop, it will be 880M all over again. High ASIC, high leakage chips... Refreshing Maxwell and waiting until 2H 2016 would be the smart way to go at this point, especially since they have to get manufacturers besides Clevo, who will find a way to shove anything into anything, on board. They currently have two companies that told them where to stick Pascal for the time being.

 

that made me giggle a bit

ok, so how abouuut....mobile pascal as clevo exclusive and maxwell refresh for all those bga jokebook manufacturers that cant take the heat?

Sent from my Nexus 5 using Tapatalk

 

I agree with you. What if that happened?

That's literally a question, not a tease.

 

phsew...in that case i really hope that clevo will be up to the production task, cuz customers would totally crash the clevo vendors shut up and take my money!

Sent from my Nexus 5 using Tapatalk

 

Clevos actions lately show that they want to overtake just being an ODM.... They clearly want to make a name for themselves.

What I can say is that currently, the cooling system in the P870DM is the only cooling system that can handle Pascal without *thermal* throttle.

 

Oh by the way nVidia. I fall not only under the judgment proof group, I fall under ADA mental health protection. Figured I would save you the time. You can bring me to court... Except my mental health disability precludes me from leaving the house. You could sue me through a virtual means but my mental health status won't allow you to get anything except my name and address. Being protected by the ADA under bipolar law, you can not force me to do or say anything. That's per my attorney. By making this post I have now admitted to having the disability that exempts me from any type of interrogation. Saves you time, saves me time. I met my contact when I was in the middle of a sudden manic episode and it was quite thoroughly documented that I lost my mind. If you're really that scared about my knowledge you can seek an injunction but my public admission of my issues can't stop what other people tell me. You also have no legal grounds to make this forum tell you anything because the owners, users, and affiliates, of NBR and their sister sites, are in fact protected by my ADA status. Long story short, taking me to court would be fruitless, fix your own leaks.

 

show 'em

btw, 200W tdp cooling required to avoid thermal throttling? oh my...

Sent from my Nexus 5 using Tapatalk

 

Wait, what? Refreshing Maxwell? Why would they do that? What would they call the line? Did I miss something?

 

The problem with a maxwell refresh is that they showed their hand already. GTX 980 is at its limits. Titan X is at its limits. Their only choices with a refresh are to even out the line. Turn the 980 into a 1060Ti, turn the 980M into a 1060, turn the 970M into a 1050Ti, revamp the 960 to have a 192-bit memory bus and call it a 1050. Make a card with ~2400 cores and call it a 1070, shift the 980Ti to a 1080, and turn the Titan X into a 1080Ti, drop the titan name or add double precision, and call it the Titan X Black.

But that just makes their line more fleshed out and robust and fills the gaps. It doesn't bring anything actually new to the table, except maybe a DP-enabled Titan X Black actually earning its $1000 price tag, and an in-between card between the 980 and the 980Ti.

But if Ethrem is correct, and Pascal is hot like Kepler or even like Fermi, this means everything I told everybody was right: Maxwell is the exception to the rule with how unnaturally cool it is. Things are going to change in the future, and people accustomed to 970M/980M in superthin notebooks like Razer Blade or Gigabtye's line are going to be set back to low end crap like 860Ms. It doesn't matter how good the low end cards are, because the minute higher end stuff come out, they become difficult to game well with, and settings need turning down. And they'll still be expensive.

 

What makes you say that (in relation to the 'more cores at lower frequency' that I was talking about)? Are you thinking they need to wait to improve yields before they launch if they need 'big dies' with more cores?

EDIT: and after reading all the remaining posts in this thread, I'm thinking you know this from your source you talked about. But is the yield thing the reason for the delay?

 

You cant really modify an entire architecture like that. You dont simply add DP to an existing die, you need to completely redesign everything more or less. You cant revamp a die to give it more bus/cores/etc. You need to literally change everything.

 

I don't understand how Pascal can run so hot, after learning from Maxwell, unless Nvidia is that stupid not to have employed similar techniques in the architecture to allow it to run cool. It can't be a byproduct of the process since the iPhone 6s also uses 16nm finfet in a fanless application and the A9 is 50% faster overall performance than the A8 which is 20nm, and A9 also runs a faster clock. I don't know where all this info is coming from but anything at this point in time is from early silicon and engineering samples to evaluate performance and not necessarily tuned for best thermal application. Requiring 200W TDP for a top end card in the mobile market would be suicide especially considering the general desire for smaller, thinner, lighter laptops. The GTX 980 is a bastard child that doesn't conform with MXM 3.0b. This would be regression in the eyes of the general consumer.

 

The P870DM was aqtually the laptop I was thinking of getting. So this beast would be able to handle the heat that Pascal currently suffers from?

 

A9 is likely using a low power optimized process. They usually are avalible first before high power processes. A small die is always going to be much easier to build larger die. If you want an example of bad finfet chips, desktop broadwell should be an example.

 

I agree with this statement. While it's impressive they were able to do this, but I still want a lower power, cooler GPU for my laptops. Even cooler and more efficient than the 980M, I want to see progress in this direction.

 

That ladies and gentlemen, is the post that wins my personal "Best post of 2015" award.

 

P870DM, damn I want one. I need to stop buying so many laptops though, it's becoming a bit of a problem as you can see.

 

Simple. Sell all your current ones and buy two P870DMs.

Then give me one.

Then you'll be so annoyed at that, that you'll not buy a new one until you need one.

Somehow, this makes sense. It better, because I want a P870DM with two GTX 980s. =D

I know you can't. But that's not my problem, it's theirs. I was simply saying that the only way a refresh would make sense is if they did something like that.

 

Do they even do SLI 980 gtx? I would be down for one but all the websites only offer the sli 980m option

 

It's not an official clevo configuration, but from what we know, Eurocom may offer it, if only for a limited time.

I'll gladly be using that. I don't even care if I'll need two power bricks at all times =D.

 

Been there, done that.

 

But there is an SLI of GTX 980, the MSI GT80S 6QF. Even though it is a lesser TDP, i am still interested in its performance figures.

 

Those are far lower TDP yes. But the ones I'm referring to are the 180W models that'll likely be going in the P775DM1, not the 120W variants of the GT80.

 

Oh I was not aware of SLi in P775DM1. Thanks for the info.

 

We are all waiting for it to ship and get benchmarked...

It's not available for order/pre-order, but it has been announced by MSI.

 

Considering the performance of the GTX 980m vs GTX 980 200W what are they trying to accomplish with the GTX 980 150W and 180W versions because they will likely not be any faster than the GTX 980m. Maybe because they're unlocked overclockers?

 

It remains to be seen what they will run like with enough power delivery - dual 330w was needed for OC'ing CPU/GPU with even 980m SLI.

If the power delivery can't meet the needs of the OC 980, then yes, it's likely there will be less difference between the 980m SLI and 980 SLI than should be possible - about 30% improvement.

 

indeed, the different TDP versions of the 980 should only differ slightly in their stock performance numbers. whatll give u true advantage will be the installed power phases in order to have more overclocking room power-wise

my take is around 25-40% performance boost 980 vs. 980M depending on TDP level, that would then give u, at least in theory, 50-90% performance boost 980 SLI vs. 980M SLI. but since only the lower TDP versions have SLI capability that number would rather be in the range of +50-60% id say... (best case scenario with perfect scaling)

 

I never said there was SLI in the P775DM1. I said, the cards going in SLI in the P870DM will likely be the 180W cards that are going in the P775DM1. Not the 120W cards for the GT80.

 

yeah I red your post again. Now i feel stupid (well more than normal).

 

thats ok, dont feel bad. ultima has this effect on people

 

I really do.

 

Aren't the 180W ones also a bit bigger than regular mxm? Like at the back I think? Would they still fit as SLI in the P870 is what I wonder

Also have we seen 120W yet? As in, are they regular sized mxm? For a P751ZM perhaps

 

I thought there was going to be 150W, 180W, and 200W. 150W and 180W apparently fitting the MXM 3.0B standard slot size. 120W would be a 980m

 

Notebookcheck said left one is 180W and if I remember correctly, someone said that's not exact mxm. http://www.notebookcheck.net/fileadmin/_processed_/csm_gtx_980_II_a771f34d57.jpg

150W is in the smallest one? Then I'd like to know if that's mxm

 

Well here's my 980m vs 980 in the P870DM-G, which looks like the 180W is slightly longer and very slightly wider than the MXM 980m.

 

180W 980 that fits in the P775DM1 is MXM-sized for the most part, but a bit longer.

150W fits in the GT72, and looks... weird. Like ASUS-mxm-card weird.

 

and then we have the mythical 120W(?) cards in the msi gt80, which no one has actually seen yet...

Sent from my Nexus 5 using Tapatalk

 

So there is a (possible) 120W one
Time to start believing in unicorns for this mythical beast

 

But like I said, why bother with 120W? The 200W 980 is about 35-40% improvement over 980m. A 150-180W would be about 20-25% faster, and 120W would be about same as 980m.

 

if its the same as 980m I at least hope it pushes 980m's price down further
And even if its only 10% better than 980m (and mxm3b) I still might consider it in the future.
Even at a higher price point if it turns out that this will be the last GPU for my system.
From what I hear in the pascal thread is that thats gonna be very hot and its giving me doubts whether it will fit into my P751.

edit: nvm, this is the pascal thread

 

Take all comments about Pascal with a super tiny grain of salt. It's all speculation and hearsay. Until we get actual review parts and reports, hold all reservation. We don't know what's out in the wild now.

 

Of course. Nothings set in stone. But more options is always good for the consumer .
Seeing how 200W and 180W are not fitting, Im happy for anything coming out that will still fit regular mxm, even a 120W version .

 

mmmm i wouldnt say that. TDP envelope doesnt have to go hand in hand with performance numbers, it only gives u the maximum of electrical power and heat u can saturate before u run into limits.

notebookcheck already showed that the 150W MSI GT72 variant and the 180W P775DM variant give u the same stock performance as the 200W model in the phoenix, so that goes to show that at stock settings, those TDP envelopes arent maxed out yet. when pushing clocks to the makes, the TDP variants would then more likely be separated by clear differences id presume...

remember: even at 120W u still have 33% more gpu cores in the die than with the 980M

 

TDP is really just a measure of how good the cooling system is instead of power consumption. TBH, I doubt the 33% core clock would make more then 10% difference at same clock. Its more of the higher VRAM clock thats affecting things.