OMG Nvidia "Pascal" 1000-series, as soon as next year?!

[Atomicnerd64]

If they remember to add A-sync compute I guess it might be a bigger improvement but I am strongly suspecting we will have another Thermi incident

"THERMI" I can't control my laughter hahahahaha

[Mikjul]

If they remember to add A-sync compute I guess it might be a bigger improvement but I am strongly suspecting we will have another Thermi incident

Fermi was never an accident IMO... it was miles ahead of the competition up until the Radeon 7k series hit, not just gaming performance but also GPGPU.

[Metro]

Just to let you know, it will not be Q1 2016, it will be Q2 2017 for high end Pascal

[hypnodog]

Just to let you know, it will not be Q1 2016, it will be Q2 2017 for high end Pascal

That would be for GP2xx. GP1xx will certainly have high end models.

[don_svetlio]

Fermi was never an accident IMO... it was miles ahead of the competition up until the Radeon 7k series hit, not just gaming performance but also GPGPU.

you do realize thermi ran hotter and slower than Terra Scale, right?

EDIT: http://www.techpowerup.com/gpudb/268/geforce-gtx-480.html and https://www.techpowerup.com/gpudb/253/radeon-hd-5870.html

[Sauron]

I wouldn't say they got greedy, sure they were saving money by lowering GK110 output but it was a good decision. No point is selling a 550mm^2 die when a 398mm^2 die has similar performance. They've been releasing mobile flagships as flagships since the 680. 

 

How does the 980ti have similar performance to the 980? The 980 should have been the 970 and there should have been a cut down 980ti called the 970ti to fill the gap between the 970 and the 980 (which would have been what is the 980ti). Then we could have had a 990 or a titan x with custom coolers and a better price. It was a good decision only for their wallets, we (the consumers) only lost from this.

[Mikjul]

you do realize thermi ran hotter and slower than Terra Scale, right?

EDIT: http://www.techpowerup.com/gpudb/268/geforce-gtx-480.html and https://www.techpowerup.com/gpudb/253/radeon-hd-5870.html

[Deletive]

That would be for GP2xx. GP1xx will certainly have high end models.

There is no secret that 2016 Pascal is their mobile die. It'll be a generation faster then what it was before but nothing to surprising when you switch to 16nm and ~/~ 50% more die space (too lazy to do the actual math) 

 

How does the 980ti have similar performance to the 980? The 980 should have been the 970 and there should have been a cut down 980ti called the 970ti to fill the gap between the 970 and the 980 (which would have been what is the 980ti). Then we could have had a 990 or a titan x with custom coolers and a better price. It was a good decision only for their wallets, we (the consumers) only lost from this.

 

 

You didn't understand the post, you were talking about the 980 and 780ti, so was I. The 980 was the flagship for the first part of 2014 till the 980ti came out. Once the 980 was released there are no need for the bigger kepler GK210 chips because they had similar/better performance then the bigger kepler chips while having a 150mm^2 smaller die and consuming less power.

 

As I said it should be no surprise the Nvidia has gone with the GX104/GX204 die as the flagship then realize the bigger server die a few months/year later. There probably aren't going to be any consumer Volta chips so I don't think this is going to change until atleast 2019. 

980 was ~25-35% performance increase then the 780 (which by the way has been that way for a long time for both AMD and NVidia) so GM204 there is no "we could have had" They have Target margins, they released a card that reach those margins. But I understand how we "lost" considering they gain more money for each Maxwell Gm204 chip with the much cheaper cost of resources but it also the reason we even have a 970 at 340$.

[don_svetlio]

[Mikjul]

Yeah luanch drivers vs a matured card hehe, we all know after the adverse launch of the 480 it absolutely mutilated the HD 5000 series, in some games OC'd it surpassed the 2 GPU 5970.

 

and i am a guy who skipped Fermi with a 6870.. LOL

[don_svetlio]

[don_svetlio]

Yeah luanch drivers vs a matured card hehe, we all know after the adverse launch of the 480 it absolutely mutilated the HD 5000 series, in some games OC'd it surpassed the 2 GPU 5970.

 

and i am a guy who skipped Fermi with a 6870.. LOL

[Mikjul]

What are you trying to disprove?

 

Not sure if you ever owned a Fermi GPU or just stayed away from enthusiast sites like Overclock.net.. Fermi when OC'd raped OC's 5800/6900 series GPU's... high power and heat is the only argument

[don_svetlio]

What are you trying to disprove?

 

Not sure if you ever owned a Fermi GPU or just stayed away from enthusiast sites like Overclock.net.. Fermi when OC'd raped OC's 5800/6900 series GPU's... high power and heat is the only argument

Fermi when OCed cooked beef jerky - a friend owned a 480 - 102*C on the core - didn't even last a year.

[Mikjul]

Fermi when OCed cooked beef jerky - a friend owned a 480 - 102*C on the core - didn't even last a year.

I have a friend rocking a 480 since launch, it reaches 86c max in the hardest games she runs... seriously 86c is nothing for a GPU anyway.

[don_svetlio]

I have a friend rocking a 480 since launch, it reaches 86c max in the hardest games she runs... seriously 86c is nothing for a GPU anyway.

Well - my friend's burned cad disagrees. could be a faulty card or maybe not - is hers the reference cooler?

[Mikjul]

Well - my friend's burned cad disagrees. could be a faulty card or maybe not - is hers the reference cooler?

EVGA SC edition yes.

 

Exactly this one..

 

[don_svetlio]

EVGA SC edition yes.

 

Exactly this one..

 

-snip-

[Mikjul]

Well all i can say... testing methodology...

 

[don_svetlio]

Well all i can say... testing methodology...

 

Well - open test benches usually have lower temps - when I open my case I get 57*C in valley - soon as I close it - boom - 70*C.

All I can say is your friend is one lucky girl if her 480 maxes out at 86*C

[themaniac]

i'll believe it when i see it

[evilarceus]