Where is the BIG fully enabled Pascal chip?

[i_build_nanosuits]

Kepler Titan Black (GTX 780ti) :

- 551mm² , 15/15 SMX enabled , 2880 cuda cores (192 cuda cores per SM)

 

Maxwell Titan X

- 610mm² , 24/24 SMX enabled, 3072 cuda cores (128 cuda cores per SM)

 

Pascal Titan X / GTX 1080ti / Tesla P100 (64 cuda cores per SM)

- 471mm² , 56/60 SMX enabled, 3584 cuda cores

 

Questions:

-Where is the BIG Pascal chips (550mm²+)

-Where is the FULLY ENABLED pascal chip (60/60 SMX, 3840 cuda cores)

 

Nvidia could potentially have a MUCH more powerful Pascal chip...if they'd make it 600mm² like the previous gen monster, at 14nm they could probably pack around 90 SMX units containing 64 cuda cores each and potentially get a GPU with around 5760 cuda cores which would make it roughly 60% faster than a pascal titan X/ GTX 1080ti

 

 

[samcool55]

Aren't they in the server stuff thingy? I don't know...

Also i'm assuming the titan XP isn't a complete chip because of yields, i really don't see another reason why else they wouldn't fully enable the gpu in the titan XP...

 

[nobiggieBIG]

2 minutes ago, i_build_nanosuits said:

Kepler Titan Black (GTX 780ti) :

- 551mm² , 15/15 SMX enabled , 2880 cuda cores (192 cuda cores per SM)

 

Maxwell Titan X

- 610mm² , 24/24 SMX enabled, 3072 cuda cores (128 cuda cores per SM)

 

Pascal Titan X / GTX 1080ti / Tesla P100 (64 cuda cores per SM)

- 471mm² , 56/60 SMX enabled, 3584 cuda cores

 

Questions:

-Where is the BIG Pascal chips (550mm²+)

-Where is the FULLY ENABLED pascal chip (60/60 SMX, 3840 cuda cores)

 

Nvidia could potentially have a MUCH more powerful Pascal chip...if they'd make it 600mm² like the previous gen monster, at 14nm they could probably pack around 90 SMX units containing

64 cuda cores each and potentially get a GPU with around 5760 cuda cores which would make it roughly 60% faster than a pascal titan X/ GTX 1080ti

 

 

quadro pascal.

[i_build_nanosuits]

1 minute ago, nobiggieBIG said:

quadro pascal.

that answer one of the question...thanks for that yeah the quadro P6000 is a fully enabled pascal chip...but it's still ''small''

 

[Guest]

The real question is... Where is the BIG Vega chips??? Shouldn't they be out by now? 600mm²+ with HBM2 mhmmhmmhm.

[i_build_nanosuits]

1 minute ago, FTL said:

The real question is... Where is the BIG Vega chips??? Shouldn't they be out by now? 600mm²+ with HBM2 mhmmhmmhm.

AMD...come on...they will launch that 300W behemoth and it will still be slower than a slightly overclocked GTX 1080 in games...a mid-range nvidia card...and they will be 8 months late to the party and they will sell it for the price of a 1080...