[Rumour] TSMC will produce the huge GP100 interposer

[coercitiv]

So is the new P100 interposer 1200mm2?

 

[antihelten]

So is the new P100 interposer 1200mm2?

More or less yes.

Based upon this render and a bit of pixel counting, I get the size of the Interposer to be 1167mm2.

 

[Redentor]

So, he was right, if we use this image to calculate the size.

http://i.imgur.com/aZLFtA1.jpg

You, guys, are so dumb ... XD

Sounds like a bigtime pile of BS. No wonder with the source.

Boom.
This bits 'n chips guy seems to be pulling crap out of thin air that even the most cursory of sanity checks would prove to be false.

insulting other members is not allowed
Markfw900

 

[Creig]

"Might be possible" is the key word.

That's actually three words.

 

[Silverforce11]

Would be a good time for those naysayers to man up and admit they were wrong instead of the source.

 

[Adored]

The bitsandchips guy gets a lot right actually, he clearly has sources.

 

[cytg111]

Would be a good time for those naysayers to man up and admit they were wrong instead of the source.

That would imply men, good luck with that.


The Anti-ADF squad (whatever ADF stands for!)

Threadcrapping and trolling are not allowed
Markfw900

 

[HurleyBird]

Would be a good time for those naysayers to man up and admit they were wrong instead of the source.

/Shakes magic eight ball

Outlook not so good

 

[xthetenth]

That would imply men, good luck with that.

Classy.

The Anti-ADF squad (whatever ADF stands for!)

AMD defense force.

 

[raghu78]

So that rumour did turn out to be true. Nvidia really pushed TSMC 16FF+ and their interposer tech to the limits . P100 is a compute monster and Nvidia wants to blow Intel Knights Landing out of the water. Vega 10 vs P100 should be quite close for FP64 performance.

 

[Cookie Monster]

So that rumour did turn out to be true. Nvidia really pushed TSMC 16FF+ and their interposer tech to the limits . P100 is a compute monster and Nvidia wants to blow Intel Knights Landing out of the water. Vega 10 vs P100 should be quite close for FP64 performance.

Peak theoretical perhaps. But actual attainable rate would be a whole different story and one reason why the industry apart from a select minority doesn't use AMD's HPC products.

 

[raghu78]

Peak theoretical perhaps. But actual attainable rate would be a whole different story and one reason why the industry apart from a select minority doesn't use AMD's HPC products.

Nvidia's dominance of HPC is due to CUDA and more CUDA software/ecosystem related than due to AMD's hardware not being competitive. But OpenCL is picking up in adoption and AMD's market share in professional/HPC space reached > 20% with the launch of Hawaii.

 

[Silverforce11]

Nvidia's dominance of HPC is due to CUDA and more CUDA software/ecosystem related than due to AMD's hardware not being competitive. But OpenCL is picking up in adoption and AMD's market share in professional/HPC space reached > 20% with the launch of Hawaii.

~30% last time the tech press talked about it, due to the Firepro wins on Apple. It lead to a big increase in Firepro sales for AMD. That's basically from nothing in 2012. So they are gaining.

Intel is also gaining, with Xeon Phi shipping with >3TFlops FP64 without requiring a CPU to drive the accelerator cluster.

NV needs big Pascal Tesla ASAP.