'

Nvidia Announces Pascal Tesla P100 Volume Production

Nvidia Announces Pascal P100 Volume Production

Nvidia Announces Pascal P100 Volume Production

Nvidia Announces Pascal P100 Volume Production

 

Nvidia has announced Volume Production of their new Tesla P100 GPU, a 600mm square GPU based on the Pascal Architecture that offers 10.6TF of single precision GPU performance and utilises HBM 2 memory and NvLink technology. 

This new GPU is designed to be used for High-Performance Computing (HPC) and deep learning applications and is easily the most powerful GPU that has ever been designed by the company, with single precision (FP32) compute performance of over 10.6 Teraflops of performance or 21.2 teraflops of half precision compute performance.  

  Nvidia Announces Pascal P100 Volume Production

 

To put things into perspective the GTX Titan X has only 7 TFLOPS of single precision compute performance, making this new GPU from Nvidia much more powerful. 

This new GPU is built using TSMC's new 16nm FinFET transistors and has a die size of over 600mm squared, making this the largest FinFET GPU ever designed. 

This GPU will have 16GB of HBM 2.0 VRAM and it will be connected to the GPU using a 4096-bit memory bus, a memory bus that is the same size as the one used in AMD's existing R9 Fury X GPU.    

The Nvidia Tesla P100 has 3584 CUDA cores, which are clocked at 1328MHz base and 1480MHz boost, meaning that this GPU has more and much faster cores than Nvidia's older Maxwell GPUs. This GPU has a TDP of 300W.   

Below is a graph comparing Nvidia's Tesla P100 to their older Maxwell and Kepler based Tesla GPUs. 

 Nvidia Announces Pascal Tesla P100 Volume Production

 

With NVLink Nvidia is able to combine a large number of these GPUs together into a single system, allowing Nvidia to place 8 of these new GPUS into a single supercomputer, the Nvidia DGX-1, for a mind melting price of $129,000.

Nvidia's NVLink allows Nvidia to connect their GPUs with 4x the bandwidth of PCIe 3.0, allowing Nvidia to easily interconnect a staggering amount of compute performance into a single computer. 

 

Nvidia Announces Pascal P100 Volume Production  Nvidia Announces Pascal P100 Volume Production

 

Right now Nvidia has not made any announcements regarding and Gaming GPUs using this new GPU, with this GPU first being used for supercomputing and enterprise applications before it can reach consumers. 

The die size of the Nvidia P100 chip is very similar to the GTX Titan X, making this GPU core a very likely contender for a future Titan-grade GPU. 

Nvidia has announced that this GPU is currently in volume production, meaning that this chip is not only being manufactured but being manufactured in quantities where Nvidia can actually make and sell products based on it in the near future. 

 

You can join the discussion on the Nvidia Pascal P100 high-performance Pascal GPU on the OC3D Forums

 

  

«Prev 1 Next»

Most Recent Comments

05-04-2016, 14:04:45

Dicehunter
1 step closer to the consumer version Quote

05-04-2016, 14:23:11

wrenaudrey
I've just watched the live stream, and there's no reveal of the GTX 1070 nor the 1080?
It wont be revealed or... it may be in the remaining days of the conference?Quote

05-04-2016, 14:24:14

Dicehunter
Quote:
Originally Posted by wrenaudrey View Post
I've just watched the live stream, and there's no reveal of the GTX 1070 nor the 1080?
It wont be revealed or... it may be in the remaining days of the conference?
Nvidia will most likely do a standalone "GTX" type conference for the gaming orientated versions.Quote

05-04-2016, 14:25:19

wrenaudrey
Quote:
Originally Posted by Dicehunter View Post
Nvidia will most likely do a standalone "GTX" type conference for the gaming orientated versions.
Which is coming these next few days?
I've just looked at their time table and there's a keynote tomorrow and thursday at 11am PDT/7pm BST. It may be one of these?

EDIT: First time following an nvidia launch tbh, so dont really know how this goesQuote

05-04-2016, 14:27:06

WYP
One thing that needs to be remembered is that Deep learning etc are big earners for NV. Which is why they can sell to them so early in production. It will be ages before we see this GPU (big pascal) release in a GTX gaming product.

This GPU will almost certainly be the future Nvidia Titan class GPU, but normal pascal will have to come to gamers first, just like with kepler and Maxwell.

I have updated the article with more information, including clock speeds and memory capacity.Quote
Reply
x

Register for the OC3D Newsletter

Subscribing to the OC3D newsletter will keep you up-to-date on the latest technology reviews, competitions and goings-on at Overclock3D. We won't share your email address with ANYONE, and we will only email you with updates on site news, reviews, and competitions and you can unsubscribe easily at any time.

Simply enter your name and email address into the box below and be sure to click on the links in the confirmation emails that will arrive in your e-mail shortly after to complete the registration.

If you run into any problems, just drop us a message on the forums.