Discussion: Nvidia Pascal

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
I may or may not be the only deep learning engineer here. The thing is, Pascal, from the get go has been made with Deep Learning in mind. In layman's term, its a breakthrough in the field of Artificial intelligence 5/7 years back. Scientists, in order to get faster computing power, started using the GPUs. See how facebook uses 8 Tesla m40 in their Big sur, http://arstechnica.com/information-technology/2015/12/facebooks-open-sourcing-of-ai-hardware-is-the-start-of-the-deep-learning-revolution/

Baidu, a chinese company, is also heavily invested in gpu cluster based supercomputing. ake a read here..http://www.nextplatform.com/2015/12/11/inside-the-gpu-clusters-that-power-baidus-neural-networks/ Titan x is the cheapest card for this and instead of tesla m40s, they have taken Titan x as their choice. Each server has 8 gpu in them. These machines run for weeks just to solve one problem. That time can be reduced to days, even hours if we can have better interconnectivity between the cards.

But main issue lies in the interconnectivity between multiple cpu nodes with 8 gpu each. For a single pc with 4-8 gpus, this isn't a issue, but when adding up more than one pc, Bottlenecked by bandwidth issue kicks in. This has been worked around using products from mellanox. But having that nvlink between cards will let us bypass all that. It is truly remarkable. T
 
I think 900 series will go dirt cheap when pascal comes out. So if you are gonna build a pc now, better to wait few months. I have high expectations for pascal, probably in summer they will release few units but at the end of 2016 maybe they will release the "Ti" versions of whatever GTX 1080. They do have dumb naming sense. But well it could outperform the two 980 Ti in SLI. 10 Times faster than maxwell then why not. 4k Gaming will be a breeze.
 


I doubt that. If history is any indicator then you will not see much of a drop in the new cards. You will see a hit in the used cards on ebay as the market will get flooded with people selling their old 900s but even that will level off over a while.

As for the performance, I hope. When the Maxwell came out it was not supposed to be much faster than the Keplars just lower power specs. The thing was, these cards where way better at overclocking which gave them the ability to blow the Keplars out of the water. I am hoping that the Pascal is going to be a major upgrade where they were focused on the speed and not just the power consumption. If that is the case then you could be right. The only other issue is to get the game developers to leverage these abilities. A three monitor 4K Nvidia 3d surround at 100+ FPS would be the cats meow. (Am I just dreaming)
 
We could be dreaming and just be getting our hopes up but well this year AMD's Polaris GPUs is also coming out. Wonder how they are gonna compete with Nvidia's pascal GPUs...I am excited about both.
 
The other thing I would like to see is the multi GPU cards come back. The last one was the Titan Z and the GTX 690. I would love to see a 1090 card but yes I know they are very impractical and the bang for the buck never justifies these cards. Maybe a Titan Alpha or Millennium 😛
 


Oooo, I like "Millenium". Or is it too much like the 90s?
 
http://techfrag.com/2016/03/08/amd-expected-to-launch-dual-gpu-r9-fury-x2-card-at-march-14-capsaicin-event/

That never says AMD said it is March 14. They just have some live cam going on that day. I believe it'll be like a week later.
 
He's right:

http://techfrag.com/2016/03/08/amd-expected-to-launch-dual-gpu-r9-fury-x2-card-at-march-14-capsaicin-event/

http://wccftech.com/amd-r9-fury-x2-dual-gpu/

EDIT: Just like turkey scratch said, if you mean it being launched as in produced then no. But you are right if your saying AMD is going to start advertising it.
 
If they can optimize the multi GPU graphics card for the 3d goggles there could be a market for multi GPU cards. I heard rumor that they were already looking into this where they dedicate one GPU per eye.

https://linustechtips.com/main/topic/92851-oculus-rift-with-slicrossfire/
http://www.reedbeta.com/blog/2014/04/03/vr-and-multi-gpu/
 
Not 300w but 375w instead (which is the standard for most recent dual gpu before AMD break that barrier once again with 295x2 which rated at 500w). While it will consume much less than 295x2 the performance was not significantly faster than 295. Many expect Fury X2 to be 16teraflops monster but looking at recent leak it seems amd decided to downclock those fury and keep the power consumption muxh lower than 295. They probably learn something with that 500w 295. If you pay attention about it amd heavily market FuryX 2 as VR solution not so much about being fastest gpu on the planet.
 
I can't seem to find an answer anywhere, but is Pascal supposed to support Async? I've seen much of the drama with Maxwell and Async, but can't find anything pascal related.
 


we will know about it when nvidia officially launch pascal. other than that speculation at best. nvidia for their part will not going to talk much about their gpu (especially their new feature) until they officially launch their card. sure they talk a lot about pascal this past two years but so far it's only about GPGPU. not gaming aspect of pascal.
 


The Titan X is can be up to approximately 9% faster in some games and resolutions...

The gtx 980 was only slightly faster than the 780 ti and Titan Black at launch however with the newer games being better optimized for Mawxell. The 980 is a much better card than the flagship keplers today.

It's possible that history will repeat itself and Nvidia will release a slightly more powerful and more power efficient small Pascal... I hope they release BIG Pascal with HBM 2 to justify my upgrade from a Titan X...

Hopefully on April 5 the Nvidia CEO with let us know the details about Pascal & hopefully the card will be more than 5 to 9% faster than the 980 ti...

 
Status
Not open for further replies.