Nvidia Pascal Speculation Thread

[Alessio1989]

Another wish: no more CB exotic limits with Pascal.

 

[itaru]

https://forums.geforce.com/default/...a-with-denver2-cpu-amp-integrated-pascal-gpu/

 

[iMacmatician]

Given the (lower bounds on) memory bandwidths of the Tegra and the discrete GPU, could we approximate the ratio of Tegra GPU FLOPS to discrete GPU FLOPS?

 

[Ailuros]

Given the (lower bounds on) memory bandwidths of the Tegra and the discrete GPU, could we approximate the ratio of Tegra GPU FLOPS to discrete GPU FLOPS?

IMHO it's 3GFLOPs per dGPU and 1GFLOP per Parker SoC.

 

[Kaotik]

Curious, why are they using PCIe instead of NVLINK there?

 

[xpea]

First blurb:

QUANTA QCT ANNOUNCING X86 SERVER WITH NVIDIA PASCAL NVLINK SUPPORT
Here at NVIDIA GTC 2016 something caught my eye immediately, we are seeing our first servers supporting NVIDIA Pascal NVLink.
Pascal is still not released, although we will be covering Jen-Sun Huang’s keynote tomorrow and have high expectations. We found a nice piece of marketing material in our bag and noticed that QCT is advertising NVIDIA Pascal NVLink technology support.

More here:
http://www.servethehome.com/quanta-qct-announcing-x86-server-nvidia-pascal-nvlink-support/

 

[xpea]

Pascal Deep Learning performance claimed by Nvidia:

Source: http://www.hardware.fr/news/14574/gtc-deep-learning-70-pascal.html

 

[xpea]

Supermicro is coming to the party too:

Source: http://www.computerbase.de/2016-04/supermicro-high-end-gpu-server-mit-pascal-gpu-noch-dieses-jahr/

 

[Kaotik]

First blurb:

More here:
http://www.servethehome.com/quanta-qct-announcing-x86-server-nvidia-pascal-nvlink-support/

It's quite surely just between the Pascals, neither AMD or Intel has said anything about supporting NVLINK with their CPUs or chipsets

 

[CSI PC]

It's quite surely just between the Pascals, neither AMD or Intel has said anything about supporting NVLINK with their CPUs or chipsets

Yeah,
and also the supercomputers with CPU.
I think the NVLINK will be used in the IBM-NVIDIA collaborations as a complete architecture.
Cheers

 

[xpea]

nevermind

 

[Ext3h]

The shared memory size has been reduced, though.

Shared memory no longer just being a slice of L1? About time for that...
Means Pascal might actually allow mixed graphic/compute loads now.

 

[BRiT]

Nvidia Pascal Announcement Details: https://forum.beyond3d.com/threads/nvidia-pascal-announcement.57763/