Pascal NVLink feature - possible use for GPU rendering?

[linvanchene]

Updated / Edit:

Nvidia released the GTC 2016 opening keynote as video series on their youtube channel

http://www.youtube.com/user/nvidia/videos

first part:

http://www.youtube.com/watch?v=ddBIF1fnvIM

- - -

Source for the following quotes and images:

https://devblogs.nvidia.com/parallelfor ... de-pascal/

- - -

Support for the GPU ISA means that programs running on NVLink-connected GPUs can execute directly on data in the memory of another GPU as well as on local memory

Pascal GP100 adds features to further simplify programming and sharing of memory between CPU and GPU, and allowing easier porting of CPU parallel compute applications to use GPUs for tremendous speedups

With the new page fault mechanism, global data coherency is guaranteed with Unified Memory. This means that with GP100, the CPUs and GPUs can access Unified Memory allocations simultaneously

pascal_8_unified_memory-300x213.png

4-GPU-CPU-Quad-206x300.png

NVLink

http://www.nvidia.com/object/nvlink.html


- - -

What does this all mean for GPU rendering?

Which are the CPUs that support NVLink?
Will we be able to connect consumer level Pascal cards and CPUs with NVLINK already in 2016?

- - -

[Tutor]

Source for the following quotes and images:

https://devblogs.nvidia.com/parallelfor ... de-pascal/

- - -

Support for the GPU ISA means that programs running on NVLink-connected GPUs can execute directly on data in the memory of another GPU as well as on local memory

Pascal GP100 adds features to further simplify programming and sharing of memory between CPU and GPU, and allowing easier porting of CPU parallel compute applications to use GPUs for tremendous speedups

With the new page fault mechanism, global data coherency is guaranteed with Unified Memory. This means that with GP100, the CPUs and GPUs can access Unified Memory allocations simultaneously

pascal_8_unified_memory-300x213.png

4-GPU-CPU-Quad-206x300.png

NVLink

http://www.nvidia.com/object/nvlink.html


- - -

What does this all mean for GPU rendering?

Which are the CPUs that support NVLink?
Will we be able to connect consumer level Pascal cards and CPUs with NVLINK already in 2016?

- - -

It appears that initially only IBM (and possibly ARM64) CPUs can take advantage of NVLink which accounts for, at least, half of that predicted 10X speedup prediction for Pascal.

[glimpse]

hopefully one day this "F1" grade technology would come down to earth & get's integrated into "familly cars" as well =).. for now that novel piece of technology is reserver for GPUs taht are sold for 10X price =p