nVIDIA announces unified memory for CUDA

[nicehat]

 

CUDA woulda shoulda. nVIDIA has announced that the new CUDA 6 release will feature unified memory support. This will allow programmers to access any resource or address within the legal address space, regardless of which pool the address actually resides in, and operate on its contents without first explicitly copying the memory over. 

 

Basically it makes programming simple by removing the need for programmers to worry about memory management as it is now automatic. This in turn makes coding more programmer friendly while also increasing productivity and code development. 

 

The funny thing is, is that this news is coming ahead of nVidia's Maxwell GPU architecture release. 

 

"From what NVIDIA is telling us they developed the means to offer a unified memory implementation today entirely in software, so they went ahead and developed that ahead of Maxwell’s release. Maxwell will have some kind of hardware functionality for implementing unified memory (and presumably better performance for it), though it’s not something NVIDIA is talking about until Maxwell is ready for its full unveiling. In the interim NVIDIA has laid the groundwork for what Maxwell will bring by getting unified memory into the toolkit before Maxwell even ships."

 

"Wrapping things up, NVIDIA will be showing off CUDA 6 and the rest of their announcement at SC13 next week. Meanwhile we’ll be back on Monday with coverage of the rest of NVIDIA’s SC13 announcements."

 

 

What do you guys think? Think this will make optimization easier and more smooth as games come out the pipeline if devs jump on this boat? You think this is a foreshadow for things to come as Maxwell may provide the hardware necessary to harness this? 

 

 

http://www.anandtech.com/show/7515/nvidia-announces-cuda-6-unified-memory-for-cuda

 

[Lord Pantaloons]

mmm Volta

[NohbdyKnows]

Volta pls

 

 

Stacked dram sounds nice

[ivhyn]

I can't wait 'til 2016

Maxwell will bring some nice energy efficiency nontheless.

[Kuzma]

Looking forward to Volta too tbh :3 but I'm most looking forward to Maxwell and Project Denver

[JLCitadel]

Looking forward to Volta too tbh :3 but I'm most looking forward to Maxwell and Project Denver

 

 

For anyone curious about it

 

Project Denver: http://en.wikipedia.org/wiki/Project_Denver

 

Maxwell: http://en.wikipedia.org/wiki/GeForce_800_Series

 

Volta: Not much really known, other then it's expected Mid to Late 2016 Early 2017

[TheMissxu]

I hope this pushes AMD to make OpenCL more accessible.

I'm torn with CUDA. It's a closed system, but it is widely used and fast enough...

[nicehat]

I hope this pushes AMD to make OpenCL more accessible.

I'm torn with CUDA. It's a closed system, but it is widely used and fast enough...

 

OpenCL will not get any attention as they have their hands full with another API thats acting like its the center of the earth....oh wait

[Immort]

Unified memory access for CUDA is of no benefit to the PC or gaming in general, so it's not really something for enthusiasts to salivate over.

It's of benefit to Nvidia APUs, i.e. Tegra processors which unfortunately have no place in the desktop or laptop market, so what this really does is improve tablet and smart phone performance.

[Kuzma]

OpenCL will not get any attention as they have their hands full with another API thats acting like its the center of the earth....oh wait

well it's not like they haven't managed to goet Adobe and a lot of other rendering software companies on board from the CUDA acceleration.

It's like I always say, why use proprietary when you CUDA used OpenCL?

[nicehat]

why you proprietary when you CUDA used OpenCL?

 

CUDA woulda shoulda its all hindsight now!

[qwertywarrior]

Unified memory access for CUDA is of no benefit to the PC or gaming in general, so it's not really something for enthusiasts to salivate over.

It's of benefit to Nvidia APUs, i.e. Tegra processors which unfortunately have no place in the desktop or laptop market, so what this really does is improve tablet and smart phone performance.

a better term of Tegra is a Soc

[ACatWithThumbs]

Unified memory access for CUDA is of no benefit to the PC or gaming in general, so it's not really something for enthusiasts to salivate over.

It's of benefit to Nvidia APUs, i.e. Tegra processors which unfortunately have no place in the desktop or laptop market, so what this really does is improve tablet and smart phone performance.

That's not correct unified memory is a must for NextGen games!(Thats the biggest bottleneck in PC gaming right now even bigger then DirectX)

All Next Gen games will be build from ground up using unified memory because thats how the memory acces on Ps4/XboxOne works.