NVIDIA Still Working On PRIME/DMA-BUF Contribution

[phoronix]

Phoronix: NVIDIA Still Working On PRIME/DMA-BUF Contribution

Aaron Plattner at NVIDIA is still working on the open-source "PRIME Helpers" patches for the Linux kernel. This is work towards ultimately better handling PRIME/DMA-BUF for NVIDIA Optimus Technology on Linux...

http://www.phoronix.com/vr.php?view=MTI3NTM

[LLStarks]

I'm glad Nvidia still cares.

They need goodwill if they ever want to use dma-buf themselves.

[tomato]

Oh, so after all it is possible for nVidia to contribute code related to desktop/laptop GPUs to Linux kernel!

must be snowing in hell or something... oh, no, wait, it's just the result of a hard stance courtesy few kernel developers

(to be clear: I'm glad they're contributing)

[dh04000]

I hope that ATi also uses this work to handle thier "optimus-like' solution instead of using thier in house solution. If all the drivers can use ONE system for sharing GPU and transfering GPU access, and ALL contribute and improve it, then this system could really become something. As much as people here hate Nvidia and Ati at times, if we could get them to help improve and use it, then we all would benefit.

[pingufunkybeat]

Oh, so after all it is possible for nVidia to contribute code related to desktop/laptop GPUs to Linux kernel!

Of course it's possible. This was never in question at all.

What they can't do is to relicence other people's code. Just like Alan Cox can't relicence the blob and make it GPL.

[tomato]

Of course it's possible. This was never in question at all.

What they can't do is to relicence other people's code. Just like Alan Cox can't relicence the blob and make it GPL.

Then why the DMA-Buf code wasn't contributed by nVidia in the first place?! It's not like the Optimus problem wasn't known for few years before! I'm not asking them to open source their blob because I know that they have other companies "precious IP" in there. Thing is, having proprietary code and contributing code to kernel is not mutually exclusive by itself. But they're in love with their proprietary crap and can't see forest for the trees (that if you get 10 competent engineers to solve a problem, 9 of them will solve it in exactly the same way).

I hope that this current work is not a one-off only.

[dh04000]

Hey, I'm not an expert in Kernel development, so can someone translate this for me?

Will these helpers allow Nvidia/Other Non-GPL Drivers to use the PRIME/DMA-BUF system in the kernel, or is this totally unrelated and doesn't chnage the non-GPL driver interacting with GPL kernel/system issue?

[asdx]

I'm glad Nvidia still cares.

They need goodwill if they ever want to use dma-buf themselves.

That's what amazes me about Nvidia (in a bad/funny way). Whether I read a story like this I ask myself how far are they willing to go with their workarounds/hacks to keep their "trade secrets" and lack of coopeartion close to themselves, instead of cooperating with the FOSS/Linux community.

*Sigh*, shame on you nvidia.

[Danniello]

Maybe nVidia cares because... Valve "asked" them to do it, because their problems with Steam for Linux on nVidia Optimus...

Little of-topic: nVidia Optimus it is disaster even on Windows. It should be simple and working out-of-the-box, but reality is that it generates plenty of problems. Most of them are with not recognised games, but recently it is even with Mozilla Firefox - version 18 is crashing all the times on hardware with nVidia Optimus...
http://support.mozilla.org/en-US/questions/946268

[tuke81]

What a shock firefox crashing

I do understand that 2 gpus in one laptop have it's strong points i.e. gaming. But that I don't understand that manufactures stuck them almost every pony market laptops out there(damn, marketing guys great new laptop with two graphics cards and überdüber gtx720m with 2GB ram for more powah...). Hopefully Haswell has strong enough igpu that it does not need other gpu to stick on every laptop outhere.