Intel, NVIDIA Hybrid GPU Switching On Linux Nears

[curaga]

@airlied

Thanks. That's the kind of info I was looking for in the news post

[airlied]

OK, I'll spell it out :P

One of the laptops wired in the way required by the Nvidia(tm) Optimus(tm) proprietary technology(tm) mark, and advertised as having such. Or the older ones, without said (tm) mark.

I know it is, after googling the model name. The point was I shouldn't have needed to, that is essential info in a news post like this.

If only it was that simple.

There are Rev1, Rev2 and optimus switchable laptops.

We can ignore Rev 1 - they required a reboot to switch

This code is mostly towards the Rev 2 switchable, which switches the physical outputs between the chips under ACPI control.

Optimus is a different kettle of fish, and is also muddied by the fact that not all Optimus laptops have a Rev 2 compatability mode, most of them do, but I think there are some out there which don't have the physical mux anymore. So the Lenovo T410s has Optimus but it also has the ability to act like Rev 2, hence why this code was possible.

Optimus isn't "wired" its a pure sw solution built on a extra copy engine on the discrete GPU which talks for PCIE like anything normal would. No extra wires.

[airlied]

You have done more than enough, am we're very grateful.

But there needs to be some incentive for them to port their stuff over, without having to make you do that work for them.

Even if we had the same marketshare as macs, they would probably still be this stubborn, so there has to be a roadblock somewhere.

What is really stopping them?

They have no customer demand. (remember you are not a customer of nvidia, you are a customer of an OEM).

[madjr]

How do you suggest we work with nvidia? I work in the open, they know I'm working on it, we don't have source to their code, they have all ours.

I'm really wondering how I can do any more to make your wish come true.

Dave.

You have done more than enough, am we're very grateful.

But there needs to be some incentive for them to port their stuff over, without having to make you do that work for them.

Even if we had the same marketshare as macs, they would probably still be this stubborn, so there has to be a roadblock somewhere.

What is really stopping them?

[curaga]

OK, I'll spell it out :P

One of the laptops wired in the way required by the Nvidia(tm) Optimus(tm) proprietary technology(tm) mark, and advertised as having such. Or the older ones, without said (tm) mark.

I know it is, after googling the model name. The point was I shouldn't have needed to, that is essential info in a news post like this.

[BlackStar]

Neither the article or the blog post mentions whether this is the "optimus" thingy or one of the earlier two-gpus-on-a-laptop thingies.

The "optimus" thingy is a windows-only, closed-source technology. Two guesses as to which thingy this article is about.

[Tgui]

Good work!!!

[curaga]

Neither the article or the blog post mentions whether this is the "optimus" thingy or one of the earlier two-gpus-on-a-laptop thingies.

[wpoely86]

It's cool that the open source drivers already have some support for this, while the proprietary drivers have nothing.

Nice work!

[airlied]

finally, some progress
hope to see this in the next version of my fav distro.

I wish they also worked with nvidia on this, since this switching is a lot more important for those using the binary blobs. I mean thats the ENTIRE reason of getting a laptop like this.

How do you suggest we work with nvidia? I work in the open, they know I'm working on it, we don't have source to their code, they have all ours.

I'm really wondering how I can do any more to make your wish come true.

Dave.