Intel Wants YOUR Linux Questions, Feedback

[squirrl]

These patent issues are just lame. I'd rather have a choice. Their free software driver is a few miles behind their closed source one for Windows.
It's sad but even the damn Apple products seem to best Xorg/GNU/Linux. I honestly don't mind running proprietary.

The only people that care otherwise is Stallman and a few handful of the developers working on the free software driver.

If 1% of the population of computer users care then what about the 99% that don't care?

[alelinuxbsd]

Question for the Intel's Linux graphics team i hear rumors that said that from Amd BD2 (bulldozer 2) should not be necessary to move data in one part of memory reserved for video because it will be processed on site simply using pointer.
Given that these shifts affect a lot the performance of the Gpu included on the Cpu, is possible known if that is possible even using the new Intel Ivy bridge or is necessary wait for new processors?

[crazycheese]

The only people that care otherwise is Stallman and a few handful of the developers working on the free software driver.

If 1% of the population of computer users care then what about the 99% that don't care?

Wrong. The only person who does not care about proprietary on this planet is you.

99% are not "care"/"not care", they are clueless. This is third state. True/False/Clueless.

Because of clueless:
- Gary Kildall was killed by Microsoft.
- IBM moved and renamed to Microsoft.
- DMCA got in.
- HDMI came in.
- Genetically manipulated food is allowed in America.
- Software patents came in.
- SOPA is incoming.

Linux and foss is something that appeared because someone was not clueless, but cared.
Apple has some piece of "care", but it is not that of FLOSS by caring about the essence, but caring about "buzz". Apple is zombie.

Anyway, you are free to live in garbage society, eat garbage food and use garbage OS.

Remember Matrix? Power of machines was literally based upon cluelessness.

In fact they WANT you to be clueless.

Intel linux driver is awesome.

[AlbertP]

Even for people who don't care about licensing, open source drivers are better. They can be preinstalled in Linux distros. You can use various generations of Intel graphics right out-of-the-box. For AMD/nVidia you need to install a driver. Intel is much easier for newbies.

[jltyper]

Even for people who don't care about licensing, open source drivers are better. They can be preinstalled in Linux distros. You can use various generations of Intel graphics right out-of-the-box. For AMD/nVidia you need to install a driver. Intel is much easier for newbies.

I think Linux distro's are trying to appeal to the reading portion the population. When there are proprietary drivers, Ubuntu offers the user a chance to download them automatically.

And one more thing, lets not forget easier <> better

[allquixotic]

I think Linux distro's are trying to appeal to the reading portion the population. When there are proprietary drivers, Ubuntu offers the user a chance to download them automatically.

And one more thing, lets not forget easier <> better

And proprietary <> better, either! The nVidia binary is a huge memory hog; if you were to get a complicated 3d program with a fair amount of shaders running and take a dump of the address space, then use a program to create a pie chart of the memory used by each shared library loaded into the address space, the nVidia binary's userspace components would be taking about 50% of the mapped memory in the address space. On 32-bit systems this leads to out of memory errors (well, more like "out of address space" errors, since you might still have available system memory), on both Win32 and Linux because the binary driver's implementation is near identical on the two platforms. In fact, this problem has plagued 32-bit Windows game developers for several years, and the general idea is that game developers are now forced to use less and less RAM in their actual game in order to make room for nVidia's userspace component. And they are forced to run memory profiles on the memory usage resultant from the rendering they're asking for, and tone it down if it crashes the nVidia blob.

Can we do anything about this, where "we" is defined as anybody who is not an employee of nVidia? The answer is no. It is completely, utterly out of our hands.

This is not the case for the open source drivers. Motivated people with technical knowledge, regardless of their corporate affiliation, can (and do) fix issues within the drivers when it bugs them. And people ensure that the drivers continue to work properly as the rest of the stack evolves around them (e.g. the kernel, LLVM, GCC, Xorg, etc.) To the extent that this occurs with the proprietary drivers (and it does occur, but usually takes between 6 months to a year for the most annoying problems e.g. gnome-shell corruption with fglrx), the general public is at the mercy of the corporation to:

(1) become aware of the problem,
(2) acknowledge the problem,
(3) deem that the problem is worthy of fixing,
(4) prioritize and fix the problem,
(5) release a driver with the fix.

I'm tired of being at their mercy, and so, the choice is obvious: use the open source drivers.

Most people who say they "don't care" about open source/proprietary while using Linux really mean, "I'm too short-sighted to be patient and let the open source drivers mature, so I'm going to use whichever driver is higher performance in the short term, even if it means I'll be stuck with bugs that are completely out of my control, and be tied to a specific version of the Linux kernel until the company decides to support newer ones".

I can't live that way. I want to be empowered to take things into my own hands and do what needs to be done. I think it would be just as fitting to call the free software movement the "empowered users movement", because that's really what the result is if free software is successful.

[ryszardzonk]

No, according to Gwenole, this won't happen anytime soon, unfortunately.

I hope that changes sometime in the future. In the mean time if one is Gentoo user they might be interested in http://forums.gentoo.org/viewtopic-p-6936196.html where I posted link to my ebuild. What it basicly does it uses most recent available version of an mplayer install package for Gentoo and adds few things among whitch there is an patch for vaapi from Gwenole tree. If one is not an Gentoo user they may go and grab the patch from here http://bigvo.dyndns.org/overlay/medi...rc4_p20120109/

On the side note If anyone knows where to grab patch for ordered MKV in mplayer I'll add it in there too

@Eugeni
Maybe there is a way for getting vaapi updated it the Mesa? I am kinda looing forward for it not say anymore "vaapi state tracker currently unmaintained"

[Ansla]

You're mixing things up, the vaapi state-tracker is for the rest of the free world, the drivers using Gallium3D, not Intel.

[ryszardzonk]

@ansla

is it really? Plase excuse my indolence

PS. maybe then it is argument for Intel moving toward Gallium then?

[jltyper]

The nVidia binary is a huge memory hog;
On 32-bit systems this leads to out of memory errors (well, more like "out of address space" errors, since you might still have available system memory)

Firefox is huge memory hog. That doesn't mean anyone is going stop using it. This thread is about Intel anyways.

I'm tired of being at their mercy, and so, the choice is obvious: use the open source drivers.

You will always be at somebody's mercy. Live it. Accept it.

Most people who say they "don't care" about open source/proprietary while using Linux really mean, "I'm too short-sighted to be patient and let the open source drivers mature, so I'm going to use whichever driver is higher performance in the short term, even if it means I'll be stuck with bugs that are completely out of my control, and be tied to a specific version of the Linux kernel until the company decides to support newer ones".
I can't live that way. I want to be empowered to take things into my own hands and do what needs to be done.

Being tied to a specific version of the kernel is only a disadvantage if the newer kernel is exceedingly better than the previous one. If you can't prove it there's an advantage, with real numbers..
How is that any less short-sighted? Are you just looking at version numbers? What about the articles on this website?