Drivers for linux are rubbish

[IsawSparks]

No, actually, I didn't say that, it's just that you're an idiot.


I told you that if you find performance so important that you'll ditch open source for a few fps and a few watts, then you should use closed source. Like the nvidia drivers, CoreAVC, Windows, Sony rootkit, XBox, knock yourself out.

You should just be aware that there are people who will not ditch open source for a few fps or a few watts, and the open source drivers are at a state where most of us don't have to.

Right, the idiot is not the person whose posts contradict themselves and tells people on a Linux forum to use Windows.

Sony Rootkit? 2004 is calling you from its DVD collection.

I'll use my hardware and software the way I want thanksverymuch. I know it may not please you that my combination of closed source drivers and open sourced everything works 100% of the time with games and HD video, but it pleases me a whole hell of lot.

In the vernacular, suffer in ya jocks mate.

[IsawSparks]

Sure. Its a little known operating system called ANDROID, which is OPEN SOURCE: http://source.android.com/

The HARDWARE vendor has only produced system images up to 1.6 despite 2.1 being current. What kind of hacker doesn't build the latest version and use that instead?

And as it happened, the jump from 1.6 to 2.1 had a similar result to an xserver upgrade -- it broke all the existing hardware accelerated GPU drivers -- they have since been worked around, but for a time, if you wanted a 7201 on android 2.1, you did without GL and video decode accel.

You never said which version of the OS, you only talked about the phone itself. You also said it was an ARMv5, when it's actually ARMv6/11.

Knowing what you own and how to communicate yourself accurately makes conversation a whole lot more constructive. I read just fine the problem is you don't communicate effectively.

It's nice that you can follow a guide online and build a new version of the Android OS, but that doesn't make you a hacker. It may make you a hack though.

Also, isn't your 7201 a multi core chipset? That might explain the performance capability. That said it's definitely not capable of proper full rate 720P MPEG2 sans GPU acceleration and the screens res would never let the video reach full rate anyway. MPEG2 is res dependant, unlike AVC and VC-1 ASP.

[pingufunkybeat]

I may be missing something here, but how does being closed source and proprietary improve the suitability of the drivers for "high end discrete GPU markets"? Assuming that you had two equally/equivalently functional drivers, one open and one closed, what benefit could being closed bring?

Simply reusing the Windows Catalyst driver, with 10+ years of engineering and know-how in it?

Unfortunately, that one can not be open sourced, for a variety of reasons.

[pingufunkybeat]

I read just fine

Apparently not

[Panix]

You guys can argue all day long but the fact of the matter not every Linux user shares the same demands or philosophies. It's supposed to be about options and choice, isn't it?

Even you open source extremists seem to admit that the driver can't do everything so the binary driver is requires for some features/uses. But, if it is a secondary priority and inferior to the nvidia driver, then there isn't a good choice, is it?

The fact it's based on commercial needs and the workstation market means the open source priority is limited and the binary choice isn't any better than the Nvidia option. What is needed, imho, is continual improvement of both drivers and some more emphasis on the binary since the FOSS drivers will never get the full support meaning 3D and licensed features. Just because you don't need a feature doesn't mean they should follow your requirements! The video card is capable of hardware acceleration among other features so it should be supported for that. Whether it's an open source method or closed, it benefits all uers and owners of the card if there is SOME option for using the card for that feature.

[bridgman]

Supporting every possible combination of hardware and distro is a lot of work that works far better when the drivers are open. Wasn't this one of the reasons why the support for older chipsets was dropped?

Support for older GPUs was dropped from the Linux drivers because it was also dropped from the Windows drivers, and without the ability to share development & testing costs across multiple OSes it wasn't really practical to maintain support any more -- the decision didn't really have anything to do with the availability of open source drivers.

We did look at the possibility of freezing the fglrx driver at 9.3 and then just adding support for new X server and kernel versions, but all indications were that relatively more users would be happy if we put effort into improving support for the older GPUs on the open source drivers instead.

I may be missing something here, but how does being closed source and proprietary improve the suitability of the drivers for "high end discrete GPU markets"? Assuming that you had two equally/equivalently functional drivers, one open and one closed, what benefit could being closed bring?

Simple. The closed driver can share code with other proprietary drivers and leverage work done across OSes (almost 100% of the PC market), but that option isn't really practical for an open driver.

[Qaridarium]

Simple. The closed driver can share code with other proprietary drivers and leverage work done across OSes (almost 100% of the PC market), but that option isn't really practical for an open driver.

catalyst 9.3 is obsolete right now because the lastest git version of the r300g can handle full openGL2.1 and ETQW for exampel run's...

Rip catalyst 9-3...

[droidhacker]

You never said which version of the OS, you only talked about the phone itself. You also said it was an ARMv5, when it's actually ARMv6/11.

Except.... that ANDROID uses the v5 instruction set. So though the CPU may qualify in spec as a v6, the differences aren't being leveraged. It is therefore to be considered as a v5.

Knowing what you own and how to communicate yourself accurately makes conversation a whole lot more constructive. I read just fine the problem is you don't communicate effectively.

The specifics are completely IRRELEVANT. The only one concerned with the absolute everything-must-be-so-BS'ing-ly detailed is YOU. The POINT is simply that a VERY WEAK CPU is FULLY able to handle the work in question.

Now you go on a rant about OS and instruction sets. WHO CARES? It is neither here nor there!

It's nice that you can follow a guide online and build a new version of the Android OS, but that doesn't make you a hacker. It may make you a hack though.

You'd better stop yourself now before you make a complete ass of yourself. I haven't made any assumptions about you and I have NOT issued ANY PERSONAL INSULTS. So unless you are interested in a PERSONAL CONFLICT that could easily see you banned for personal hostility, I strongly suggest that you BACK OFF.

Also, isn't your 7201 a multi core chipset?

Not quite. It does have two cores, but one of them (256 MHz ARM9) is strictly for the radio and cannot be leveraged by the OS.

That might explain the performance capability.

Huh? Compared to WHAT? We're talking about netbooks as an absolute MINIMUM. I don't think that there's any question that there's not a phone around that is even as powerful as the weakest current netbook!

That said it's definitely not capable of proper full rate 720P MPEG2 sans GPU acceleration and the screens res would never let the video reach full rate anyway. MPEG2 is res dependant, unlike AVC and VC-1 ASP.

720? Might. It doesn't take that much more than 480. Of course the simple fact that it can handle an SD video in software pretty much means that ANY x86 manufactured in the last 10 years should have no trouble with HD MPEG2.

And now that you mention it... It doesn't even have an mpeg2 accelerator..... MPEG4/H263/H264 only.

[mirv]

catalyst 9.3 is obsolete right now because the lastest git version of the r300g can handle full openGL2.1 and ETQW for exampel run's...

Rip catalyst 9-3...

ETQW runs on the open source drivers? Sweet, did not know that. They really are further along than I thought.

[pingufunkybeat]

Support for older GPUs was dropped from the Linux drivers because it was also dropped from the Windows drivers, and without the ability to share development & testing costs across multiple OSes it wasn't really practical to maintain support any more -- the decision didn't really have anything to do with the availability of open source drivers.

That makes sense.

I didn't know that the support for older drivers was also dropped. Still, they can be installed on a recent Windows machine, because things don't change as often in Windows-land.