Does anyone know when OpenSource ATI GPUs power options are fixed?

[Adarion]

I don't know about the prices in your country. But since there are so many Germans here, let's take that country for example ...

No! Germany is not a good example. Electronics here are way higher priced than anywhere else in the world. It really sucks here sometimes. And we currently have a stupid gouvernment.
Besides: It's hard to compare these laptops. Sometimes Windows comes with preinstalled Adware (the yellow plague (Symantec) as trial version) to make it cheaper, then you will find hardware differences aplenty (HDD/SSD, RAM, BIOS chip, opt. drive, all the chips, screen matte or glossy) and even then you still have different manufactureres, normally Dell and Lenovo tend to be higher priced than ECS, Asus and all these guys. Partially they are better crafted, that is something you can't see when just looking at the normal tech specs. Also a lot of chips types are not given, which makes me sometimes really angry. As Linux user you want to know exactly what is used for Wlan, Lan, softmoden, Audio, SuperIO, BIOS chip (flashrom anyone?) etc. so you can check for compatibility. Also here, these chips can make a difference in price. Oh and then you also have the case, the housiung, some use Al/Mg some use just plastics. And the fans. Some try rise margins by adding cheapo fans that make a lot of noise. And you never read about that unless you check user reviews. So I think it isn't easy to compare Notebooks that way.

Sadly AMD wasn't treated well in past times by notebook vendors. If you finally found something than the vendors tried to keep it as cheap as possible and combined the AMD stuff sometimes with horrible chipsets or even worse - glossy screens (OMG-the-horror). Luckily situation seems to improve and there is more choice. On a notebook a lot of things have to play with each other it's not just the CPU and GPU.

The idea of better power management as GSoC project sounds nice but afaik enrolling phase is already over. But then I don't know if company bound projects are accepted by G or by the students.

[Panix]

This I take just an example for the complaints and crying going on here.

Okay, okay.
Listen, people. (Or read.)

1.There were numerous polls on phoronix where you could cast a vote "what is most important to you in GPU driver development".
A lot of people wanted 3d stuff first.
(I always voted for power management by the way.)

2. Now let's imagine AMD-ATI had started with power management first. 1000s of people would be crying about the lack of 3d accel and whatsnot.

3. Power management is among the most complicated parts. So it takes some time until it runs nicely everywhere on all the many chips. Look for the explanations agd5f gave somewhere here in the forums.

4.You all are welcome to take a manpage, a book and some articles from the net "how to write Linux Kernel drivers or X.org or mesa drivers" and hack away.
What? You can't? No time?
You're also free to donate to pay more developers. There are not many enterprises actually hiring devs for freedom driver development like AMD does.

5.There are things called "software patents". "intellectual property" and other things that might range from silly to more serious. The problem is, the US and some other countries are very strict about this stuff.
So AMD has to go through a long minefield when releasing specs and writing code.

6. Okay. Go on complaining and threatenting to buy non-AMD stuff.

I wish you a happy time with your intel graphics.
BUT: Don't come crying here!!
a) if it doesn't provide you enough GPU power when you need it
b) if you're bankrupt after buying their stuff
c) if you are having so much fun with you Imagination Tech, PowerVR Poulsbo/... GPU that you're bursting into fucking tears because NOTHING works.
In case you didn't know: Intel buys a lot of stuff. Also from these folks. And sells it as "intel" graphics.
Later people wonder why nothing would work. happened. So yeah, go trust a company that sues harmless people for using a simple English word.

7. Planning on anything else, yeah?
Okay, go ahead.
I personally have seen a lot of GPUs and I am so sure I do not want them.

And that is not just because I'm a stupid fanboy.

Yes, you are. You're an idiot.

I should have stopped reading after #1 but let's answer your stupid points.

#1. So, pathetic AMD won't support their products in Linux because of a stupid poll?!? REALLY?!?
Next.
#2. So, they're incapable of doing both?!? In other words, they're incompetent? That doesn't bode well for getting AMD hardware then.
Next.
#3. So, the guy who still can't get power management for hardware that's over 5 years old should just concede that AMD can't support something that's over 5 years old because it's too complicated? What does that mean for customers that have bought brand new AMD mobile gpu laptops and cards?!? Wow, should I even bother with any of your other STUPID points?!?
Next.
#4. MAYBE AMD SHOULD SUPPORT THEIR HARDWARE IN LINUX?!? MAYBE SOMEONE MIGHT TRY WRITING SOMETHING FOR THEIR HARDWARE WHEN THEIR OWN PEOPLE DO?!? MAYBE THEY MIGHT STOP BEING SOOOOOOOOO PRO-WINDOWS AND ACTUALLY SUPPORT LINUX FOR A CHANGE?!?
#5. More BS excuses.
Next.
#6. Yeah, it's a good idea to NOT buy AMD hardware, at least, not graphics hardware, if someone wants to use Linux. Since, AMD obviously doesn't have a policy of supporting Linux, you have two other sh*tty options or alternatives but at least there will be some support with the other companies. Nvidia, at least, doesn't profess to have open source support like the lying AMD does. Intel at least has open source support even if their graphics hardware isn't that good.

But, I think Intel is investing in other computer hardware. Oh yeah, AMD is, too, but their cooling and efficiency sucks, too, with their processors and boards from what I read. But, they go out of their way to assert Linux support when it's a bunch of crap.

How often does one read here of problems and lack of support for each release of drivers and features that never arrive?

'Fanboys' of hardware usually aren't that bright. Why be a 'fanboy' of any company?!? It makes little sense.

[bridgman]

Hi Panix;

Just in case you have forgotten (it seems that you might have done so), there are two drivers not one :

- an AMD-developed driver which *does* include all the things you say we don't provide and which *is* developed at the same time as the hardware (or at least the same time as the Windows driver)

- an open source driver developed by the community with support from AMD, which isn't fully caught up with new hardware support and features but which has added ~10 years of new hardware and features in ~5 years so catching up the rest of the way seems pretty likely (with the caveat that DRM issues are always hanging over our collective heads)

[DaemonFC]

Hi Panix;

Just in case you have forgotten (it seems that you might have done so), there are two drivers not one :

- an AMD-developed driver which *does* include all the things you say we don't provide and which *is* developed at the same time as the hardware (or at least the same time as the Windows driver)

- an open source driver developed by the community with support from AMD, which isn't fully caught up with new hardware support and features but which has added ~10 years of new hardware and features in ~5 years so catching up the rest of the way seems pretty likely (with the caveat that DRM issues are always hanging over our collective heads)

The first one would be great if its sole purpose wasn't to hide what it was doing from the user, and therefore becoming malicious software.

[bridgman]

Oh good, since that is by no means its sole purpose. The "proprietary" aspect is just a consequence of sharing common code written for use across all OSes.

[darkbasic]

there are two drivers not one :

There is only one bridgman: fglrx kernel/xorg policy support is just a joke.

[DaemonFC]

Oh good, since that is by no means its sole purpose. The "proprietary" aspect is just a consequence of sharing common code written for use across all OSes.

Hiding things from the user including whatever Microsoft and MAFIAA boss Chris Dodd don't want you to tell them, I mean.

[bridgman]

If you mean "not exposing internal details from other OSes in our Linux driver" I guess that is correct. Is that what you mean ?

The obvious downside of code-shared drivers is that you do need to keep at least some of the driver source proprietary -- traded off against the ability to provide more features and performance than you could deliver with the same resources and a Linux-specific code base.

In the end each HW vendor makes a choice based on their primary markets; NVidia offers proprietary-only, Intel goes with open source-only, and we support both.

[jrch2k8]

bridgman haters are gonna hate and i suspect many of these haters don't use r600g from a long time ago or are nvidia fanboys, as an user of r600g i am extremely grateful for this superb driver(it obliterate that piece of junk called fglrx in almost any comparable area).

here some experiences:

GPU: AMD Radeon 4850X2 sapphire (my most power hungry monster GPU)
r600g: profile / mid GPU0 -- profile / low GPU1
Mesa: 8.1 git + custom flags / pcie2/ tiling 1d-2d/ swapbuffer off
Kernel 3.4-rc2
software:

KDE 4.8.2 composite / GL2 shader / EGL: r600g obliterate fglrx so badly is sad, i mean everything just pop in the screen without any sort of render error
Firefox 13 aurora GPU accel: r600g flawless in everything that can be accelerated + html5 youtube flawless too, fglrx problematic/crashy but sometimes render faster some webgl test but not that much to be worth
Chromium 18 + git GPU accel including 2d canvas: same as above except fglrx crash the browser when 2d canvas and some other flags are up
Wine 1.4 + Lineage2 Harmony: r600g is uber stable never had a crash with these game so far and render is flawless / i assume fglrx should give more FPS but my eyes can't tell
mplayer2 + ffmpeg git: 1080p piece of cake and flawless (phenom II x4 CPU tho) with fglrx xvba never worked properly and xv well tearing feast, the new option helps but cripple performance in other areas so not worthy
Wayland + QT5: r600g orgasmic i can't wait until it replace X11
temperature: r600g 70c after hours of gaming fglrx 67c
power at wall delta: around 10w which for such power hungry GPU is really good if you ask me

so i don't deny that for weak mobile devices can make a difference but i can't believe is so beasty as some claim it is, first check you have properly set the power modes (mid is good enough for most cases) in rc.local, get a very recent kernel so you get the ASPM fixes, check your default cpu governor is in ondemand or powersave, check you have an updated BIOS cuz many mobile devices come with some crappy buggy ACPI implementations and if you are lucky they published a fix later after release(i mean very lucky), check your screen have backlite support in linux, etc.

and well thanks for the community and AMD support for the r600g is really shaping as a awesome driver

btw i have several laptops in the office that uses the radeon 4650 mobile with r600g and after set properly battery life is maybe 5%-10% less than with fglrx which again i consider pretty good for such early state of the PM in gallium but i never seen a laptop porperly configured that drops more than that so my question is you can post what the F is this laptop hardware and OS? cuz well is easy to whine (assuming this device exist and you have actually that problem and you are not a troll) and talk crap but without any hint you are not asking for support you are asking for witchcraft or budu

btw in 3d performance i can always reach near 65% of fglrx using r600g which is more than enough for my use but well fglrx wins here for now

[darkbasic]

jrch2k8, how many cards did you test? 4870/50 is probably the best card ever for R600g... I bought lots of AMD cards just to test them with the open drivers, power management didn't always work.
Even with the same card, a vga from another vendor or with another bios may not work (for example pm on my HD5870 was as good as with catalyst except the flickering but other users had issues with dynpm).
We need a working, tear free dynpm enabled by default. This is first priority in a laptop (but last in a desktop IMHO).