Radeon Power Management Still An Incomplete Mess

[necro-lover]

You forgot one thing. The direct genetic manipulation, while giving desired results much faster compared to selection, is not stable or predictable.
Because, when using selection, the plant tests it itself - by surviving or not. Several plant generation loops and analysis, and done. This is not the case with direct manipulation.
It takes a LOT of resources to simulate the plant development and calculate the impact. It takes a lot of money and time to run test cases to stabilize the whole changes and iron out even tiny quirks.

Only then direct genetic manipulation is safe.
This is not being done. Instead money making is done - safety does not interest people, they are used to dump wastes without thinking of consequences, like babies peeing in own bed.

This is why genetically modified food possesses danger much beyond the scope of a nuclear meltdown.
Tschernobyl case, with tremendous efforts, was managed.
But you can't manage where genetic damage is driven, you can't control where the bee flies.

http://www.youtube.com/watch?v=nIJbA5BJFiQ

sure but BUT natural breeding(selection) with modern techniques means genetic analysis and based on the analysis you do a natural breeding is technically the same than a direct genetic manipulation.
the only point is: natural breeding sounds better but its technically the same.
you can select any natural-mutation based on a radiation damage on the genetic with natural breeding and in the end there is nothing "natural" its only a effect of the radiation and mutation.

Direct genetic manipulation is the same like a mutation because of (natural)radiation its only faster.

Natural mutation or direct Genetic manipulation do not chance this: "the plant tests it itself - by surviving or not."

The plant still tests itself by surviving or not.
this basic rule is not off just because you manipulate the genetic code.

(only you use the BAD stuff termination technology then its broken by design)

[Panix]

A lot of crap.... Seriously, not sure what ppl are rambling on about but I read at least two complaints about power management with the FOSS drivers. The Radeon feature (Xorg) page won't load but I remember that it listed PM as working.

Yet, people here who have laptops with at least the mobility 3200 radeon card in their laptop can't get it to work with proper management. How old are those cards? If it's so difficult to get OSS drivers working, you have a problem.

Not sure why few here aren't just saying the facts which are: AMD just supports Windows. They don't invest in Linux. The resources and personnel just isn't there. Call them out instead of these ridiculous tangents some posters go on.

[Sidicas]

A lot of crap.... Seriously, not sure what ppl are rambling on about but I read at least two complaints about power management with the FOSS drivers. The Radeon feature (Xorg) page won't load but I remember that it listed PM as working.

It is, but the equivalent of "PowerPlay" is not enabled by default because AMD let laptop manufacturers do whatever the hell they wanted with the hardware. My HP laptop doesn't use AMD's reference clocks on the GPU, nor does it put into the BIOS what clocks the GPU should run at.. So enabling the generic power management of the Radeon (r300g) driver does not work well at all. The laptop either burns up quickly at the reference clocks or it isn't stable at the lower voltage. However, with a little bit of tinkering and manually setting clocks, I've got it working *VERY* well, much better than the Catalyst drivers as I have a lot of control over the laptop's performance and thermals whether I'm on battery or plugged into the wall.

HP ships their own Windows XP (only) drivers for my laptop and they haven't been updated since 2005. The drivers slightly underclock the GPU and GPU memory so the laptop doesn't get too hot. There is no way for any drivers to know about these kinds of things under Linux because the information really isn't stored anywhere that I know of. So of course, power management for Radeon is a mess.. AMD needs to get their shit together and force laptop manufs. to actually make laptops that can handle the heat of their chips at reference clocks or at the very least, keep track of when a laptop manuf. ships laptops with underclocked / undervolted GPUs from the factory.

Yet, people here who have laptops with at least the mobility 3200 radeon card in their laptop can't get it to work with proper management. How old are those cards? If it's so difficult to get OSS drivers working, you have a problem.

Not sure why few here aren't just saying the facts which are: AMD just supports Windows. They don't invest in Linux. The resources and personnel just isn't there. Call them out instead of these ridiculous tangents some posters go on.

I'd say the HD 3200 is still fairly new considering how far the open source drivers have been lagging behind. Yes, that's mostly because AMD put only half a dozen guys on their open source driver team, so it is AMD's fault. At least they're still better than nVidia who put a whopping 0 guys on their imaginary open-source team.

[bwat47]

A lot of crap.... Seriously, not sure what ppl are rambling on about but I read at least two complaints about power management with the FOSS drivers. The Radeon feature (Xorg) page won't load but I remember that it listed PM as working.

Yet, people here who have laptops with at least the mobility 3200 radeon card in their laptop can't get it to work with proper management. How old are those cards? If it's so difficult to get OSS drivers working, you have a problem.

Not sure why few here aren't just saying the facts which are: AMD just supports Windows. They don't invest in Linux. The resources and personnel just isn't there. Call them out instead of these ridiculous tangents some posters go on.

Lacking working dynamic power management is not 'working'. Right now all you can do is manually set a profile via the commandline and that hardly works. Don't get me wrong, I certainly don't blame the oss driver devs, they are doing great work with the resources they are given, this is AMD's fault for not realeasing good enough documentation and no code drops.

[Sidicas]

Lacking working dynamic power management is not 'working'. Right now all you can do is manually set a profile via the commandline and that hardly works. Don't get me wrong, I certainly don't blame the oss driver devs, they are doing great work with the resources they are given, this is AMD's fault for not realeasing good enough documentation and no code drops.

On my hardware (Mobility X700), the dynamic power management works.
It's just that you wouldn't want to use it simply because every time it reclocks / revolts the GPU it causes the screen
to flicker and go black for a second. Certainly not something you want to have happening very often. The Catalyst
drivers for my hardware only do it when the AC power is connected / removed and that's all. I prefer using a profile so I can trigger it myself. Wish I could find a nice on/off KDE Plasma widget that can easily be tied to scripts.

[ownagefool]

I hope this answers your question.

It just makes you look nuts mate.

Listen, its fine for you to be annoyed with AMD half arsing parts of their OSS policy, but suggesting Nvidia is the alternative makes you look stupid.

The lay of the land is pretty simple:

> If you care about OSS support above all else, Intel wins.
> If you care about the best performing OSS supported cards, then an AMD probably wins.
> If you care about performance above all else, then Nvidia/AMD blobs win.

You can argue that Nvidia blob support is superior to AMD, but thats no reason to argue you should use an Nvidia card if all you care about is OSS support. Use some common sense and at least back Intel or something.

[bwat47]

On my hardware (Mobility X700), the dynamic power management works.
It's just that you wouldn't want to use it simply because every time it reclocks / revolts the GPU it causes the screen
to flicker and go black for a second. Certainly not something you want to have happening very often. The Catalyst
drivers for my hardware only do it when the AC power is connected / removed and that's all. I prefer using a profile so I can trigger it myself. Wish I could find a nice on/off KDE Plasma widget that can easily be tied to scripts.

I don't really consider that "working" the flickering was so bad it was unusable, and in addition even with dynpm enabled the card still ran significantly warmer than with catalyst. Even forcing it on low profile it still ran hotter than with catalyst. The card I have is an hd2600 mobility, and without catalyst the temps/battery life is really terrible and the fan goes crazy.

[D0pamine]

so here is the scenario i'm thinking of... some dual booting ubuntn00b fires up his/her laptop and wonders why its getting hot and the battery doesn't last very long..... perhaps because the GPU is being used constantly to render the desktop?? IS COMPOSITING REALLY NEEDED??????

the GNU Network Object Model Environment is looking less and less GNU-like and more like a product of microslop

[bwat47]

so here is the scenario i'm thinking of... some dual booting ubuntn00b fires up his/her laptop and wonders why its getting hot and the battery doesn't last very long..... perhaps because the GPU is being used constantly to render the desktop?? IS COMPOSITING REALLY NEEDED??????

the GNU Network Object Model Environment is looking less and less GNU-like and more like a product of microslop

I consider compositing a requirement for a modern desktop. I don't care so much about the effects, but the removal of stuff like tearing, flickering, and visible window redrawing are all benefits of compositing.

[curaga]

Tear/flicker can be prevented by the driver, as the open radeon driver has done for years by default. No compositor needed for those.

Visible redraw - could be solved by double/triple-buffering in the app, possibly in the driver, but this one is the easiest to solve by a compositor.