The State Of Open-Source Radeon Driver Features

[intellivision]

Hi, I am going to break the rule for just one post to you.

I too encountered exactly same situation of "stubborn management", "corporate viewpoint" and "stalled things/minds/mindsets".
The morale I got however, is extremely different from yours.
I just let the founders of iD Software explain it:
http://rome.ro/games_ddici.php

GL

crazycheese, I've been following your posts and every one of them has been an incoherent mess.
Perhaps the Phoronix forums aren't for you, I'd say 4chan would be more suited.

[BO$$]

I think bridgman is the most informed one to tell us the answer to this question if he may: What would be technically easier for you and the linux driver team? Stabilising fglrx or getting the open source driver on par on functionality and performance with the fglrx?

Everybody seems to be assuming that if fglrx would be open sourced all our problems will end. I don't know how big is the fglrx but if somebody drops 6 millions lines of code on you (I don't know how big fglrx is just assuming a number), you aren't exactly going to become productive with it right away. In fact most devs that already work on the open source driver would continue to do so since they already know the code and are already productive with it.

Somebody else said that if you want to use the old 2000 3000 or 4000 lines of products you should stick to Ubuntu 12.04 not 12.10. I knew that. But most probably somebody coming to linux today will download the 12.10 since it's the latest and will find that the graphics drivers perform inadequately.

Another one said that why does linux have to want better market share. I've heard Torvalds saying that he doesn't like the fact that the only place where linux is not successful is the desktop. And I've heard many others saying that they would like better market share. In fact with better market share a lot of linux problems like being ignored by driver developers would end.

[necro-lover]

I think bridgman is the most informed one to tell us the answer to this question if he may: What would be technically easier for you and the linux driver team? Stabilising fglrx or getting the open source driver on par on functionality and performance with the fglrx?

Everybody seems to be assuming that if fglrx would be open sourced all our problems will end. I don't know how big is the fglrx but if somebody drops 6 millions lines of code on you (I don't know how big fglrx is just assuming a number), you aren't exactly going to become productive with it right away. In fact most devs that already work on the open source driver would continue to do so since they already know the code and are already productive with it.

Somebody else said that if you want to use the old 2000 3000 or 4000 lines of products you should stick to Ubuntu 12.04 not 12.10. I knew that. But most probably somebody coming to linux today will download the 12.10 since it's the latest and will find that the graphics drivers perform inadequately.

Another one said that why does linux have to want better market share. I've heard Torvalds saying that he doesn't like the fact that the only place where linux is not successful is the desktop. And I've heard many others saying that they would like better market share. In fact with better market share a lot of linux problems like being ignored by driver developers would end.

I use the Radeon driver with the latest alpha kernel and latest mesa GIT ppa. And I'm happier with my computer and its more stable in the overall experience than my catalyst experience and I really tried to use the catalyst for years!

So your "Logic" is broken please consult your Borg collective for a reassimilation to fix your brain with nano-robots.

Or get some Vulcan lessons in Logic !

These FOSS individual Anarchists will always fight back in resistance

Because they don't think that resistance is futile.

[Hamish Wilson]

I think bridgman is the most informed one to tell us the answer to this question if he may: What would be technically easier for you and the linux driver team? Stabilising fglrx or getting the open source driver on par on functionality and performance with the fglrx?

I am not bridgman of course, but since this is a public forum I can respond anyway. Again, I have to ask why you see these as being mutually exclusive goals. One does not take resources from the other, and both are actually happening (if somewhat slowly). Catalyst is much better than it was back in the day (and certainly much better than it was before ATI was purchased by AMD) and the free drivers have been improving by leaps and bounds (I know, I have been following them for almost all that time).

By the way, there is not "one linux team" as I understand it, but two separate initiatives within the same company. So they are not even tripping on each other's toes.

Everybody seems to be assuming that if fglrx would be open sourced all our problems will end. I don't know how big is the fglrx but if somebody drops 6 millions lines of code on you (I don't know how big fglrx is just assuming a number), you aren't exactly going to become productive with it right away. In fact most devs that already work on the open source driver would continue to do so since they already know the code and are already productive with it.

Indeed, releasing the Catalyst code would not solve much of anything in the short term. I do not see anyone even suggesting that - and if they are they are indeed misinformed. Getting access to certain functions of Catalyst probably would be useful though.

Somebody else said that if you want to use the old 2000 3000 or 4000 lines of products you should stick to Ubuntu 12.04 not 12.10. I knew that. But most probably somebody coming to linux today will download the 12.10 since it's the latest and will find that the graphics drivers perform inadequately.

That was me, and in that instance I think you can blame Canonical. For a distribution that targets new and in-experienced users they certainly do not do themselves much favours by promoting their six month or die release model as being stable. Fedora at least never had that pretension, and I think Canonical should be promoting the LTS releases as being the best for new users. But that is not my or bridgman's call, and my point still stands that if you want a stable interface for blobs you should choose a stable long-term support release.

Another one said that why does linux have to want better market share. I've heard Torvalds saying that he doesn't like the fact that the only place where linux is not successful is the desktop. And I've heard many others saying that they would like better market share. In fact with better market share a lot of linux problems like being ignored by driver developers would end.

Again that was me, and that was me asserting my own personal opinion. My personal goal is to have a free and open system that suits my needs and does what I need it to do. Marketshare for me is not that large concern, although I do agree getting more users and companies on board could be helpful. Torvalds may have other ideas. Besides, citing Torvalds opinions as support for your own can be a rather risky proposition, given his rather abrasive manner and penchant for strong forceful declarations.

Also, Torvalds has not been all that friendly to blob manufactures in the past that are unwilling to work with his kernel infrastructural:
http://www.bit-tech.net/news/bits/20...slams-nvidia/1