R500 Reverse Engineering
David Airlie has mentioned to me this week that a new reverse engineering group has begun work on the ATI/AMD R500 series. David is not taking part in this group (due to existing NDAs with ATI), but this project is current led by Jerome Glisse. Jerome was one of the open-source r300 drivers, and he also has another one or two others helping out.
At this time it doesn't look like they have made much progress or even have a Wiki/project page, but upon getting any information I'll pass it along. According to David, they are using the tools from Nouveau for hardware traces. With the mode setting being the main change, David doesn't expect it to be much work.
In regards to David's wannabe-open-source 2D R500 driver, AMD has not re-evaluated their position and the driver will not be made public (from what I know, I do not expect it to come out anytime in the near future). Granted with Jerome's new reverse engineering efforts, Airlie's 2D driver will soon become obsolete.
At this time it doesn't look like they have made much progress or even have a Wiki/project page, but upon getting any information I'll pass it along. According to David, they are using the tools from Nouveau for hardware traces. With the mode setting being the main change, David doesn't expect it to be much work.
In regards to David's wannabe-open-source 2D R500 driver, AMD has not re-evaluated their position and the driver will not be made public (from what I know, I do not expect it to come out anytime in the near future). Granted with Jerome's new reverse engineering efforts, Airlie's 2D driver will soon become obsolete.
Posted on February 21, 2007 at 07:20 PM in Graphics
Tags: ATI, AMD, R500, Reverse Engineering, Nouveau
Archive/Print This Page
Subscribe To This RSS Feed
R500 Reverse Engineering
> Granted with Jerome's new reverse engineering efforts, Airlie's 2D driver will soon become obsolete.
if he's so good at it, i can hope that it will soon obsolete the fglrx ;-)
great news! is it going to be developed separately, or included in current opensource radeon driver?
if he's so good at it, i can hope that it will soon obsolete the fglrx ;-)
great news! is it going to be developed separately, or included in current opensource radeon driver?
Posted by yoshi314 on February 22, 2007 at 05:20 AM
R500 Reverse Engineering
And how can it make fglrx obsolete? It relies on fglrx to get information? It can only ever play catchup.
Posted by timbo on February 22, 2007 at 07:25 AM
R500 Reverse Engineering
I do not know for certainty, but I believe the R500 driver will become part of the open-source Radeon driver once completed.
Posted by Michael on February 22, 2007 at 07:34 AM
R500 Reverse Engineering
> And how can it make fglrx obsolete? It relies on fglrx to get information? It can only ever play catchup.
once it gets more feature-complete it might draw more people to it than ati's driver. ati does not put much effort into implementing new functionality to their drivers. it seems that developers are too busy with bugfixing (or adding support for new cards).
the number of people using open r300 driver is steadily increasing, one reason is the abandonment of r200 cards in ati's linux driver, other is kernel/x.org/aiglx/exa/randr up-to-date support. also lack of ati support for non x86/x86_64/linux systems gives them some advantage.
it could be the same with r500 driver, unless ati does something about the state of their driver - the 8.34.8 gives an impression that it breaks more things than it fixes (looking at the problems on phoronix forums).
i wonder if ati cards are harder to re than nvidia ones.
once it gets more feature-complete it might draw more people to it than ati's driver. ati does not put much effort into implementing new functionality to their drivers. it seems that developers are too busy with bugfixing (or adding support for new cards).
the number of people using open r300 driver is steadily increasing, one reason is the abandonment of r200 cards in ati's linux driver, other is kernel/x.org/aiglx/exa/randr up-to-date support. also lack of ati support for non x86/x86_64/linux systems gives them some advantage.
it could be the same with r500 driver, unless ati does something about the state of their driver - the 8.34.8 gives an impression that it breaks more things than it fixes (looking at the problems on phoronix forums).
i wonder if ati cards are harder to re than nvidia ones.
Posted by yoshi314 on February 22, 2007 at 08:02 AM
R500 Reverse Engineering
"ati does not put much effort into implementing new functionality to their drivers. it seems that developers are too busy with bugfixing (or adding support for new cards)."
That is actually a somewhat inaccurate statement. AMD does deliver new features to the fglrx drivers, just not every month due to their release cycle.
"i wonder if ati cards are harder to re than nvidia ones."
The r500 rev engineering should actually be easier seeing as all previous generations have been reverse engineered, so there is not nearly as much work to be done as Nouveau. One of the things though will be getting manpower working on the reverse engineering, since Airlie cannot participate on the r500 driver.
That is actually a somewhat inaccurate statement. AMD does deliver new features to the fglrx drivers, just not every month due to their release cycle.
"i wonder if ati cards are harder to re than nvidia ones."
The r500 rev engineering should actually be easier seeing as all previous generations have been reverse engineered, so there is not nearly as much work to be done as Nouveau. One of the things though will be getting manpower working on the reverse engineering, since Airlie cannot participate on the r500 driver.
Posted by Michael on February 22, 2007 at 08:28 AM
R500 Reverse Engineering
At this point, I'm not even concerned about 3D that much, I just want to have acceptable performance on my x1300 and be able to use it with my widescreen monitor.
Posted by Brandon Sharitt on March 1, 2007 at 10:47 PM