RadeonHD 1.2.2 & 1.2.3 Drivers Released

[he_the_great]

I've got a theory.

What if AMD is just playing us? So that we only get full support long after the hardware is made obsolete?

I'm going to take a different approach to this than, bridgman.

The posting was on RadeonHD drives. My understanding from phoronix is that it is headed by RedHat folks and the like, not AMD. On top of that, they have already been given NDA spec documentation before any public release.

On the fglrx side of things, it supports the new hardware, just as well it does the old. It does not have CrossfireX support for the 4xxx series, nor any of the predecessors.

[yoshi314]

They are fooling us, because in 2 years I don't know what to do with an HD 2600 XT gpu. so in the mean time they convince us they are "good" and are working so we buy their cards.

i don't think so - the drivers are catching up more quickly to newly released cards.

just because the 3d docs aren't available yet doesn't mean that it will be the same for every next generation of cards.

once first 3d documentation will be out it will be much easier for ati to release subsequent 3d docs for next generations of their cards. right now the people preparing the docs have to pass them for IP review, because they don't know for 100% what can they release and what they cannot.

btw i don't the the

in 2 years I don't know what to do with an HD 2600 XT gpu

part. will your card die or something?
i still have an x300se card around and it's quite useful for me. what year did it come out, again?

[cruiseoveride]

hahaha

the good, the bad, the fglrx

nice one!

[Yfrwlf]

@cruise: Like was stated, AMD is helping out, to some degree, with the RadeonHD driver for instance, but regardless it's easy to make the claim that you're making but it's possible you're right. But, it also could just be that it's low priority for them because Linux still has low market share for several reasons (some of which are unfair, some should be downright illegal IMO, like forcing Windows to come as a package with new computers). Either way, yes, Linux does get the shaft, whether it's intentionally not getting very much help or not. AMD is giving a lot of same day support of their hardware for Linux in their closed source side though, but so has Nvidia to some degree. AMD's open source side has improved much more than Nvidia's open source side though, so it's not all just BS PR.

[lucky_]

@bridgman
I'am still curious about two things :
It's said that work is going on in private repos under NDA for the 56/7xx, but the IP review is still an issue.
- Did you receive any feedback from the IP gurus after your last attempt ?
- How can you be sure that the code that hits you internal repos won't conflict with the IP review (suppose you use subsets a,b,c in your repos, but the IPists allow you to you use subsets a and c only). Will this mean that after the IP review you will need a cleaning code review ?

[Opteron]

Hmm,

Added support for RV770, RS780, M82, M86, and M88.

So no RV730 in this release, yet.

cheers

Opteron

[bridgman]

@bridgman
I'am still curious about two things :
It's said that work is going on in private repos under NDA for the 56/7xx, but the IP review is still an issue.
- Did you receive any feedback from the IP gurus after your last attempt ?
- How can you be sure that the code that hits you internal repos won't conflict with the IP review (suppose you use subsets a,b,c in your repos, but the IPists allow you to you use subsets a and c only). Will this mean that after the IP review you will need a cleaning code review ?

The nature of the IP reviews means we get tons of feedback. It's not a question of "sending it off to the IP review" -- a bunch of senior technical folks get together and talk about the info we plan to release and the risks associated with releasing that info. Each of the issues raised gets investigated, trying to (a) figure out how big and real the risk is, (b) look for ways to eliminate or reduce the risk while still being able to do useful open soruce development work, and (c) in parallel, determine how much of the risk exists based only on information which already exists. For both of the reviews so far we ran into roadblocks before getting into the details of the 3D engine -- this time I'm hoping to get everything through.

Any time we do development work under NDA we assume that a code review and revision may be required before the resulting code can be released. As long as the amount of code involved is relatively small (as it is in this case) that's not too much of a problem, particularly since we usually can say "here's all the info but only use these bits for now" during development. Before we start NDA work we evaluate the risk of the "IP we don't release" permanently tainting the developers and making it difficult for them to do their ongoing work, and if the risk seems at all high then we don't even release the info under NDA. One of those little memory-wipe wands from "Men In Black" would be real handy but for now all we have is NDAs

[Anato]

ADM's actions makes me angry. They talk about open source support and releases some documentation now and then, but clearly priorities are elesewhere.

There is clear disparity in priorities with closed source drivers and open source documantation. If AMD realy wanted to help us (and itself) it would put more effort to releasing documentation faster. I don't necessarily mean full 3d, but at least 2d acceleration and video so that drivers would be usefull for desktop and hdtv.

I'm currently considering to by Radeon HD 4850 to my gaming machine which is also in linux TV use. And 780G board to work machine which is also MythTV server. Currently I have to decide which one (AMD/ATI or Nvidia) sucks less in open source front

[Louise]

The nature of the IP reviews means we get tons of feedback. It's not a question of "sending it off to the IP review" -- a bunch of senior technical folks get together and talk about the info we plan to release and the risks associated with releasing that info. Each of the issues raised gets investigated, trying to (a) figure out how big and real the risk is, (b) look for ways to eliminate or reduce the risk while still being able to do useful open soruce development work, and (c) in parallel, determine how much of the risk exists based only on information which already exists. For both of the reviews so far we ran into roadblocks before getting into the details of the 3D engine -- this time I'm hoping to get everything through.

Any time we do development work under NDA we assume that a code review and revision may be required before the resulting code can be released. As long as the amount of code involved is relatively small (as it is in this case) that's not too much of a problem, particularly since we usually can say "here's all the info but only use these bits for now" during development. Before we start NDA work we evaluate the risk of the "IP we don't release" permanently tainting the developers and making it difficult for them to do their ongoing work, and if the risk seems at all high then we don't even release the info under NDA. One of those little memory-wipe wands from "Men In Black" would be real handy but for now all we have is NDAs

Each time you explain what goes on behind the scene, I am amazed how much work and money there is being spend!

So thanks a lot for that =)

[Louise]

ADM's actions makes me angry. They talk about open source support and releases some documentation now and then, but clearly priorities are elesewhere.

There is clear disparity in priorities with closed source drivers and open source documantation. If AMD realy wanted to help us (and itself) it would put more effort to releasing documentation faster. I don't necessarily mean full 3d, but at least 2d acceleration and video so that drivers would be usefull for desktop and hdtv.

I think AMD have been VERY good to Linux, taken in consideration the marked share Linux have for gaming. If that marked share can be measured at all.

Based on the many articles here at Phornoix, and just the comments in this thread the thing that stalls the release of specs is IP that someone else holds.

Imagine if it is Intel that holds the IP for something in the R600 specs?

Edit: Or worse: What is it is Microsoft have have IP's in the R600 specs?