AMD Promises Open Source Graphics Drivers

MoxFulder writes "Henri Richard, AMD's VP of sales, has promised to deliver open-source drivers for ATI graphics cards (recently acquired by AMD) at the recent Red Hat Summit. A series of good news for proponents of open-source device drivers. In the last year, Intel, the leading provider of integrated graphics cards, has opened their drivers as well. But ATI and NVidia, the only two players in the market for high-performance discrete graphics cards, have so far released only closed-source drivers for their cards. This has created numerous compatibility, stability, and ethical problems for users of Linux and other open source OSes, and prompted projects like Nouveau to try and reverse-engineer NVidia drivers. Hopefully AMD's decision will put pressure on NVidia to release open-source drivers as well!"

I could not read the summary

I'm sorry, I could not read the summary. I have worked in R&D... I got as far as "VP of sales has promised" and had a panic attack.

Nice

They're just trying to get them some press. Unfortunately Linux gamers are an edge case. People needing video card support on Linux above vanilla SVGA as a whole is an edge case.

Re:Nice

No, some other users just want fully operational 2D graphics with dual head support. More especially for dual DVI cards where the external TMDS is not supported under X.

Dual head

Now there's something to do if you had a million dollars.

Re:Nice

Well, if there is a good video card support on linux, linux gaming will just strengthen. It isn't a godgiven that you can play games only on Vista.

Re:Nice

The problem is, most Linux desktop users use it to develop or manage it as a server. They won't pay for a game. They will not pay for anything at all, most of the time.

I, a Linux user, am more then willing to pay for the same commercial games that are available for Windows. Matter fact I have payed already when I purchased UT, UT2k4, Quake 4, and Doom 3 which I have installed exclusively to play on Linux. I have no idea where you got the idea that most Linux users are unwilling to pay for software let alone games. Did you poll every single Linux user? Or did you form your ignorant opinion out your ass? I'm sure there is a large number of Windows users that don't pay for their software hence the existence of warez groups offering commercial software for free. And what about the large number of people unwilling to pay for Windows or games which drives many software companies to develop draconian drm & copyright protection measures?

Re:Nice

Why should software development companies waste money developing games for Windows when they could get a far larger market share by making games for Wii, Xbox 360, PS3, DS, PSP, etc.? The PC gaming market is much smaller than the console games market, and Nintendo is helping widen the gap with the Wii and DS which appeal to non-gamers as well.

Re:Nice

What about that optical mouse you're using?

[looks down...]

Logitech.

[looks over at the 'ol Windows98 box...]

Logitech.

[goes downstairs, looks at wife's laptop...]

Logitech.

'Nuff said.

Re:Nice

I'm not sure why you seem to think that undermines the previous posters point.

As they said, the console market is far larger than the PC market, and that console titles outstrip sales of PC titles significantly (and that they sell for more). World of Warcraft and Linage don't change that. Selling a million plus copies of a game in the first year is expected of a half decent console game (games EA's Madden do that and more in their first couple of weeks - games on the PC don't sell that well with anything like that frequency (nor do they tend to retail for as much).

Added to that, is of course the risk factor.

You might imagine that releasing an MMO is a better way to make money than a console game, but that's not borne out. MMO are far risker projects, requiring many times the capital investment, much longer development cyles (years longer), a more complicated business structure and business plan and have massive monthly outgoings (rather than just paying a small amount for a tiny helpdesk team and letting the staffers go once the game is out). When the risk is higher, you of course keep less of the return (and again, there is a lower return when you have high on-going costs).

As EA has discovered, it's a lot more economical to stick to releasing incrimental upgrades of existing tiles (from Madden and Fifa to the Battlefield series) than to take risks with PC MMO's, which ultimately fail far more often, and bring in less profit as a percentage of both investment and revenue even when they are successful. EA have even shut down 'successful' PC MMO's down, because they were not successful enough by their standards - there just wasn't enough of a return on their investment, that is, they were better off spending that money in developing a new console title, because it was almost certain to give a better RoI.

The overall numbers of gamers for Lineage 2 and WoW is large, but it's probably not as profitable as many think - a third of WoW subscribers are in China, and they pay just a fraction of the amount US and EU players pay. Several million more people in the US, EU and AUS have played the likes of Halo 2 on the X-Box than have played WoW on a PC or Mac.

All the above is is why there are so few MMO's, compared to console titles, as a business MMO's are simply less profitable (because they bomb more often every n attempts, and when the do go wrong they do so more spectacularly, in that a bigger hole is left in the publisher and/or developers wallet).

Re:Nice

* You != an average Linux user.
        * Loki Software proved the lack of market for Linux games 5 years ago when they shut down in 2002.


I think we also have to take into account the fact that the Linux landscape has changed drastically in the last five years. How popular was the iPod in 2002?

Re:Nice

You've missed 1 key point: 5 years ago!

Linux is a MUCH better desktop OS than it was 5 years ago. Coincidentally, that was about the same time I tried to use Debian as a desktop. It stunk and I quickly dropped it. Then 2 years ago, I found reason to try it out again. Slackware was pretty good, but still iffy for a desktop.

Now I've got Kubuntu. It's amazing, and definitely a good desktop OS. The home PC I have ordered was chosen based on the idea that it would only run Linux, and Windows didn't matter. (This one is going to be my 'game' PC in the living room now.)

Loki was too early. If they tried the same thing now, they'd have a LOT better success.

As for the 'waiting' issue... Was that the only issue? Or did Loki fail to advertise that they were going to be releasing that game in a few months? Because if I didn't know about it, I'd just pick it up for Windows, assuming Linux would never get it. Maybe there were other issues as well, that don't come to mind immediately.

Loki didn't prove anything except that they didn't make it.

Re:Nice

Loki Software proved the lack of market for Linux games 5 years ago when they shut down in 2002.

Actually, you are incorrect. Loki Software died mainly because of managerial incompetence and mismanagement. You can read about some of that here. [linuxtoday.com]

And, as others have pointed out, the Linux desktop has matured a lot in the last 5 years. Even if Loki died because of a lack of customers (which is not the case), the same would not necessarily happen today.

Re:Nice

It's pretty simple, really: * You != an average Linux user. * Loki Software proved the lack of market for Linux games 5 years ago when they shut down in 2002.

Chalk me up to another non-average Linux user. I've got a 3 foot wide bookshelf with boxes of commercial Linux games I've bought (most of Loki's offerings while they were still operational, stuff from LGP, NWN, iD's offerings, etc). I wonder how many of us it takes to make a market.

Loki had a lot more problems going on than the lack of a Linux market at the time. They tried to be too big, too quick and go for too many AAA titles at once. Between having to pay large upfront license costs to port games (often six figures or more) and royalties from every sale on top of that, they just didn't have a business plan that met their market. They would have been much better off as a porting house rather than a self publisher (much like Ryan Gordon/icculus does now).

On the other side of the scale, LGP is working on a lot of B grade games. Some of them are very good but they're very, very slow and methodical in their porting. I've beta tested games for them which took more than a year to release after I got the first beta. They need to get stuff out the door if they want to be serious. Throw in Tux Games charging $50 [tuxgames.com] for the exact same box you can buy in the discount bin for $15 (ok, here's $18.82 at walmart [walmart.com] and you might skew the numbers because people aren't buying from them so they "don't get counted" as a Linux sale. In fact, you can pick up NWN, Quake 4 and Doom 3 from Walmart for the price of one game from Tux Games with shipping.

IMO, a lot of the problem is simply the game industry not understanding the linux market properly. A market exists but you can't go at it Loki style or you're doomed to failure, not because of the market, but because the business plan doesn't add up. Software houses should look toward portability when they design a game and the cost of a single developer to handle the Linux port of it would be pretty cheap in the overall development of the game (I haven't exactly done a poll but I wouldn't be surprised if you could find Linux geek willing to work for less than the average game coder just for the privilege of being able to get paid to program a game for linux). Another part of the problem are the publishers who dictate to the game houses what they're going to release so even if they want to do a linux version, it may not be possible.

Re:Nice

They're just trying to get them some press. Unfortunately Linux gamers are an edge case. People needing video card support on Linux above vanilla SVGA as a whole is an edge case.

Having solid drivers isn't just "an edge case". Go install the default ATI or Nvidia driver on a recent linux distro then upgrade it to a non open source one from the company. It's like day and night. I noticed a huge difference between having a default driver vs company made one, silly things like dragging a console with transparent background is no longer a pain, it's smooth. The desktop feels fast and I don't even have any 3d desktop installed.

Then you got things like multiple monitor support. My Feisty install without closed source drivers just wouldn't work. It kept resetting the screen resolution after reboots, wouldn't recognize my second monitor, I couldn't even force it, it was a black screen. Once I installed the closed source driver, shazam! All my video worries are gone. Now I am happily using a 2560 x 1024 dual monitor setup with hardware acceleration.

Also you got 3d desktops like Beryl. With eye candy being a major selling point in some operating systems, 3d features will become important if desktop linux wants to get more popular. I hope all graphic card companies will develop good drivers for Linux.

Re:Nice

Having solid drivers isn't just "an edge case". Go install the default ATI or Nvidia driver on a recent linux distro then upgrade it to a non open source one from the company. It's like day and night. I noticed a huge difference between having a default driver vs company made one, silly things like dragging a console with transparent background is no longer a pain, it's smooth. The desktop feels fast and I don't even have any 3d desktop installed.

Agreed... this is why I was excited about possibly having open-source drivers, and posted this article. My current box has onboard NVidia, and a low-end ATI discrete PCIe card... frankly, I can't wait for *one* of them to have open drivers. Although using the binary drivers improves 3D performance and a lot of strange display bugs, as you point out, it's a huge pain to keep them up-to-date with kernel upgrades since they can't be bundled with the main kernel. I don't like putting a big binary blob in my kernel, which by all reports is out-of-date with respect to a lot of other kernel subsystems, and may open up security holes.

I don't do 3D anything (word processing, programming, web browsing mainly), but baseline unaccelerated SVGA is definitely *not* acceptable: 2D graphics acceleration is necessary for a smooth and productive desktop experience. The open-source 2D acceleration is actually pretty good at this point, but of course it simply DOES NOT WORK with a lot of the latest ATI cards in particular.

The current pace of open-source driver development is positively glacial, largely because most of the people who have sufficient documentation to easily improve the drivers are under NDA. Read this incredibly frustrating blog entry [livejournal.com] from a developer who's under NDA with ATI... using only a few hundred lines of code, he has patched the open-source Radeon driver to support most of the newer ATI cards... but ATI has spun its wheels for months without allowing him to release the code.

Mainstream gaming

I agree that hard core Linux gamers are an edge case. However, most of us would like to be able to play Penguin Racer or Tux Kart occasionally. Useless eye candy like 3D window switching effects help relieve boredom as well. This doesn't require the latest hot graphics card with dedicated cooling towers. However, it would be nice to have stable drivers that track kernel evolution for entry level 3D cards - sufficient for simple games and effects. The present situation is that old low end Vanta Nvidia cards (suitable for Tux Kart) still require proprietary drivers - and Nvidia is losing the motivation to keep them updated (they did patch old drivers for the security hole mentioned on Slashdot a while back).

IMO, using binary blobs that run in the card, not in the kernel (i.e. downloadable firmware), are a reasonable way for vendors to hide trade secrets while keeping the card updateable and the kernel driver open source. As long as shared memory between the graphics card and main system is restricted to a window, bugs in the firmware shouldn't cause security holes in the kernel. In fact, one benefit of micro-kernel architecture is that isolated drivers that run in their own process and address space, can run in an intelligent I/O card instead.

The IBM Series/1 was built on the principle. All I/O was done by intelligent cards with a common API: submit Device Control Block with command, memory block, and parameters to start an operation. Receive vectored interrupt and find results in updated DCB and memory block. Interrupt included address of DCB, so interrupts were trivially "object oriented".

Re:Mainstream gaming

It sure is nice when GLX works and you can do CAD, modeling, simulations and 3D programming(OpenGL) on a Linux box. So there are practical uses beyond gaming for those fancy 3D cards.

Re:Nice

Not any more. Compiz and Beryl are becoming the standard way of drawing onscreen in much the same way as aero and quartz. That means unless you have decent 3D you will stuff up desktop performance. Gamers might have much higher demands, but the days of 2D chips being adequate for desktop use are over.

Re:Nice

Nice to see you dont know squat about linux.

3d acceleration and the Video acceleration is used daily by EVERY linux user (short of text based server installs.

What you just said is as redicilous as saying "Vista users dont need anything but 2d Svga."

I run Wxvga all the time WITH 3d and guess what I dont play games in linux at work.

And I am not a "edge case" but a typical linux user.

Re:Nice

The high-end is still reserved for gamers, researchers, and people doing visualisation. A modern (cheap) GPU, however, does a lot more than a framebuffer. The most obvious thing it does is compositing. Pretty much every application does some form of alpha blending (see those icons on your toolbar?), even if it's with a 1-bit alpha channel, and there's no reason this couldn't be done in hardware. At the windowing system level it's even more important. Draw every window to a texture and let the GPU handle the shadows (not just a gimmick; on OS X is't a huge visual clue as to the active window) and overlaying.

Pixel and vertex shaders are a whole new ball game. There's a lot of text on my screen. All of it drawn from truetype fonts. A truetype font is basically a series of bezier curves. Microsoft Research released a paper a few years back where each of these curves was approximated to a triangle [microsoft.com]. A vertex shader program then inspects each of the rendered triangles and corrects the error between the triangle and the bezier. This allows an entire font to be uploaded to the GPU and rendered at any resolution with very little CPU load or RAM usage (compare this with Apple's hack of just storing a table of glyphs in the video RAM, which doesn't scale very well).

Pixel shaders can be used for a lot of things. With pixel shaders you can perform a lot of convolutions in hardware, giving some nice effects. You can use a pyramid algorithm to perform a number of things, like bi-cubic filtering, blurring, etc in a fraction of a second.

Sure, you could do a lot of these on the CPU, but the GPU is going to do them a lot faster, and probably use less power (important for mobile users).

Even without needing the 3D support, it's useful to have all of the features working correctly. Power management is a big one, since the kernel needs to be able to save the state of the GPU somewhere before turning it off, and Linux uses a lot of hacks to try to avoid needing to do this.

eye candy and scientific workstations

With Compiz, Beryl, and XGL, excellent 3D graphics support has become a mainstream issue. Furthermore, Linux is widely used in science and engineering, and those users use excellent 3D graphics as well.

I'll believe it when I see it

$SUBJECT. If AMD really means it, it bodes well for the future - I always hoped that their openness with the Linux community over the x86-64 porting effort wasn't a one-off.

The big question though is whether or not they will try for mainline inclusion, or if they will go with an out-of-tree effort.

Ethics? Still, nice to hear.

This has created numerous [...] ethical problems for users of Linux and other open source OSes,...

Damnit, Jim, I'm a computer user, not a philosopher! But honestly, I think most of the people COMING to Linux in the Desktop world could care less about these "ethical" issues. Once again, it's just another thing that some of the Linux community puts above having things Just Work(tm). However, since some of these closed-source drivers aren't working for some, it's nice that AMD wants to open theirs so that eventually they can be modified until they work. A win for everybody, actually.

Ethics? Yes.

I think most of the people COMING to Linux in the Desktop world could care less about these "ethical" issues.

True, right now they don't care. But that doesn't make it any less important to develop Free drivers.

Richard Stallman had his realization that Free software is necessary based on his experience with a printer driver [wikipedia.org].

Re:Ethics? Still, nice to hear.

I think most of the people COMING to Linux in the Desktop world could care less about these "ethical" issues. Once again, it's just another thing that some of the Linux community puts above having things Just Work(tm).

Under Linux, most things Just Work(tm) because people with those ethical issues took the time to do something about it. You can't possibly claim that GNU or Linux exist in an amoral vacuum.

Re:Ethics? Still, nice to hear.

Free Software has never been about ethics, it's always been about making stuff work. One of the pre-requisites to making stuff work, however, was the availability of the source code and the rights to fix or adapt it, and redistribute the changes (so not everyone needs to do the same fix themselves). BSD started with a bunch of guys trying to fix UNIX. GNU started with RMS trying to fix a printer driver. Linux started with a guy trying to fix Minix.

The last one is the most interesting, since fixing Minix ended up meaning completely re-writing it because (at the time) the license didn't allow redistribution of modified versions (only patch sets, and those were growing unwieldy).

To an outsider, it might seem that ethics or ideology were the motivating factors, but in reality it's just a desire for things to work. The problem with binary-only drivers is that they might kind-of work now, but at some point they might not and then there will be nothing we can do about it.

Re:Ethics? Still, nice to hear.

No. Open Source is about making stuff work. Free Software is about ethics and freedom.

can they?

The agp specification is proprietary and you need to pay (heavily) for the spec. Releasing their driver source would be like giving away the agp spec. It might not be legal.

Seeing is believing

Don't buy ATI until they have followed through with that promise. As far as I am concerned, they have until July, when their new low end card becomes available. If there are no Linux drivers for that card then, I will buy an NVidia based card.

Re:Seeing is believing

Good thinking!

Punish that damn ATI for not having an open source driver. Punish them by buying hardware from another company that doesn't have open source drivers!!!

This is great

I only buy Nvidia because it just runs better under Linux even though ATI is better on Windows. I happen to run both and I want the best of both worlds. My guess is this is partly because of the change of momentum towards Linux on the corporate desktop over the last year.

Some people will be sure to downplay this, but I think this is really the beginning. It will take time, but I expect that Linux desktop graphics will closely compete with the Windows desktop soon.

Nvidia, this is your wakeup call. Follow suit, or my next graphics card will ATI.

Hopefully it will help corporate adoption of Linux

Some of the most popular corporate laptops (Dell D600 I am thinking of you) have perpetually had nasty graphics drivers that have resulted in much suffering on the part of users.

For the D600, I have a rather nice choice of either good performance and much graphical corruption (weird cursors and such) with the official ATI drivers, or horribly slow performance and no corruption with the existing open source drivers.

Oddly enough this only happens with some Distros, so I am sure there is a magic setting I could change some where, but honestly, I shouldn't have to! Especially just to get basic desktop functionality.

Don't Promise it. Do it.

Really it isn't hard. Identify the code you own, replace the code you don't, put on a GPL header and release.

Promises are cheap.

Actually, I hope they don't.

Hopefully AMD's decision will put pressure on NVidia to release open-source drivers as well!

Actually, being an ATI fan, I hope NVidia doesn't release open source drivers. ATI will get a lot of goodwill with this, at the very least, and at most, if Dell's experience with Linux is good, perhaps they'll get more sales. Go ATI. :)

an alternative reading...

VP of sales: We have to delay the r600 again!!
CEO: WTF?!?!? Ok, ok. Let me think a moment... with all those problems, it seems that we are not developing _anything_... I got it!! You just have to go to one con, one full of hippys with long hair, and make a new press release stating something wonderful of our new cards... No! much better! something wonderful of all of our cards. That is. This will give us some good press and nobody will remember the delay
VP of sales: what kind of press release can afford such incredible thing??
CEO: I don't know... well, these hippys are always requesting new and shiny drivers that works with their toy operating system. Just promise that.
VP of sales: But... we promised that 18 months ago, and still...
CEO: And?
VP of sales: Well... but our customers are not stupid.
CEO: We'll discuss that later. Just give me a month without news of r600 and I'll remember you in the next stock options party.

They're probably just press whoring, but...

I'll bite, though. I've been a Matrox and Nvidia user ever since I gave a hoot about the card driving the pixels in my machines and stopped buying Trident and S3 cards (so.... maybe 10 years now). For some reason, I've never given ATI a second look, aside from the fact that they seem to be the chipset of all rack-mount servers I've ever used.

I'm in limbo now, though. I'd love to upgrade my Geforce FX 5200/128M card, but I just won't until I can get dual-head and acclerated 3D under FreeBSD/amd64 with an *open source* driver (it matters to me -- I'm sick of binary blobs). Period. The first company who gives me hardware and open sources the drivers (or at least the damned specs) will get my business the minute a driver is available for my platform, and I'll become their biggest volunteer fanboi/astroturfer to reciprocate the deed (on- and offline).

I'll believe it when I see it, though. I ain't holding my breath.

Day one Sale

I havent bought an ATI card ever (unless you count the GC and Wii). I would imediate buy one if they had robest OS drivers. Currently I always buy Nvidia.

It is that simple really, not about gaming nessesarily, more about trust, I trust OS more.

Intel driver Open Source?

Last time I looked at the Intel driver source, there were a ton of calls into the video BIOS. Not something I would call an "Open Source" driver. This may have changed since then,- I really hope so.

Why is it important to have more source you might ask. Well, for one thing it would be really nice if we can get rid of the video BIOS altogether. A full source driver which shows how to switch video modes is a very good start to accomplish this (although not necessarily enough).

And then you might ask, why do we need to get rid of the video BIOS? Well, when evaluating graphics chips for an embedded systems, I found out that the video BIOS can spend an insanely long time initializing stuff and displaying stuff that we don't want/need (some like several seconds). In general, video BIOSs are over-engineered and do waaaay more than needed.

If you are aiming to build a near-instant-on system, and/or something that doesn't look like a PC, you want this sort of flexibility. If AMD steps up to the plate, that would be awesome.

Re:Intel driver Open Source?

Last time I looked at the Intel driver source, there were a ton of calls into the video BIOS. Not something I would call an "Open Source" driver. This may have changed since then,- I really hope so.

Why is it important to have more source you might ask. Well, for one thing it would be really nice if we can get rid of the video BIOS altogether. A full source driver which shows how to switch video modes is a very good start to accomplish this (although not necessarily enough).

Look into the new "modesetting" branch of the Intel driver, currently moving towards the default. It moves all the work of modesetting and other related hardware manipulation from the video BIOS into the driver, and avoids the video BIOS entirely. This does indeed give the benefits you describe in your post. Some of this modesetting code also moves toward sharing between drivers, to support modesetting for all Xorg video drivers. (Some of it consists of driver-independent code, such as dealing with funky monitors.)

Vague...

Anybody got any more details? They talk about the lack of a timeline, but "graphics drivers" is also vague, and could mean 2D, or just another small subset of features.

I'm certainly not going to go out and start buying ATI cards until all the details are worked-out.

Time to sign the pledge!

Looks like it's time to sign the pledge to support open, 3d graphics drivers http://www.pledgebank.com/open3d [pledgebank.com] and put some pressure on graphics card manufacturers.

What about licensed functionality?

If there's licensed proprietary code in the drivers that ATI/nVidia doesn't own, it'd take a great deal of lawyers and possibly price hikes to make it possible. Just because something's closed source doesn't mean to say the only reason it's not open source is laziness or hatred of F/OSS...

Nouveau-equivalent for ATi?

Does anyone know if there is a Nouveau-equivalent for ATi? The FGLRX package feels so wrong!

Excellent

This is turning out to be a good year for good news! It was only when I started thinking of things this is better than, that I realised we've also had the apparent collapse of DRM for music (it's not over yet, admittedly, but if you were around between 1987-1990 you may remember that from Glasnost to the collapse of the USSR was also a slow motion thing...), the apparent flop of Vista, the imminent failure of the whole Palladium/TCM foundation of Vista's treacherous computing, Dell shipping Ubuntu pre-installed... whatever next? Darl McBride's corpse dragged through the streets by an angry mob of lawyers? Bush / Cheney resign? :)

Damn you AMD!

It used to be so easy, I didn't even have to consider ATI's video cards. Now I have to start to keep up with two lines of confusing model names.

I think I'm going to cry

In 1997 I had a 486 with a 2.5 GB hard drive, which why that might seem behind the curve, my previous machine was C64, so it was a big upgrade for me. My choices seemed to be limited to Windows 3.1, Beos, OS2, or this little project called linux (I guess freebsd too, but I actually use my computer :p ). So I got one of my fancy friends to use his isdn line to download me a Mandrake install cd.

A decade and like 30 distros later I'm still a loyal user. In the past couple years I have watched the great evils of the world morph into our new allies (ie.. IBM, Novell, Intel). But this all came with a underlying Faust kind of deal. Each kernel was tainted with philosophical contradiction. That video driver was nothing more than a cheap whore while the wife was out of town. A big ugly binary among pristine virgin innocent source included apps. All you who think linux can't play games might want to note opengl, sdl, and openal are finding there ways onto your nazi boxes more and more. And bling, ha, one word beryl, Mac aint got shit on us. We're the beast in this mother fucker. Oh and by the way with all that bling and a tweaked kernel, the newest gnome and plenty of other gentoo goodness I'm at about 196MBs of ram used. We'll see where games can be played.

By the way in case anyone from nvidia is reading this, please go fuck yourself you "nvrm: xid" bug ignoring sons of bitches!

Slashdot greatest hits

ATI Committed To Fixing Its OSS Problems [slashdot.org] was posted only a few days ago (that one came from Chris Blizzard's blog) and the cautious tone is backed up by other Red Hat summit reports [livejournal.com]. However, since we're here why don't we pick out the highlights (along with overlooked gems) from last time?

• Someone contacts ATI's sales about this announcement [slashdot.org] and is told to hold tight for more news (this is consistent with what was posted to German news site heise.de [heise.de] link found via liquidat [wordpress.com])
  
• Thank you for your current drivers ATI [slashdot.org]
  
• Slashdot shouldn't speculate on what ATI's roadmap [slashdot.org]
  
• We want open source ATI drivers, not Linux drivers [slashdot.org]
  
• ATI don't support vblank in their binary drivers [slashdot.org] (open source drivers do support vblank but it's hard to find the cards)..
  
• ATI? You mean AMD! [slashdot.org]
  
• Marketing... Ha! [slashdot.org]
  
• This is because of Dell starting a Linux push [slashdot.org]
  
• Doing a Windows graphics driver wrapper wouldn't be feasible... [slashdot.org]
  
• Reverse engineering modern graphics cards is hard... [slashdot.org] because the API driving them is complex [slashdot.org].
  
• Intel have released open source graphics drivers BEFORE their next card was released [slashdot.org]!
  

Elsewhere on the web folks are wondering whether this means that the a new GPGPU will be accessible but the actual graphics driver itself will remain closed. AMD/ATI has also announced open source drivers before which translated into more stable and more frequently released Linux binary x86 drivers...

H.264/VC-1 acceleration

What some might have missed would be the opportunity to sell gfx cards due to all of a sudden being able to do H.264/VC-1 offload/acceleration in linux for HTPC usage.

The first manufacturer to offer this would get a lot of sales due to people using their cards for HTPC. I know I would, because currently I can't play 1080p VC-1 and barely h.264 properly, even on a Core2Duo.

Too little, too late.

ATI has opened the driver because they realise it's now next to worthless anyway: The open source people have created something that just works better in every way (except maybe speed), and they did it just stabbing blindly at registers. Their in-house programmers have reams of documentation to work with, so they have no excuse whatsoever for being behind. They should be fired, after the suits that perpetuated the whole "we can't open the code" farce for a decade.

nVidia doesn't get as much flak because they at least pretend to care. That, and they don't leave early adopters with bricked hardware (not even 2D drivers) for over half a year.

NVidia can still do this for little risk

NVidia can still come off okay too, if they start now. They don't have to risk their IP either. If they contribute money and/or access to older video cards, then they have fostered the development of Nouveau. If they give the team the resources to succeed without giving them access to the source code to the proprietary drivers, then they have fostered the reverse engineering without committing their own staff or risking lawsuit from revealing potentially infringing/stolen intellectual property.

At the very least, they should review the Nouveau contributors and declare them to be free of NVidia influence (a condition of fending off an IP lawsuit). This may not do anything to foster the development directly, but it certainly gives the open source advocates all kinds of warm fuzzies that the company is genuinely interested in making quality open source drivers.

So, I guess next time I recommend PCs using ATI!

I always try to be fair and make exceptional recommendations and deals for the folks at work.

A couple years ago I turned a CFO projected enterprise $8M deal into a $5.5M deal while getting hard-drives sizes doubled, RAM doubled, all CRTs swapped to same size LCDs ... upgrades, an additional 100 desktops (900 total), 35 HP Intel Servers, 20 SUN-Cisco nodes, and three Alcatel-Lucent Omnicore switches. Yep also the cable, patch-panels, wire-racks, transceivers, and all the other required hardware and software trinkets (let me think, was that Gates-Arrow, Ingram Micro, or CDW we made the deal with? Dang, I forget...). Total screw-ups were kept at less than 0.5% of cost which the vendor we went with resolved at no cost. Saved $2.5M ... had a ~$27K problem resolved for $0.

Maybe next time I will look at the MB-graphic or cPCI cards and decide ATI is easier to support over the lifecycle requirement. It will have to prove a better business decision, but I will look, and if all is about equal well ATI will win my recommendation.

I still have another year before I need to seriously start thinking again about big problems, but I won't forget to look at the ATI and NVidia lifecycle supportability issue. The recent distributed content management and storage network was set too a non-proprietary architecture for lifecycle compliance requirements ... scaleability and upgradeability as CFO/CIO infrastructure like to call it.

I am seldom questioned ... when I make a recommendation %~$, most of the technophobes in management remain silent and just hope I screwup.

Yeah, yeah. Let's see....

1. Does ATi (and hence AMD) have the right to opensource all the code in their drivers? If they outsourced or otherwise bought code in - maybe not.
2. What level of functionality will these drivers provide? I would point out that Matrox open sourced their drivers - but then squirreled away a number of the features in a binary HAL.
3. Open source can mean a lot of things. If you're Microsoft, for instance, "open source" can mean "you can look but you can't touch".

It's great AMD are saying this - and it's a lot more than ATi would have done - but as has already been said elsewhere in the thread, I'll believe it when I see it.

I don't think everyone understanding the argument

I don't think everyone understands the argument here. There is a problem with closed source drivers. It's not just ideology. Closed source drivers means you can end up with no drivers for a device for your version of the kernel. Even if drivers for some different kernel version exists. A good example of this is old devices. If the manufacturer still exists, they probably don't care to do the work to update drivers for a device they no longer sell. Maybe there should be a device/kernel interface that stays the same for all time, but I think as a rule, people want the best interface possible, with open source drivers so devices can be kept up. You then of course get the advantage of open source so you can fix/work-round bugs (or improve it!).

Bur for which cards?

I couldn't find anything that said which cards they were targeting here. Is it anything =Radeon 7500? =9200? All Radeons? How about FireGL's? Granted, I'll be happy if they release code for anything, but nothing's actually going to help me out unless it's code for newer cards since the computer I'm using has PCIe and not AGP.

Is this real?

Strangely this statement is not listed on their press-releases page? //Svein

technically speaking

Technically, even VESA drivers are also open source drivers for ATI cards. And that's provided that the announcement even means that they are going to be releasing open source drivers at all.

So, the question is: what exatly are they going to be releasing?

In general, I just wouldn't pay much attention to these announcements either way: it's open source once it's actually been released, no sooner.

Xenon

Will this mean we can expect the opening of drivers for the Xbox360's GPU as well?

Excellent news!

This is excellent news! I'll actually be in the market someday soon for a new graphics card to replace my current 64MB NVidia chipset. I'm tired of the flaky NVidia drivers that freeze up my system at random intervals. Bring it on AMD!

SD

If tey do this

Will it be possible to defeat any DRM they have built into their cards? I thought they employed Macrovision or something, but can't remember the details.

Bravo!

Bravo! I was extremely happy to read this. The advantages, for both consumer and manufacturer, will be plentiful.

I know this will never be read, but just want to go on the record.

Dell

Assuming this happens...

I wonder if Dell is a big driver behind this. Think about it: Dell's going to make Ubuntu an option in some of their computers. It will be cheaper for Dell to sell those models with only hardware that Linux supports well (lower support costs, fewer configurations to test), and it will also be cheaper to have the same hardware be on both Windows and Linux installs.

Result: DAMIT cleans up their Linux drivers and they sell lots more hardware to Dell.

Result: Win for us Linux weenies.

MLC!

Yeah!

One sale made...

... if AMD/ATI actually carries through. I plan to put together a high end box this summer, and was planning to get an Nvidia card as their closed source Linux drivers suck a little less than ATI's closed source drivers. But if AMD actually has open source drivers for their high end cards by then I'll buy AMD. I don't care if Nvidia's comparable card gets 15% higher FPS.

Too little - way too late.

I've noticed that Lenovo dropped ATI chips in favor for Nvidia... could it be a coincidence that they also support Linux on their laptops? No... they fielded many calls from pissed off linux users who rely on Lenovo to leverage their purchasing power to get some better video drivers.

Instead of waiting until ATI gets around to it - they just switched suppliers for some of their laptops to give their customers a choice.

Im staying with Nvidia

AMD customer: "ati drivers for linux suck"
AMD: "I know we will open source our crappy incomplete ati drivers and let the community create the rest of the drivers! It will also promote the use of ATI cards in Linux and we wont have to spend a cent!"

im still sticking with nvidia their drivers may not be open but atlest they sponser Linux distrubutions and provide drivers that actually work!

GPU's merging with x86

Open sourcing the drivers actually make perfect sense, when you consider that AMD and Intel are working on merging the CPU and GPU. When GPU's can be accessed through the x86 ISA, there will be no need for drivers, so handing over the source to soon-to-be legacy video card technology is the first step on the road. Intel also sees this, which is why they're doing the same thing. It will be interesting to see what Nvidia does to keep themselves relevant, will we get a third CPU manufacturer or will they stick with the sinking ship of discrete GPU's.

The final step in the evolution of Fusion is where the CPU and GPU are truly integrated, and the GPU is accessed by user mode instructions just like the CPU. You can expect to talk to the GPU via extensions to the x86 ISA, and the GPU will have its own register file (much like FP and integer units each have their own register files). Elements of the architecture will be shared, especially things like the cache hierarchy, which will prove useful when running applications that require both CPU and GPU power. [anandtech.com]

Re:I have a question...

video performance is kind of like trying to drive in a city where maps (actual correct as of this moment ones) are top secret burn before reading
some stuff you can just guess (main library police office fire department city hall factories over that away homes over this away) but some things you need to get a chunk of map (like 2nd 4th 9th streets all have half streets next to them and 8th st is north only) then you get into wierd stuff like both an airport and a seaport with situations where you can get a ticket if you drive the wrong car to one of the ports (or even that you are driving a CAR AT ALL).

plus some of the OEMS do tricks like have a flash upgrade change bits of the map or change chips on newer versions of the same board.

in short it becomes a game where not only do you not have the names of the players but you are not sure exactly what game is being played

Re:ATI cards are terrible

Don't support Windows but I have installed ATI 'drivers' on a Windows box for a relative. Their 'drivers' include a media player and rely on the .NET framework, their Linux drivers display similar incompetence and even copyright-infringement (needless bundling of Linux AGP code).

So true. Their Windows drivers are horrendous, and their web site used to be so confusing to navigate. Their drivers for Linux are okay, at least their is not crap included like a media player. Unfortunately I'm stuck with dealing with them right now (my main PC is a notebook). Every time I've used Nvidia after using ATI for the same thing, Nvidia was always more stable and the driver support is so much simpler. Now ATI is just playing catch-up with Nvidia, as it always has been. They are hoping to get ahead with this announcement.

Re:People keep on saying "Linux" but...

...wouldn't this be useful as far as Windows goes as well? This could create better drivers for everyone, not just Linux users.

People don't expect open-source drivers for Windows. However, they do expect working drivers and Microsoft tries pretty hard to make sure they don't start generating BSODs. [Like they did for Vista.]