LXer has an interview with project lead and creator of the Open Graphics Project, Timothy Miller. “I’m not sure exactly how large that market is, but we seem to be creating a movement. Individuals and businesses interested in open hardware are coming to me all the time, asking about it.”
As he said, coming out with an ASIC is expensive. Even the first computer to come out with a polygon-accelerated desktop chipset is now depending on ATI for theirs now.
This is interesting. It was only yesterday that I read the mailing list archives mentioned in the article and noticed Timothy Miller’s first post on the subject. I wondered what became of it. Seems like they are really making progress.
The real question is, of course, how much are you willing to pay extra for a graphics card that supports the Free Software ideology? I doubt they can cut the costs low enough to compete with low-end Radeons ($50 and less), for example, at least not for the first year or two.
If it has rudimentary 3D support (some basic acceleration, enough to run TuxRacer), great or excellent picture quality, good backwards compatibility and compatibility with existing standards (like VGA, VESA 2.0 or 3.0), I think I could pay $200. I need not mention Free drivers, as they are a given… This is four times more than the cheapest non-second hand card on the market, which is more than enough for the uses I put my computer to.
Here’s hoping we will one day see a final product.
I agree, I would happily pay $200 for a video card that had a completely open architecture.
From what I read of the article it seems like TuxRacer should run on what they are trying to make. I don’t believe that TuxRacer makes use of any custom shaders.
I’d pay a price premium, but in terms of performance it would have to offer at least on par, if not greater performance than existing parts that OSS drivers can be had for.
In other words,(and yes, I know that specs are not 100% open) ATi’s r200 class chips,(and below) and any Intel chip have OSS driver support. There are a handful of others. It would have to roughly equal or beat them.
Let’s hope they can actually bring a product to market, I’d like to dip my toe in the water.
I don’t care about an OSS driver if it’s gonna give me performance equivalent to an ati rage II+. I’ll stick with my firegl 8800 in that case.
The proposed GPU design has a 128-bit memory bus using 200MHz (DDR400/PC3200) DDR SDRAM memory. That’s 6.4 billion bytes per second or 1.6 billion 32-bit pixels per second. Some people have told me that that’s about the same performance you get from the Radeon 9200. (It’s a little more complicated than just 1.6 GP/S, but it would require a lengthy explanation.)
Hmmm……………….
I’d probably wait until the second generation then.
The card I have now is ALOT faster than a 9200.
But drivers could be what makes/breaks things. I’m running in a partially OSS driver, a fully supported driver for a 9200 chip might change things. While buggy, ATi’s drivers do have a *C*L*E*A*R* performance difference……. in 3D.
We’ll have to wait and see. But the proposed design…. That isn’t bad. Not bad at all.
Remember the Osborne? If you wait for the second generation, there may not be a first. The OGP will be especially sensitive to this.
To many users of OGP cards, Free Software drivers are critical. Adequate performance is important, but many people will switch to OGP cards on principle. It’s a community effort to create them, and it’ll be a community effort to make sure enough are sold so that the project remains sustainable. We want to design a product that is a GOOD product, but there are practical limitations that people will just have to deal with. Take consolation from the fact that many features of modern GPUs are hardly used at all by X11 and OpenGL, while the OGP design will be fully utilized.
Think about freedom, and think about the future. Buying an open graphics card is an investment in your future ability to get Free Software support and retain control over your own computer.
No, I don’t remember the Osborne. Got a link?
————-To many users of OGP cards, Free Software drivers are critical. Adequate performance is important, but many people will switch to OGP cards on principle.———
Therin lies the problem, at least to me. I’m not a linux/OSS user because of principles and ideology per se, I use it more because of it being a better product. I don’t have to constantly work on it and fix it. It just works.(most of the time. nothing is perfect) It’s faster, I don’t have to upgrade my hardware as much anymore. Security too. And I really like free upgrades.
OSS driver support to me is a matter of practicality, but if the performance is a downgrade from what I already have then it doesn’t match that practicality.
—————–Think about freedom, and think about the future. Buying an open graphics card is an investment in your future ability to get Free Software support and retain control over your own computer.————
But I already have that. It’s not complete, but it’s good enough to suit my purposes. As I stated earlier my chosen card is a firegl 8800. I chose it because of the partially open source drivers *as well* as the fact that the card is pretty fast. Even with handicapped drivers, games and day/day usage is very smooth.
Yes, full OSS drivers are a matter of interest, but if it’s not met with greater performance I’ll have a hard time getting excited. Sorry.
But that’s not to say that the full OSS drivers on a spec-wise inferior card couldn’t offer greater overall performance than a spec-wise superior card with handicapped drivers. Look at how much difference nvidia makes with their drivers, even on their MX line of cards.
Only time will tell. But when that time comes, I’ll be watching phoronix.com to see how the benchmarks stack up. If they’ve got it, I’ll get it.
For raw performance, OGP’s design will be slower than cutting-edge designs. What are you trying to do that needs to be so fast? The OGP design is spec’d to be as fast as it needs to be for a very good desktop user experience. All it needs to be is snappy and faster than your apps are trying to draw. Any faster is just a waste, because all it translates into is more idle time for the GPU when doing normal tasks. Do you need more idle time?
Living by benchmarks alone is a bad idea.
It is no way clear to me that an OGP design will be slower than the present closed-source cutting edge designs.
How many of those design really deliver a high percentage of thier rated specs? Just look at the size of the drivers they are delivered with to take account of the diffirent software packages out there, and if they don’t have the optimized code for a certain package of software how much of the top performance do you really get?
A full open sourced software and hardware card will let programmers get right down to *ALL* the registers if needed to get the best performance out of the card. There is no way to do that with a closed sourced design.
When shipped a number of programmers will be able to use the card to 100% of it abilities, at present with with exception of a few software companies under NDAs you just can’t say that of any closed sourced card out them.
This really counts for 3D, what do the specs says verses what we see delivered in the real world? An OGP design has the protential to deliver in the high 90’s percentage of it’s abilities.
Closed Source Drivers are highly optimized, that is in fact the reason that manufactors don’t want to open their drivers.
If you read the interview you will also see that they don’t have a design for 3d capabilities yet and probably won’t even support modern 3d features like programmable pipeline.
It is that simple. If you want Graphics Hardware capable of 3d Games, this one isn’t for you.
Honestly, I think you misread what I’ve typed over the past few posts.
———For raw performance, OGP’s design will be slower than cutting-edge designs.———–
My firegl 8800 is already slower than cutting-edge designs. It’s almost a 5 year old design.
———What are you trying to do that needs to be so fast?————
Nothing in particular. But if(when) I’m going to upgrade, I’m going to upgrade in every sense of the word. Performance, drivers, performance, features, power usage and oh yeah….. performance.
———-Do you need more idle time?————-
The idle time isn’t high on my concern list. My concern is when I decide to fire up a game to play. I don’t want to be in chop chop land.
I won’t be playing any doom3’s or quake 4’s, but when I do decide to play a game(my most recent favs have been neverwinter nights, bzflag, and I’ll be ordering X2 soon) I want to play at high resolution for better clarity.
———Living by benchmarks alone is a bad idea.———-
So is “living” solely by open drivers.
I’m trying to “live” by an entire package.
well, thats all on the premise that there is good in humans.
When i think about if i would buy a graphics card for 200$ (and i think that is a rather optimistic price for a small scale production) that probably has no 3d acceleration at all or only at radeon 9600 level, i know the answer to the question if i’m going to throw my geforce 6800 out for it… when hell freezes over.
But maybee the world will prove me wrong, isrealians and palastines love each other, the KKK helps rebuilding New Orleans and this graphics card gets bought out of friendlyness before it is on par with competition.
On the other hand, i do see chances for them in the ebedded market if they manage to come out with a product within the next two years, before nvidia and ati are getting more active in those markets.
———-or only at radeon 9600 level————
If the OGP offered R9600-level performance I’d buy one.
well, i mentioned that because it is around 60$ now, though i doubt they will be above radeon 9000 levels, which in my opinion already would be the minimum acceptable.
But honestly, if you want open source drivers, why not buy a mainboard with integrated ibm graphics? Then you still have the option to build in a recent graphics card to play games in windows.
Of course, that is the pure 2D performance. Hardware 3D is a completely different story, that is, if they will include hardware 3D at all.
I don’t think people are talking about 2d performance if they compare to radeon 9000 cards.
All I ask is for a card that can run Xgl etc. smoothly then I’ll be happy. The 3d games and other 3d intensive stuff can come later.
BTW, it would help exposure of this article if people would digg it:
http://digg.com/linux_unix/Interview_With_The_Open_Graphics_Project…
Go open graphics
.
I pay over the odds to use software that works anyway (windows tax + linux distro + compatible hardware), so why not for this?)
If just more than half of the people now using only open source software (GNU folks, OpenBSD users etc.) and/or not having 3D support for their small alternative OS would buy these cards, would(n’t) that be enough for the project to keep going? I’m sure that many such open source / niche OS users would be happy to have even some basic 3D support (hey, they haven’t needed high performance 3D cards before so why would they insist having them available immediately…) If the price won’t be all too high, and quality is ok, many such people will buy one of these cards.