Talking with Timothy Miller

barryman_5000 writes "Timothy Miller has written plenty of drivers for the open source effort and now kerneltrap has an interview with him on his newest effort for an open graphic card. He talks about his background, struggle with secretive 3D vendors and more."

Comments

Hey Hemos...

[BJH]

...what happened to this story? It's not like we haven't seen dupes before...

Re:Hey Hemos...

[tijmentiming]

Yeah indead! i replied to it, but now it's gone!

Re:Hey Hemos...

[spac3manspiff]

Now be a doubleplus-good slashdotter,
There was never a dup, anyone who thinks there was one has commited a though crime.

Linus was never seen on the cover of Business Week.

Re:Hey Hemos...

[tijmentiming]

yeah, like you have root access

June 2005?!

[jrutley]

I can't wait to buy one!

Re:June 2005?!

[wild_berry]

I cant work out if you're being sarcastic. You didn't mention running Duke Nukem Forver on it, so I think you may be serious.

And thank you for being the first post to talk about the card and/or article, not spelling or grammar.

In the Article, Mr Miller says he's heard quotes "I'd rather buy a used Rage128 from eBay". When this card becomes available, my Rage128 (Pro Ultra) will be on ebay.co.uk. Just so you can be ready ;)

I don't know if it will actually be an upgrade to my system, or a sidegrade, but I think this is like buying Fair Trade goods: those traded to afford a broad selection of producers some of the money thrown around by the rich developed world. I.E. I will buy one because I believe in the principle over (perceived lack of) the features.

duh

[Mr_Silver]

He has observed that there is a growing trend by graphics hardware vendors to provide less and less information to free and open source operating system developers.

This is because the graphics card market depends on vast amounts of R&D and producing a product that is technically superior to everything else out there. Essentially being continually ahead of the game as your competitiors try to catch up.

As much as OSS advocates would not like to hear it, opening up the graphics card specifications to all and sundry would be the equivilant of pooring your R&D down the pan. Selling support for graphics cards doesn't keep you in business - making a product that kicks the ass of your competitors (and them having difficulty working out how to beat it) does.

Re:duh

[pe1rxq]

There might be a lot of R&D, but also a lot of crap.

Thinking that you can hide this precious R&D in software without anyone seeing is nonsense. (The software interface is all that is needed for writting drivers). Your competitor is going to need at most a few weeks more before they dissassembled everything. If that is enough for them to steal your market your card wasn't as far ahead as you thought.

This 'black magic beyond us mere mortals' attitude is exactly what is to blame for this kind of thing spreading. (ie NVidea not releasing specs to their ethernet chip which ofcourse contains a lot of expensive R&D) Most stuff simply isn't as impressive as those companies want you to think.

Jeroen

Re:duh

[runderwo]

This isn't insightful or interesting. Apparently the parent has never heard of NDA contracts.

Re:duh

[tribulation2004]

Okay, for the sake of argument, I'll agree.

Now what's stopping them from releasing specs for cards which no longer ship (let's say graphics cards that are older than one year)? The problem is really that video cards supported by open source drivers are all 3+ years old (ie. obsolete). The performance gap between competitors is not 3 years; keeping specs secret after the performance gap has evaporated does not protect a company's lead.

Closed-source drivers do not allow for easy debugging/tuning, making the cards themselves useless if the manufacturer abandons support for the driver.

Re:duh

[farnz]

All a decent specification document reveals is how to tell a particular chip what to do, not how that chip does it. Any competitor knows what a chip is supposed to do (you can get that information by disassembling a binary driver, or by monitoring the bus while you send the binary driver commands), so the only "IP" you risk losing is the discovery that your "hardware" features are implemented by the device driver.

Re:duh

[DrSkwid]

making a product that kicks the ass of your competitors

only if you choose that arena

This Open Graphics Card will kick the ass of its competitors as far as a lot of prospective purchasers are concerned, and I am one of them.

It won't be sat in my XP box running Half-Life but it will be sat in my terminal and get the hours and hours of use it hopefully will deserve.

2048x2048 and up - yay, now that's kicking ass.

Re:duh

[radish]

What's an NDA got to do with anything?

How can I possibly write and publish an open source driver to an interface, if that interface us under NDA?

Re:duh

[Muttonhead]

Welcome to the new Slashdot. It is dumbed down. I think it started 9/11/2001. If you post outside a narrow range of tolerances you will be modded off topic. It's full of corporate trolls, just look at the gleeful Real Player adding mp3 story from yesterday. Microsoft almost gets more billing than Linux. I suspect AIBots or something similar posting crap messages trying to influence public opinion. I suspected when Slashdot was bought by a publically traded company it was the beginning of the end. I think I am correct. Malda should be publically humiliated.

I'm looking for the old Slashdot (a new place to learn/discuss) but haven't found it yet. Any ideas?

BTW, I'd give your message a +1 if I had the points to give.

Re:duh

[Hyksos]

This is exactly what I've always thought. It's not impossible to disassemble a program (driver in this case). It's very hard, but not impossible. I'm waiting for some super-geek to disassemble nvidia's linux driver and re-write the whole thing under an open license. He'll get his ass sued of course, but then the cat would already be out of the bag.

DVD-Jon, fancy a new (and proper) challenge? :)

Re:duh

[Lisandro]

As much as OSS advocates would not like to hear it, opening up the graphics card specifications to all and sundry would be the equivilant of pooring your R&D down the pan. Selling support for graphics cards doesn't keep you in business - making a product that kicks the ass of your competitors (and them having difficulty working out how to beat it) does.

Well, that's fine; i don't want the silicon blueprints for their beloved R&D. I just want specs on the interface which lets me use that particular hardware. As much as graphic vendors would like us to beleive, there's not much that can be "stolen" from gfx card specs. Don't take my word for it; just check the ones available for older cards and see how much you can get from there.

I think GFX vendors are reluctant of releasing specs for a number of reasons. One, it leaves them in a controlling position, since they dictate what you will and won't be able to do with your beloved card. Two, some parts of GFX cards might contain licenced technologies (stuff like MPEG decodig, perhaps? texture compression?), but still, we can do without. And three, almost every major GFX vendor has been caught cheating in their drivers (oh, oh, "optimizing"), which leads me to beleive more than one common GFX card might be software crippled. Hell, ATI had a card in which you could unlock four pipelines with a small program.

Desiging GFX hardware is hard, and writting driver is too. Yet, why can't you release specs for hardware we bought? There's an amount of zealotry to the OSS desire of open-source-for-everything, but if anything benefits from open source, that is system drivers. GFX cards or anything else.

Re:duh

[Lussarn]

As much as OSS advocates would not like to hear it, opening up the graphics card specifications to all and sundry would be the equivilant of pooring your R&D down the pan

They have found a company willing to produce an OSS friendly card, because that market is not taken at the moment. How is that "pooring your R&D down the pan".

Linux,t he BSDs and smaller OSes needs documented hardware. If the big players can't deliver new companies will fill that gap.

Re:duh

[gl4ss]

* (The software interface is all that is needed for writting drivers). Your competitor is going to need at most a few weeks more before they dissassembled everything. *

if it's that easy... surely it wouldn't be much of a problem that they don't release the code.

as to the "open source gpu".. it's going to be hideously expensive. chip producing takes real money and is cheaper the more you do them - it's expensive, very expensive. it's easy to come up with the idea for such a product, but to make it happen and make it happen so that the end product is faster than a 6 year old voodoo is going to be VERY difficult. so much that it would really be more point in buying gpu's that are old enough that the specs are out there for writing those drivers(cheaper and you'd get a faster product - and you really would have that product too and not just some wish-vapor).

Re:duh

[LWATCDR]

"Trouble is that it's not worth the effort for free software. Plus you end up selling hardware (by providing drivers) for a company that couldn't give a shit less about you."
I find that last part a big unfair. Nvidia and ATI both provide Linux drivers for their cards. I have heard that ATIs suck but they do atleast provide them.

Re:duh

[PedanticSpellingTrol]

Eh, just do it compaq-style with two super-geeks. One reverse engineers the existing driver and documents the spec, then the other writes drivers from that spec without ever being allowed to look at the original driver

Re:duh

[Spy der Mann]

This is because the graphics card market depends on vast amounts of R&D and producing a product that is technically superior to everything else out there. Essentially being continually ahead of the game as your competitiors try to catch up.

I'll probably be modded down for this (as redundant lol), but I'll just explain to you. The inner workings can be kept as secret as you want. But the SPECS (like the API) have to be public.

When the card designers keep their specs secret, it's not about competition. It's possibly about having some contract with Microsoft etc, so only Windows can have decent drivers. Otherwise it seems illogical that they would hide this stuff.

In summary, there's nothing wrong about having open specs.

Re:duh

[renoX]

> Hell, ATI had a card in which you could unlock four pipelines with a small program.

But remember that those four pipelines may work *or not* as it is also an easy way to disable faulty units, thus allowing the HW makers to sell at reduced price something which should have been trashed otherwise..

Re:duh

[Lisandro]

Actually, it's more common than you think. Some nVidia-powered GFX cards brand were softmoddeable too; i remember reading something about the 6800 and software locked pipelines.

The 9700 -> 9800 ATI softmod was supposed to be doable on 9700s that were, like you said, 9800 with faulty units. I know four people that did this mod and all had the card running just fine - only one had artifacts, and was due to the GPU running much hotter, which was fixed by sticking a larger heatsink. ATI recalled that particular model after the mod was public, so i really dunno...

Re:duh

[trevorcor]

The way I understood it (from something Alan Cox mentioned in the original LKML thread that started this project), the two big card vendors have so much cross-licenced IP that they're both terrified to even consider making any information public; it would open them up to lawsuits from the other (legitimate, or maybe not so much).

Re:duh

[Brandybuck]

Precisely. Releasing the specs doesn't compromise your onboard hardware secrets. "To do X shove Y into register Z." But there's this recent attitude that you can't let anyone know that there's a register Z.

Re:duh

[Brandybuck]

But that's still no reason to disable them in the software. Simply don't connect up pin and you're done. Just like with the 486SX.

Re:duh

[renoX]

Maybe it's just cost 1c less to disable them in software than in HW?
OK, the number of users which disable the protection will be much higher, but I suspect than this number is so small that it's "lost in the noise".
Also this number can't be measured while the cost saving by disabling a functionnality through software vs pushing the chip an additional tool to disable some pins is very easy to compute.

Re:duh

[LWATCDR]

What you do not get is that most Linux users want "working" drivers. All graphics cards come with free as in beer drivers. I think you fill find this card goes no where. It will be slower and more expensive than the "closed" cards from nVidia. While it would be nice if nVidia would release the information on their video cards at least give them some credit for producing good drivers that do work under linix.

Re:duh

[WaterBreath]

I just want specs on the interface which lets me use that particular hardware.

I want to say up front that this is spot on. The card company doesn't lose any money by allowing other people to work on drivers for their software. In fact, they can only sell more. If they are hiding higher-tech features on cheaper cards, they should be hidden further behind the interface, not at it.

That said, I will address another part of the parent's post.

almost every major GFX vendor has been caught cheating

I don't know if you really feel that this is "cheating", but if so, what would you suggest as an alternative? What they do makes perfect economic sense. They've put in the money on R&D for a hi-tech product. Once the design is implemented, manufacturing cost differences between the multiple levels of cards are negligible. But they still paid to research the hi-tech stuff, and so if they just charged less for the expensive features, they couldn't recoup costs.

So if they didn't disable those higher-tech options, then they would simply have no cheaper cards available. Which means they'd sell less, and possibly not recoup the costs of their research. Instead, they disable the "most expensive" features and sell a cheaper version, which opens a new market to them: people who want brand-X cards, but aren't pay an arm and a leg for a top-of-the-line card.

The other option is to make two separate designs, and completely exclude the ability to support those features from one of them. But that's even more design and testing work that really isn't necessary. Remember, the cards work because of the research that went into the features, so they're charging you for use of the features, not physical presence of the features.

It's really no different than airplanes that have multiple classes of seats. 1st class gets better service. And even though coach is on the same plane, and technically has access to the same resources as 1st class, they paid less, so they don't get to take advantage of them. It's a cost effective way of offering services at different price levels, so that large portions of the market aren't excluded.

This should make sense to anyone who has taken a microeconomics class. And anyone who hasn't should be asking questions before they make judgments.

Re:duh

[Lisandro]

I don't know if you really feel that this is "cheating", but if so, what would you suggest as an alternative? What they do makes perfect economic sense. They've put in the money on R&D for a hi-tech product. Once the design is implemented, manufacturing cost differences between the multiple levels of cards are negligible.

I need to apologize; i mixed my thoughts a bit here (typed on a rush at work :). Graphic cards vendors HAVE been caught cheating with their drivers, doing stuff that they're not supposed to (the infamous "quak.exe" incident? 3Dmark? etc etc etc) and that might be a reason why they don't open source their drivers; there might be stuff there that we're not supposed to know about.

As for the open specs parts, well, you have a point. This happens with a lot of electonic devices nowadays, where a simple firmware update usually unlocks a ton of features not available before. The thing is, you don't need a deeper understanding about your, say, digital camera other than the USB data transfer specs in order to write OSS drivers for it. For a GPU, you need the whole deal.

It's a tough position for manufacturers, but it makes little diference for me, as an user, as i'd rather have OSS drivers which i know are continuously being patched and have a lot of eyes over them.

Having said that, atleast nVidia offers excellent closed-source drivers for Linux/BSD, rivaling their Windows counterparts in performance. I never had an issue with them, but that won't keep the zealots happy ;)

Re:duh

[WaterBreath]

Sorry for the somewhat harsh response due to mixed signals there. We were obviously (now =) talking about different kinds of cheating.

The thing is, you don't need a deeper understanding about your, say, digital camera other than the USB data transfer specs in order to write OSS drivers for it. For a GPU, you need the whole deal.

Good point. I hadn't thought of that. I guess I see the threat now, and the difficulty of ensuring that an open interface doesn't also allow work arounds for the crippling I talked about. Also, if their "secrets" are patentable they shouldn't be worried about hiding those, but I can understand wanting to keep some non-patentable, "flash-of-inspiration" tricks hidden.

atleast nVidia offers excellent closed-source drivers

I don't suppose you'd know the location of some Geforce 3 drivers for Linux? I've been having trouble finding them. Maybe I just need to work on my googling skills?

Re:duh

[Brandybuck]

Don't actually remove the pin, but simply don't have it connect anywhere on the PCB. I only used the 486 as an example, not a model. You're going to have different PCBs anyway, because they will be labelled differently. So that one cent cost evaporates.

Besides, any company that seeks to save one cent on a fifty dollar cost board deserves the flak they receive when someone hacks the driver.

Re:duh

[Lisandro]

nVidia has a very nice treat: Unified driver architecture, which means drivers are the same for all GeForces and TNTs, from the original TNT up to the latest FX offerings. In practice, it means more or less they bundle all drivers in a single download (a 8mb binary driver? sheeze) - but they work just fine. This is for all platforms too; you can get them here.

As for OSS drivers, both X.Org and XFree86 include a basic TNT/GeForce driver with some degree of 2D acceleration but no OpenGL support.

Re:duh

[Tough Love]

as to the "open source gpu".. it's going to be hideously expensive. chip producing takes real money and is cheaper the more you do them - it's expensive, very expensive

Read the article. It's a FPGA design, not a chip. Or do you prefer to spout first, get caught later?

Re:duh

[Tough Love]

I think you fill find this card goes no where. It will be slower and more expensive than the "closed" cards from nVidia.

I thought that at first as well. But I changed my mind after looking at the design drafts, the card specs, and the quality of the people involved. This is going to be the classic enthusiast's card. It's reprogrammable at the logic level! The uses aren't limited just to Linux video.

The point of this thing is to open it up to widespread hacking. I'll gladly pay considerably more than a commodity card to have it hackable. Just think, this thing could spawn a hardware demo scene for one thing.

I'm pretty sure there are enough like-minded people to make this project commercially viable. It doesn't have to take over the world, it just has to fly, then the sky's the limit. There is of course nothing stopping a spinoff card from being developed in hard silicon after the original is thoroughly debugged, which would bring the unit cost way down, and be much easier to cost justify. But as far as I can see, this is going to fly just fine as an FPGA design.

By the way, it's an OpenGL card. It won't support programmable shaders, at least in this version, but it will be capable of running Quake III. Of course, you will be able to implement your own shaders at the gate level if you are smart enough and some people certainly will be, I look forward to some mind-blowing demos. You could also set the gate array up to do something entirely unconnected with video, such as run a kick-ass synthesizer or image processing or encryption.

I want it now.

Re:duh

[runderwo]

The ATI open source driver was developed with NDA documentation. If it weren't for the NDA, it wouldn't exist at all.

iD-eal project for Carmack

[DrSkwid]

JC should stick some of his $ behind this project instead of making rockets.

Re:iD-eal project for Carmack

[Chanc_Gorkon]

I bet he would not touch this with a 10 foot pole. Why? Because Armadillo is FUN for Carmack. This project whle fun to alot of us, would likely not be fun for Jon. Why? Cuz he sites in front of a computer at least 10-12 hours a day already writing code. Volunteering for this would make him do it longer.

Re:iD-eal project for Carmack

[dpilot]

JC already stuck his TIME behind the original 3D drivers for XFree, back in the Utah-GLX days. He was there early on, and has done his time.

That was also a key part of my deciding to buy a Matrox G400 several years ago.

Re:iD-eal project for Carmack

[catch23]

Your statement is equivalent to people saying stuff like "Stop spending money on yourselves and send that money to all the starving people in Africa!" He does the rocket stuff for fun. He enjoys it more than writing graphics related code. I'm sure there are skills that you are talented in, but I'm sure you don't enjoy doing your skills when you're on vacation. Making rockets is like a vacation to him. I really enjoy software development, but when I go on vacation, I'd rather not be sitting in front of the computer.

Re:Erm...

[koi88]

"Has wrote?"

Should be: "Has writed"

duh-IP Security.

[Anonymous Coward]

Funny A similar argument can be made for any IP, including software patents (especially software since that's all the product is).

Re:duh-IP Security.

[Aurix]

IP Law gives them the right to sue for people illegally using/gaining access to the material.

Sure, there's only so much they can do to protect their IP, but why shouldn't corporations have the right to sue the shit out of people stealing their R&D?

Re:Grammar, pleeze

[YankeeInExile]

...what he said.

I fear you're going to lose your audience of yore with loose use of the English language.

Re:Grammar, pleeze

[adeydas]

Well if you read their FAQ's you will see that they don't quite look after grammar since its the content that matters. Since they don't look into the matter too deeply, no matter their are so many dupe stories here at /. You can mod me down now.

Re:Erm...

[KillerDeathRobot]

No, "Has writified."

At least, that's what my Don King Dictionary says...

Re:Erm...

[BJH]

"Hev Writnod"

Re:Erm...

[nomadic]

Didst writeth?

Re:But....

[Saeed al-Sahaf]

No, but it _will_ run OpenKnopixCatFoodBasementMasturbater beta v. .0006.6.6. And *IF* you are not running OpenKnopixCatFoodBasementMasturbater beta v. .0006.6.6, then obviously, you are a looozer.

Tim

[Ratbert42]

Same Tim

Alternative OSs

[DrSkwid]

These platforms, both free and non-free are valuable alternatives to the Microsoft monolith

I hate this. I don't use an "alternative" OS any more than I drink an alternative to milk or live an "alternative" lifestyle.

I know it's grammatically correct but it's the hidden implication that does my head in!

Re:Alternative OSs

[radish]

What's the implication? I don't see one. Alternative means "different", and Linux is different to Windows. Any bias you have about the use of the word Alternative is yours, and you'd be well served to rethink it. Many bad things have good alternatives.

Re:Alternative OSs

[ggvaidya]

I know it's grammatically correct but ...

For once , today. I think the editors deserve a round of applause!

[It's alright, Hemos, you know we love you :D]

Re:Alternative OSs

[Phil246]

compairing microsoft to a monolith scares me. i mean what if its of a ratio of 1x4x9 in dimensions?!
will jupiter get turned into a sun , only for it to supernova because one of the monoliths crashed? :)

Re:Oh for Pete's sake

[nomadic]

The funny thing is I can see the original editors having poor grammar, but all the newer editors seem to be the same way. So many unemployed English PhDs they could have hired, but I guess it's cooler to give your grammatically challenged friends cushy jobs where they don't really have to do anything for their salary.

All we need now is a petition for its support!

[essreenim]

This is really interesting

..and all graphics manufacturers that did publish specs have stopped doing so..

Y. nVidia are probably the most helpful to the community yet have 2 sets of Linux drivers - the OSS ones and the closed (official) ones.

A: Does the project have an official name?

Timothy Miller: Depends on what you mean by "official". We're calling ourselves the "Open Graphics Project",

cool, now Ill be able to play the doom3 engine with OpenGL and OpenGP ..

All geometry and vertex processing will be done in software in the host computer.

This is a bit disappointing. Ever played a game in S/W mode? Nightmare - last century. At least its only a part of the processing though.

Keep in mind that no graphics card on the market can fully support Doom III, with all features turned on, at a high framerate. So the fact that a card like this couldn't handle it shouldn't surprise anyone.

True. I still turned up all my settings for Doom3 though and just played at a lower fps on an nVidia Geforce FX 5700 256. It was playable (on Linux with the latest nVidia Linux drivers)..

Anyways, I'm getting one as soon as it comes out, if it comes out!!

Re:All we need now is a petition for its support!

[Tough Love]

"All geometry and vertex processing will be done in software in the host computer"

This is a bit disappointing. Ever played a game in S/W mode? Nightmare - last century. At least its only a part of the processing though.

It's also a fairly cheap part and well suited to SSE processing. You wouldn't want your processor to be sitting completely idle, wasting all those expensive transisters would you?

Something to remember

[Johan Veenstra]

When you're reading ./, try not to get too annoyed by bad grammar. Virtually all past and present editors are terrible at spotting even the most obvious language deformations. Let's face it, bad grammar is here to stay. Let's just hope it's not contagious.

What are you talking about?

[agraupe]

"Has wrote" is perfectly cromulent grammar!

Some people these days...

Re:What are you talking about?

[mrmagos]

I would *SO* mod you up right now if I had the points. Some people's children....

Re:Erm...

[verloren]

No, I believe the correct phrase is "Done wrote"

It's not really their fault...

[dteichman2]

barryman 5000 wrote that. Not the /. editors. Just because you hate those editors because of the fact that they reject every article that you send them DOES NOT mean they are responsible for spelling and grammar errors. The job of the /. editors is to approve stories that are interesting and relevant to slashdot.org

barryman 5000, pleez goo bak an laern Eanglich.

Re:It's not really their fault...

[William_Lee]

I guess /. has redefined the definition of "editor" and forgot to tell the rest of us.

editor
SYLLABICATION: editor
PRONUNCIATION: AUDIO: d-tr KEY
NOUN: 1. One who edits, especially as an occupation.
2. One who writes editorials.
3. A device for editing film, consisting basically of a splicer and viewer.
4. Computer Science A program used to edit text or data files.
ETYMOLOGY: Late Latin ditor, publisher, from Latin ditus, past participle of dere, to publish. See edit.

edit
SYLLABICATION: edit
PRONUNCIATION: AUDIO: dt KEY
TRANSITIVE VERB: Inflected forms: edited, editing, edits
1a. To prepare (written material) for publication or presentation, as by correcting, revising, or adapting. b. To prepare an edition of for publication: edit a collection of short stories. c. To modify or adapt so as to make suitable or acceptable: edited her remarks for presentation to a younger audience.
2. To supervise the publication of (a newspaper or magazine, for example).
3. To assemble the components of (a film or soundtrack, for example), as by cutting and splicing.
4. To eliminate; delete: edited the best scene out.

Re:It's not really their fault...

[Johan Veenstra]

In that case, they should reject submissions with such glaring grammatical errors: "sorry your submission has been rejected, because of your sub standard use of the english language".

But I guess that's too much to ask from editors, who do not even check the content (dead links, substance), and continue to post duplicates.

gonna party like it's 1999

[justins]

All geometry and vertex processing will be done in software in the host computer.

Geometry acceleration was a newfangled, fancy feature five years ago. Something that previously required thousands or even tens of thousands of dollars worth of hardware. Now it's a considered a given, and there is no reason to buy a card without it.

This card is going to suck. :(

Re:gonna party like it's 1999

[dteichman2]

Why the hell is this even a possibility? Pick up a fucking Radeon card (any model #) and put it in your computer. Every Radeon card seems to work fine on any flavor of any OS. You can buy a Radeon 7000 now for cheap. Guess what? All of them have built-in Geometry and vertex processing.

Try playing and Half Life mod in "Software" mode and tell me the kind of performance you get. That's a lot better than you'd get from this concept card (which happens to be very backwards).

Please excuse me. I need to get back to Intel about my new idea for a "Open Pentium." It'll run at the speed of a 486, but at least everyone will know exactly how it works. Although, something tells me that people would rather have decent performance, and not wait a week to install a modern OS.

Re:gonna party like it's 1999

[the_maddman]

But they're implementing this in a FPGA. Geometry processing would take up to many gates in the FPGA. Just fitting what they have listed already will be a challange. If they can make it work, I'll buy one. Maybe the next generation will be an ASIC and have more gates to spend on features.

Re:gonna party like it's 1999

[Anonymous Coward]

Have you tried the intel embedded GFX - the Prescott was quite good and the Grantsdale is positivley great.

Intel's approach is the same as is used here - there chipsets don't do any Geometry of vertex processing, they're essentially a pixel unit. Intel run all of the vertex processing on the CPU.

Of course that's also in Intel's interest as it allows them to get extra value from their CPU's. Geometry vertex processing is perfect for parrelisation through SSE type extentions.

In DirectX speak the vertex shaders are run on the CPU but the pixel shaders are run on the GPU - the new Grantsdale chipset supporte pixel shader model 2.0 and below IIRC.

What's more suprising is that in many cases you won't see a performance hit using the grantsdale over a full GPU with vertex unit.

5 years ago this was a great step forward (about when I started comercial work on GFX programming hardware) TnL (as it was called back then by nVidia who gave it us first in the GeForce series of cards supported by Direct X 7.0.

However these days quality demands that we light and do many many operations at the pixel level rather than at the vertex level and it's much cheaper to have a bump map than 1000 extra verts no matter which way you look at it. Therefore most algorithms for doing things from will hit the pixel units hard but the vertex work is trivial in comparison.

With programable shader models the vertex units and pixel units on GPU's are becoming more and more similar. I'm pretty sure within a generation (2 at most) they'll just be shader units that can be switched from pixel to geometry utilisation on the fly with a clever memory architecture to make the pipeline feed back into itself rather than a standard waterfall type flow.

Now from TFA Tim was talking about stalled CPU's and ways of using those cycles. I can tell you that developing a video game (on any platform, PS2, XBox but especially PC) these days you're very very unlikley to get stalls waiting for the vertex units and if you do rearranging your sequence of rendering will overcome that. However you're much more likley to get stalls waiting for the pixel units. Using techniques like defered rendering and having a nice big command buffer however means that these CPU cycles can be used for useful work - why not do the vertex processing on CPU then? In my current project I think we throw in some extra AI processing where we can in such stalls so we would see a slight performance hit from using a software pipeline. However I think it's rare for anyone other than games developers to really squeeze so much out of the system and many games developers don't bother if they don't need to.

Infact for most scenes I've tried the performance of the pixel only grantsdale is almost as good as a middle to top end ATi or nVidia part.

I can see no reason for the Open Graphics Project board to perform significantly differently to the Intel chipsets - although what they're targetting is a little behind the grantsdale that's their top end as they don't support programmable models.

Essentially what I'm saying is the world has moved on and if you want 5 year old features buy an old GPU or one which has to be backward compatible, otherwise if you want what's important now then get a board that is designed for todays requirements - not what they were 5 years ago.

This open board is very sensibly designed to perform well in todays market with todays needs whilst being a manageable project in terms of scope and duration and I for one applaud it!

I would be interested to know how well intel support it's chips linux and how open they are about their chips specs. It'd also be interesting to see what their stance on their existing written software vertex libraries (heavily _x86 optimised but probably never going to port) being available to plug into other driver sets as a more generic vertex processing library.

Re:gonna party like it's 1999

[Taladar]

Every Radeon card seems to work fine on any flavor of any OS.

Even though I have a NVidia Card I hear lots of Radeon Owners complaining about the bad quality of ATI's drivers for Linux.

Re:gonna party like it's 1999

[mallan]

Yes, I've had a lot of problems w/ my Radeon cards (original Radeon and a 9200 in my laptop) - both the Open Source and proprietary drivers have lots of problems (although I've had much more trouble w/ ATI's binary only drivers)

The thing that everyone seems to be missing is that the specs for r100 and r200 Radeons *are* available, and even after 5 years of open specs, the open source community still hasn't been able to write a high performance, high reliability driver for the early Radeons. Writing a good 3D driver is a *lot* of hard work. This guy and anyone who's going to help him would be much better off putting their efforts into bullet-proofing the Radeon driver - it'll be far cheaper and far more capable than a home-brew custom FPGA based card. At best, this card is going to have the performance of a Rage128 (which has open specs) or a TNT2, if he's lucky. Why would anyone pay $200 to get the performance of a $10 card? It's silly.

-M

Re:gonna party like it's 1999

[CaptnMArk]

He was probably thinking of Radeon = 9200 which has open source drivers.

(I have one, and can confirm that these drivers are more stable than nvidia ones. I did play some quake 3 with them some time ago. Performance is lower that you can get with nvidia though, ie. no doom 3).

Re:gonna party like it's 1999

[Brandybuck]

This card isn't being created for gamers. It's been created for desktop users. KDE, GNOME, Qt and GTK+ will want hardware 3D acceleration to do all sorts of fast rendering. But they won't need in-hardware geometry and vertex processing.

Yeah, it will suck playing Doom III on this, but that's not the target market.

Re:gonna party like it's 1999

[justins]

This card isn't being created for gamers. It's been created for desktop users.

Then the question becomes "why bother?" If you don't need the most modern OpenGL features in silicon there are cards which are documented quite well enough to have completely open-source drivers. (In fact there are cards like some of the Radeons that actually HAVE some of these features in 100% open source drivers)

It's nice that somebody has a pet project, though.

Re:gonna party like it's 1999

[Anonymous Coward]

Because some of us would like to enjoy 3D accelerated desktops in Linux without having to resort to eBay to find replacement units.

This isn't a pet project, it's about keeping the future open for all FOSS platforms.

Re:gonna party like it's 1999

[Brandybuck]

OpenGL != games

As the article said, if you bothered to read it, "2D is a special case of 3D". The 2D specific work that will be done is only optimization for the 2D case.

Mac's Aqua desktop requires a 3D accelerated card, even though there's nothing on that desktop that is 3D.

Re:gonna party like it's 1999

[justins]

What the hell does that response have to do with what I posted?

Re:gonna party like it's 1999

[Brandybuck]

To clarify, you use 3D to do 2D stuff. For example, transparency and compositing, or rotation and scaling. You could of course do this in 2D, but by doing it in 3D you get both the 2D and the 3D.

There are completely documented 2D cards. But they're not on the market anymore. And they won't do this advanced 2D rendering.

Re:gonna party like it's 1999

[justins]

Sure, I understand that, but there are also thoroughly documented cards like the Radeon 8500 (and whatever the newer version of the same is... the 9100?) which offer these features with open-source drivers.

Which isn't really QUITE the point, the point is, even the substandard (when compared to the binary drivers) open source drivers for some newer Nvidia and ATI cards will offer everything Miller's card is meant to do, cheaper and faster.

It's hard to see what market niche he's filling, aside from the hobbiest gadget freaks. I obviously wish him luck, although I can think of much more useful projects for someone who is willing to go through the rigamarole of bringing a card to production in an open source way. How about an 802.11g card that doesn't require loading proprietary microcode from the driver?

Re:gonna party like it's 1999

[dteichman2]

I've been running one box with a Radeon 7000 and RHL 9. It's been running 2 years. No reboots.

Re:gonna party like it's 1999

[Brandybuck]

How about an 802.11g card that doesn't require loading proprietary microcode from the driver?

That's the next project :-)

Re:gonna party like it's 1999

[dteichman2]

Er... my laptop has a Radeon 9200 in it. Doom 3 works just fine. Performance is fine with some reduced quality and effects.

Re:Grammar, pleeze

[DrSkwid]

Just like when you write a book and your editor just sends the manuscript to the printers.

Do you even know what an editor is ?

Re:Yadddaaa

[dteichman2]

Umm.... the whole point of Open Source Software is that EVERYONE should give away their collective work. The resulting work "pool" cuts development time for other work. Thus, promoting more work and improving the quality of all software.

Re:second damn article in a row

[Atrax]

Modded redundant? That'd be about right. Grammar on slashdot is fucking redundant.

nice one.

Re:Grammar

[musselm]

My favorite is when a grammar-correction post gets something wrong as well, such as:

"Sorry, it's just that seeing ... and then this not an hour later were too much."

The word "seeing" is a singular noun. It should read "... was too much."

Enjoy!

Minority OS

[bill_mcgonigle]

But you do use a minority OS. And vendors do discriminate against their minority-OS customers. Draw whichever parallels you see fit.

Re:Erm...

[WillerZ]

Nope:

dun writ
-or-
dun writted

would however have been acceptable.

oh yes, you have found them out

[justins]

blah blah

Your competitor is going to need at most a few weeks more before they dissassembled everything.

blah blah

Most stuff simply isn't as impressive as those companies want you to think.

Which is why driver software has never had anything to do with the success or failure of graphics cards in the market.

Re:oh yes, you have found them out

[farnz]

Buggy drivers hurt, regardless of the standard of the hardware; if the hardware is any good, the specs needed to write a driver do not contain any "super precious IP"

Exactly

[gandell]

Especially since it's /. , not ./. I'm sorry, old chap, but I had to. :)

Re:Exactly

[Johan Veenstra]

At least there was nothing wrong with my english :) ./ is just so much more nutural to type. I use it a lot more often: ./my_program params

Re:second damn article in a row

[Atrax]

fair enough. you're right.

sometimes poor grammar just makes me too annoyed to see beyond the red mist.

I would by this card in a second.

[Anonymous Coward]

Why because I am a zeolat?

NO!

I beleive in Free software, but this is a very personally selfish reason!

I like the PowerPC platform and like to screw around on it. I know that x86 is cheaper and faster, but for what I use a computer for the Ibook is plenty fast.

However I will never buy another mac product again because newer cards are either Nvidia or ATI. The current Ibook has a ATI 9200, which is supported by Open source drivers, which means that it works with PowerPC and x86.

Nvidia and ATI binary drivers are only for x86!

So I can not every have 3d acceleration on a powerpc machine again? If I buy this OSS-friendly video card I can.

I want stability, I want freedom to use non-standard hardware setups.

I want to get 3d acceleration in OpenBSD, too!! Not just Linux!

OpenBSD is very secure, but it's worthless for the blender 3d stuff I use because no nvidia drivers or ATI drivers work with it.

The point of open source is choice. And there are very real technical reasons behind keeping all the software I use free, too.

Read this artical about "pointless ideology" and you will understand what I am talking about.
http://lwn.net/Articles/100098/

it is very important to have the ENTIRE OS Free, and not just have it no-cost.

Re:I would by this card in a second.

[Wiz]

Nvidia does have FreeBSD drivers though.

OpenBSD isn't optimised for real performance either, more so security. So perhaps running blender would be better suited by FreeBSD.

But yes, it is x86 only.

Re:I would by this card in a second.

[Metteyya]

Good point there, Richard (or should I say st. IGNUcius?).

Now, seriously - outside x86 and binary-only drivers people really start to understand the whole purpose of Free and Open.

Thx mods

[barryman_5000]

I type "has wrote" and nearly 40 comments about how dumb I am. Next time I won't even submit news and you people can go to kerneltrap for yourselves ;) Thx mods for fixing it to "has written"

Re:Thx mods

[Johan Veenstra]

Ofcourse, next time you could take another 2 seconds to check your submission for errors.

Re:Thx mods

[Yaa 101]

Don't worry, you will grow a thick skin sooner than later, at least you have a item submitted, I bet most of the grammar winers did not... ;)

Re:Grammar

[Neuroelectronic]

THe english language was destroyed a long time ago. By Slang and Ebonics and L337 talk and people from Luisiana ---- I spell like a programer.

whither sound?

[spoonyfork]

How can he be motivated to work on a universal open sound card driver instead? All my video cards have worked fine under linux... sound cards have not fared as well. :|

sweet jesus

[nomadic]

DOES NOT mean they are responsible for spelling and grammar errors.

But...but...they're EDITORS. EDITORS!! Oh god my head's about to explode, too much wrongness in one post!! What's the matter with you?!?

Petition If you're planning on buying one of these

[iplayfast]

Make you visible to TS new open 3d videocard

Let the company know what the demand is.

FOSS

[Deanalator]

I would have no problem dropping 100 or so dollars on a card that could do 2000fps in glxgears with an open source driver in the main kernel tree (on par with my current geforce3 ti200).

These shoddy nvidia drivers really bug me, and it would be nice to see a hardware accelerated opengl X enviroment sometime in the next 5 years (before longhorn), and that is never going to happen unless we can get some real hardware support.

Re:FOSS

[gzunk]

You think the nVidia drivers are shoddy... Just be thankful you haven't got to use ATI drivers.

Re:FOSS

[Brandybuck]

Actually, the OSS drivers for ATI aren't that bad. Unfortunately you need an older card and you're not going to get any sort of optimization. As the article stated, a lot of people are still going to go the route of a $20 Rage 128 off of eBay. But some of us are excited about the possiblity of a non-cripple video card that is less than two years old.

A question not asked in the interview

[jmichaelg]

You couldn't get card driver specs from the card manufacturers but did you approach ATI or nVidia about building a card aimed at Linux using their chips?

Re:But....

[Theovon]

Sure, I don't see why not. With open specs, it can be ported to ANY platform.

Re:Driver's License?

[Theovon]

Anything Tech Source writes, in terms of software/drivers, will be released under a license which basically says "pick whatever open source license you want to use for this".

Incorrect:

[jimbro2k]

Not "has wrote"...
The correctest usage is "He done gone wrote"...

Lots of people have done this

[Theovon]

People have been asking nVidia and ATI to open their drivers for a LONG time. They still don't want to do it.

Re:Lots of people have done this

[jmichaelg]

The guy is talking about manufacturing video cards, not just writing a driver. In short, he's talking about putting real money on the table. If he has in fact asked nVidia and ATI to enter into a business relationship and they still demur then something else is going on.

Re:Lots of people have done this

[lachlan76]

If he's gonna use the stock nVidia drivers, why bother manufacturing the card?

I'll buy one

[starseeker]

I'm not a huge gamer. bzflag is about the limit of my occasional forays into games these days. If it can:

- accelerate all the eye candy I enjoy
- make things like alpha transparency and video rendering fast and smooth and not impact system performance
- allow me to manipulate 3D plots or complex CAD objects in three space in real time smoothly

then it does what I need from a graphics card. If it can make bzflag run smoothly, so much the better. And I suspect a middling card with excellent drivers will stack up OK for normal worka against a really fast card with iffy drivers. Plus, if this is a success they might make better cards in the future.

Guys, let's make this the standard card for non-gaming open source boxes. Especially if it's a quality piece of work. That counts for quite a lot, too - solid hardware is a blessing if you don't have the $$ to casually replace it.

Re:I'll buy one

[wormbin]

Me too.

Stability and reliability (over time) are the most important features of my linux box. Open source specifications means the product can never be "end of lifed" by the manufacturer. When nvidia or ati decides to stop supporting old cards in their new drivers then you are one kernel upgrade away from an non-functional graphics card.

I'm guaranteed to buy one card to check it out. If it works well, I'll buy one for each of my linux boxes.

Re:I'll buy one

[Tough Love]

"Guys, let's make this the standard card for non-gaming open source boxes. Especially if it's a quality piece of work."

It's going to be able to run a lot of good games. Not Doom 3 or anything that relies on programmable shaders, but it should be able to handle Quake III level engines.

Re:Grammar

[x40sw0n]

The english language isn't dead... just evolving. Check out Jeff Noon's Cobralingus [ISBN: 1899598162](or really any of his stuff, he is an excellent writer). He has a very ...organic... view of language. needle in the groove doesn't even use standard punctuation. lots of "/"s to indicate pauses etc... in fact i don't think there is a single period or comma in the whole damn book. Nymphomation is my personal favorite.

It seems that the ignorance is yours!

[Theovon]

Contrary to what you say, this card is designed a priori to accelerate 3D graphics. The rendering engine is designed to be OpenGL compliant.

It seems that you didn't read the article and have no idea what this project is all about. None. At all.

What about games?

[Edulix]

I've read most of the interview (sorry, it's time of exams and I don't have much free time available). Quote:

JA: How well should these specs be able to handle the graphics intensive games that are currently out on the market, and those yet to be released?

Timothy Miller: Keep in mind that no graphics card on the market can fully support Doom III, with all features turned on, at a high framerate. So the fact that a card like this couldn't handle it shouldn't surprise anyone. [And cotinues...]

I was thinking about buying a new graphic card so that I could play some new games, such as Half-Life 2 or Doom III. I'm not a gamer nor I know much about graphic cards, but I like to play videogames from time to time. I really like the idea of an open graphic card, and I certainly would wait and pay 200 for this one.

But I need to know: which games would work fine in this graphic card? Will it play Doom III or Half-Life 2 *without* all the bells and whistles at a decent rate? If so, define to us "decent rate", also :-).

Thanks for your time and good luck with the project,
Edulix

Re:Grammar

[lumpenprole]

That's one way to look at it.

Another way might be that it's a constantly evolving open source project with a practically infinite number of contributers.

You may not appreciate the forks, but that doesn't mean you have to be a dick about it.

Wireless

[Ratbert42]

Now can you write some decent drivers for WiFi cards?

I have four words for you

[glMatrixMode]

R-T-F-A

It's beginning to sound good

[pherthyl]

My initial reaction to this project was "Bah, who cares about open source, my NVidia card works fine for under 100 bucks canadian".

But the more I read about this, the more enticing it sounds. I don't play games on Linux at all, so I don't care about that. And to have a nice driver, that is optimized for the new features in X11 like XRender and stuff would rock. The longer I use Linux, the less I want to bother messing around with compiling modules, so I don't even bother using the official NVidia drivers. Sounds like this will perform much better than the generic, 2D only, NV drivers.

Who knows. Might actually buy one of those. But his projected price point at $200 is too high. Even in my best "Stallmanesque" spirit, I can't justify spending over $300 canadian on this card.

Re:It's beginning to sound good

[Tough Love]

Who knows. Might actually buy one of those. But his projected price point at $200 is too high.

Sentiment seems to running around a $150 price point as a reasonable compromise. I know that for me, $150 versus $200 makes the difference between grabbing one right now or thinking hard about it.

"fully support doom iii"

[bani]

Keep in mind that no graphics card on the market can fully support Doom III, with all features turned on, at a high framerate.

nonsense. nvidia 6800 series can do it just fine. 6800 ultra can support 1024x768 with full doom3 features turned on, and still get faster than monitor refresh (>100fps).

sounds great!!!!

[jonwil]

But why go with PCI only for the first release?
AGP is what everyone wants.

I suspect it is because developing glue logic to talk to the PCI bus is easier than developing glue logic to talk to the AGP bus...

Still a market

[glasse]

This card, as it is stated in TFA, isn't meant to compete successfully in the category of "renders Doom III at appreciable speeds". It's not the result of years of R&D, and it's not bleeding-edge. It's a chance for a company to service the niche market of people who want a decent-quality card, and are willing to pay for the freeness of it. Making a product that kicks the ass of the competitors in the category of how Free-Software-compatible it is, is the idea, and I think Techsource can compete successfully with ATI and nVidia here.

Ethan

Re:But....

[DominoTree]

Can someone tell me how FP is redundant?

It seems that the only productive answer...

[leonbrooks]

...is to provide something which encourages them to Open their specs by terrifying them even more with the consequences if they don't. A card which is fully programmable might do just that.

Not necessarily in its first incarnation, but maybe in round 2 or 3, if it gets 1% of the available market either by providing programmability features which the others can't - that get used - or by leaving expensive but seldom-critical parts off to make a cheaper chipset which gets picked up big time by ASUS or some other motherboard maker.

Remember that saving $5 on a motherboard might represent 10-20% of the manufacturing cost. If the chipset has 95% of the performance for $2-$3 less and needs no heatsink or fan where a competitor's chips do, that might just do it. Or using 30% less watts to do pretty much the same job as a chipset from the big boys might make it attractive to thin-client and laptop makers.

One programmability feature I would like to see is the ability to stack the hardware gadgets arbitrarily - e.g., if you're not going to be stretching the horsepower of the thing, plug in a second DAC and video output to run two screens (or 3, or 4) from the one card; another e.g., if you are going to stretch the hardware, perhaps you can stack 3D engines so one is precalculating the next row (or band of rows) while the other's displaying the current row (or band).

Re:It seems that the only productive answer...

[Lisandro]

You know, i think the same. I had high hopes for the S3 DeltaChrome GPU line, which is DX9+ capable and has some very interesting features, like being able to process video in realtime using GPU horsepower and HDTV support. All of this at a very friendly price and using relatively simple cooling solutions, as S3 focuses on bundled / onboard solutions.

Yet, the reviews for it basically conclude it's a promising technology with poor drivers, which on GPU-land can make or kill a product. I still have hopes for them to release OSS drivers for it; over time, i think it could gain a good deal of acceptance among Linux/BSD users. Matrox did it, and sold a lot of cards to OSS users, but the 3D performance of their hardware was a bit lacking.

For some reason, i always liked S3 (even while bitching about those damn ViRGEs). Please, if someone at S3 is reading, we love you. Do it, and we'll root for the underdog!