Reasoning for closed source

[Gizmot]

Hi guys,

just wondering if you guys could enlighten me on the logic of closed source drivers (also includes BIOS btw).

I have been on pondering this issue for years and still havnt found a good answer. It is more than possibly based on my lack on deep understanding on drivers coding.

The way I see it, as an hardware manufacturer you need to pay for the actual hardware and also to write drivers for said hardware. Then, you need to write a different driver for each OS, for every new model and feature, keep track of bug reports, etc
At the same time, you know that there is a bunch of very talented FOSS guys bashing their head on how to reverse engineer some hardware.
So then, in God's name, dont they just open their close source drivers, pay some developpers (and get a crap load of free work from FOSS developpers) and developp a generic multiplatform driver that would be easy to port on easy platform?
I'm sure some people would say IP issues, but then, thats you own hardware... Not giving the competition some usefull (?) info about your product? Call me an ignorant mech.eng, but I would assume that the biggest challenge in hardware design is the actual manufacturing...

anyway, fire away your opinion!

thanks

[deanjo]

Hi guys,

just wondering if you guys could enlighten me on the logic of closed source drivers (also includes BIOS btw).

I have been on pondering this issue for years and still havnt found a good answer. It is more than possibly based on my lack on deep understanding on drivers coding.

The way I see it, as an hardware manufacturer you need to pay for the actual hardware and also to write drivers for said hardware. Then, you need to write a different driver for each OS, for every new model and feature, keep track of bug reports, etc
At the same time, you know that there is a bunch of very talented FOSS guys bashing their head on how to reverse engineer some hardware.
So then, in God's name, dont they just open their close source drivers, pay some developpers (and get a crap load of free work from FOSS developpers) and developp a generic multiplatform driver that would be easy to port on easy platform?
I'm sure some people would say IP issues, but then, thats you own hardware... Not giving the competition some usefull (?) info about your product? Call me an ignorant mech.eng, but I would assume that the biggest challenge in hardware design is the actual manufacturing...

anyway, fire away your opinion!

thanks

Simple short reason, competitive IP, not every company wants to give away their trade secrets that give their product an advantage over their competitors. Better performance over their competitors hopefully results in better sales which in turn results in better ROE for their investors. Like it or not the industry is driven by a monetary system where the investors want maximum bang for their investment.

[deanjo]

One other thing, I would have to disagree on the hardest thing being the actual manufacturing. That is actually one of the simplest parts, R&D and marketing are far more difficult as they ultimately decide if a product is successful or not.

[deanjo]

and the Gurus answers: "this shitty windows-driver? no EXA no KMS crappy ugly sourcecode holy crappy mother of the BUG no! no ! no no! "

You know you really can't blame windows drivers for limitations of X. Nor do I believe that you have any clue about what the windows source looks like so your comment is based on pure speculation. Face it X is a really old solution with a lot of old world code that was put in as a after thought for *nix. If this was not the case efforts like Wayland would not exist.

[Gizmot]

Simple short reason, competitive IP, not every company wants to give away their trade secrets that give their product an advantage over their competitors. Better performance over their competitors hopefully results in better sales which in turn results in better ROE for their investors. Like it or not the industry is driven by a monetary system where the investors want maximum bang for their investment.

I understood that...
but, for exemple, assume that amd/ati was to open up their source code tomorrow. Would Nvidia ''jump'' over the source code to find all the ''tricks'' hardware wise? Is there actually anything of value for the hardware developpement?

[Gizmot]

One other thing, I would have to disagree on the hardest thing being the actual manufacturing. That is actually one of the simplest parts, R&D and marketing are far more difficult as they ultimately decide if a product is successful or not.

I do agree,
except that I was refering to the R&D of manufacturing.

[deanjo]

I understood that...
but, for exemple, assume that amd/ati was to open up their source code tomorrow. Would Nvidia ''jump'' over the source code to find all the ''tricks'' hardware wise? Is there actually anything of value for the hardware developpement?

Maybe they would, and maybe they would not, but the last thing investors want is uncertainty.

[deanjo]

i blame fglrx only because they try to run an windows driver on top of linux/X...

It's a end to a mean my friend. If you cannot do something natively you have to pull tricks. Remember that no generic solution can beat a optimized one.

[Gizmot]

Maybe they would, and maybe they would not, but the last thing investors want is uncertainty.

thats the 10000$ question I'm asking.

Because on one side, you have uncertainty, and on the other you have the certitude to obtain the cooperation and loyalty of all the FOSS community (with also a much lower cost for drivers developpement...)

[deanjo]

I do agree,
except that I was refering to the R&D of manufacturing.

Well lets put it this way, I design boats as well as my computing life, manufacturing wise that is simple as it takes typically less then 4 hours to assemble a boat but to gain that extra 5 mph out of the hull it took 18 months of analysis, design and simulations to find that "sweet spot spec" that allowed that performance difference. Assembly wise it made no difference. I guess it all depends on the product being designed. Ask any farmer as well about what they would do about their combine if they had a chance to talk to the engineer.