Marek Does Major Rework Of Gallium3D Blitting

[russofris]

Sharing code is always the "easy" solution, its rarely the most efficient solution however. Thats why i've always been a little leery of Gallium3D. I know that NOW, R600g has surpassed what R600c had in performance in features, but I wonder what levels of performance R600c could have hit, writing AMD-specific code instead of blanket, vague, sharable code. If the Nouveau is working on their own solution that they believe is better for Nvidia cards, then I say let them do it. Either it'll work out fine, and everyone wins, or it wont, and they'll probably adopt R600g's style if possible.

In addition, sometimes competition between two different implementations will foster future innovation. Perhaps one of the two implementations will turn out to be better than the other from an architectural or performance standpoint. Either way, we should all be happy now that we have working implementations. Let's allow some time and vetting to determine which implementation is best.

F

[Ericg]

I know that sometimes the differences in the underlying hardware architecture are so vast that it calls for a different implementation, but anyone else thinks that the graphics driver developers should make more of an effort to share code? The Gallium3D architecture is in itself a big leap in the right direction, but it is not a magical bullet, I have the impression that they simply don't have any active line of communication between projects to discuss possible sharing of code/effort and end up reinventing the wheel over and over again.

PS: Marek is just amazing, thanks for all the awesome work.

Sharing code is always the "easy" solution, its rarely the most efficient solution however. Thats why i've always been a little leery of Gallium3D. I know that NOW, R600g has surpassed what R600c had in performance in features, but I wonder what levels of performance R600c could have hit, writing AMD-specific code instead of blanket, vague, sharable code. If the Nouveau is working on their own solution that they believe is better for Nvidia cards, then I say let them do it. Either it'll work out fine, and everyone wins, or it wont, and they'll probably adopt R600g's style if possible.

[BadgerRush]

Shouldn't they share more code?

Nouveau is currently working on its own implementation.

I know that sometimes the differences in the underlying hardware architecture are so vast that it calls for a different implementation, but anyone else thinks that the graphics driver developers should make more of an effort to share code? The Gallium3D architecture is in itself a big leap in the right direction, but it is not a magical bullet, I have the impression that they simply don't have any active line of communication between projects to discuss possible sharing of code/effort and end up reinventing the wheel over and over again.

PS: Marek is just amazing, thanks for all the awesome work.

[curaga]

I hope he has a nice early Christmas because of the donations and he doesn't need to depend on Bridgman's strained Linux budget

I don't think he's being paid by AMD...

[Sidicas]

$100 USD sent

Bah, Paypal reversed it some hours later.. They want me to send them some photo ID and such, even after I already gave them my bank account# and routing number.. I'll have to send it again later.

[gwelter]

I just registered to say that I paypalled him ?20 to thank him for his work. I hope he has a nice early Christmas because of the donations and he doesn't need to depend on Bridgman's strained Linux budget

[Sidicas]

See his forum profile for the paypal link.

$100 USD sent

[bornslippy]

Radeon continues to move forward, well done Marek. Performance is just about good enough now, we just need better power management...

[dogsleg]

Wow! Another hit!

Marek rocks! Thank you!

[uid313]

Superhero

Marek is a radeon superhero!

He keeps impressing!