As the title says
http://www.bit-tech.net/news/hardwar...idia-for-amd/1
This is highly amusing as physicsX is nvidia's biggest selling point
As the title says
http://www.bit-tech.net/news/hardwar...idia-for-amd/1
This is highly amusing as physicsX is nvidia's biggest selling point
That's interesting, other press releases that I read on that claimed he was hired by AMD to work on Fusion (I know AMD and ATI are essentially the same company, but I would think there would be more crowing by ATI if they'd actually got him in to work on a rival to Physx).
ATi fully support OpenCL (an open source physics on gpu engine) it's just that it has no support or backing atm.
OpenCL did look like it was going to be a big thing, however the failier of OpenGL3 and then nVidia buying up PhysX has basically left it supportless and dead in the water atm.
the last thing I want to see is ATi trying to push their own propriety physics engine.
[rem IMG]https://i69.photobucket.com/albums/i45/pob_aka_robg/Spork/project_spork.jpg[rem /IMG] [rem IMG]https://i69.photobucket.com/albums/i45/pob_aka_robg/dichotomy/dichotomy_footer_zps1c040519.jpg[rem /IMG]
Pob's new mod, Soviet Pob Propaganda style Laptop.
"Are you suggesting that I can't punch an entire dimension into submission?" - Flying squirrel - The Red Panda Adventures
Sorry photobucket links broken
Actually opencl is a general use computing api, so a better parallel would be CUDA. Ati were supposedly porting the bullet physics engine to opencl, but I haven't heard anything for a while now.
My money is on a physics engine on Microsoft's directcompute rather than open source on opencl, but either way I think it'll probably be a while before we see an alternative to physx actually being used in games for gpu based physics.
It's a shame havok's effort seems to have stalled but I suspect that's been influenced by Intel's scaling back of it's discrete gpu project.
I wouldn't say PhysX / 3D are their biggest selling points, but they are definitely the technologies nVidia tout as their Unique selling points. Although frankly this move won't make any difference at all nVidia's marketing of PhysX as it's a mature proprietary technology, and unless ATI employ a new head of Developer Relations, developing a rival physics engine will be pretty immaterial, because ATIs effort simply won't make it into enough games...
I highly doubt AMD would hire someone versed in Physics acceleration to work on silicon engineering problems.
It's far more likely that he'll be tasked with developing a standard physics processing API, or porting an existing API to OpenCL. That way AMD can score kudos with developers by saying; "Hey look! we care about you and your customers, we've got an API you can influence the development of, as well as use as you see fit, on both AMD and nVidia cards, aren't we awesome?!".
Lets face it, the nVidia kool-aid barrel is getting a bit thin these days, and it wouldn't be hard for AMD to gain some trump points.
I agree absolutely, I just don't think it'll do them any good unless they massively improve their developer relations team. Even if it is open and cross-hardware compatible, it'll still be seen as a rival to PhysX, and nVidia, with their already massively superior dev relations, will simply put more resources into pushing PhysX. So (IMNSHO, obviously) it will take better ATI dev rels to make this a significant move.
I wish my drivers worked. It's awful having to work out where to type when the entire screen is black.
Last edited by this_is_gav; 02-06-2010 at 05:31 PM.
What do you use cuda for? (I meant to ask this from Matty)
ehhhhhhh (02-06-2010)
There are currently 1 users browsing this thread. (0 members and 1 guests)