And CUDA....
wheres ati's version of cuda ? /fail
There are more games to worry about than software that makes use of CUDA & the most common being Folding@home which many don't bother with.
The GPU needs Multi threaded sequential programmed software which is harder & which brings us back to the lack of multi threads apps in general even for multi core CPU which is easer to do anyway & we still don't have many.
So unless there is a CUDA apps that is important to a user then CUDA is of little use right now as there is virtually no software for it.
ATI do have there OpenCL 1.0 (Open Computing Language) version in the works plus Hydra Engine.
http://forums.overclockers.co.uk/showthread.php?t=17958083
With NV there is no Multi GPU with out a sli mobo that may change soon with CFireX/SLI mobo NV buckling dew to CF on AMD/Intel mobos.
With NV you need to have 2 or more cards from the exact model for SLI or the other will of only be any use for PhysX or CUDA & not help out in SLI.
With ATI & Intel using Hydra Engine with the option to not need to be the same model or even the same family all will help out in MultiGPU gaming all will help out in OpenCL 1.0 (Open Computing Language)& physics is in DX11 & will run on DX11 cards in the mix both of which are open for anyone to use with out royalty & are not owned or controlled by any gfx company unlike NVs CUDA & PhysX.
http://www.khronos.org/news/press/releases/the_khronos_group_releases_opencl_1.0_specification/OpenCL 1.0 (Open Computing Language) is an open, royality free standard that can be used in multiple platforms which allows for parallel programing of modern processors. It was created by Khronos Group (and others) but the idea purposed for OpenCL was by Apple several months ago. It was developed and ratified by 3DLABS, Activision Blizzard, AMD, Apple, ARM, Barco, Broadcom, Codeplay, Electronic Arts, Ericsson, Freescale, HI, IBM, Intel Corporation, Imagination Technologies, Kestrel Institute, Motorola, Movidia, Nokia, NVIDIA, QNX, RapidMind, Samsung, Seaweed, TAKUMI, Texas Instruments and Umea University thus creating OpenCL 1.0.
In a nutshell OpenCL allows developers to take full advantage of a diversity of multi-core processors (GPU, CPU, Cell, DSPs, etc) through an API that coordinates parallel computing and a programming language that specify those computations.
Problem is that NV want to push Proprietary CUDA so OpenCL is competing, so as to be on the safe side Nvidia will use CUDA to run OpenCL just incase OpenCL turns out to be the more popular.
http://www.nvidia.com/object/io_1228825271885.html
I could say Where's NV version of Dx10.1 ? /fail, but that is just as unimportant as CUDA at this present time for most.
Last edited: