To Flex or not to Flex... that is the question...

Hi PhysX Gurus,

Well I’m extremely impressed with the Flex demos (https://developer.nvidia.com/flex) I compiled for my box and am very impressed & intrigued by the Flex concept that separates it from PhysX: “that EVERYTHING is a system of particles connected by constraints”… way cool! :-)

The two-way coupling that Flex can bring between rigid bodies, soft bodies, clothing and fluid is a huge differentiating factor for the game I’m working on and am drooling at the possibilities…

Unfortunately the fly in the ointment is that thus far Flex only runs on recent NVidia hardware. This severely limits the percentage of people who can buy my game to about 20% of the market (excluding those with AMD cards or NVidia cards older than GeForce 7xx series, mainstream laptops, etc)

However in the 2+ year-old post at http://physxinfo.com/news/11860/introducing-nvidia-flex-unified-gpu-physx-solver/, one of the Flex developers mentioned a port to DirectCompute with a possible CPU implementation later on. Both of those are critically needed for Flex to have a chance to make it in general consumer games…

Q1: When do you think Flex will have a CPU implementation?

Q2: With the planned port of Flex from CUDA to DirectCompute, would Flex then run on AMD cards?

Q3: What changes will be made to Flex in PhysX 3.4?

Q4: When will PhysX 3.4 be released?

Thanks for taking the time! :)