A brand new video has come up showing a PhysX fluid similuation from NVIDIA.
Physx is a marketing gimmick.
how is it a gimmick? But it is marketing.. I hope AMD has success with OpenCl(Which is an open standard no matter card you use it works )If dev can use the true power of Modern Gpus Frames per second will explode, Graphics, Physics & particle effects will be on a whole new level! This is a gimmick! - http://www.youtube.com/watc... This is not! - http://www.youtube.com/watc...
sorry but I agree. Physx is a gimmick. Go back a few years pre dx11. Nvidia was hoping to cash out on multiplat games by adding physx to them. AMD started pushing dx11 and Nvidia wanted nothing to do with the technology. They were going as far as downlplaying it in hopes of cashing in on those physx mutliplats. AMD gained dx11 traction and Nvidia was forced to play to their tune. In short they wanted nothing to do with innovation. they pinned their hopes on extra features in multiplatform titles. Back then Nvidia had it all. Imo Physx is the reason AMD/ATI were able to make the huge comeback they did.
gotta love the power of gpu's. This is why I expect the next console generation to feature an extremely power gpu with very little cpu support, all coding for physics, ai, etc will be done via opencl, compute shaders, and cuda.
You don't know much about how computers work if you think that way. CPUs need to be powerful enough because one way or another, everything goes through the CPU. The instruction is fetched by the CPU. If the instruction pertains to a task that needs to be done by the GPU, the CPU still needs to decode the instruction, do math on it, re-encode the answer, send it to RAM, and send it to the GPU. The CPU needs to keep up with everything, otherwise it becomes a bottleneck and the computer user needs to lower the video settings of the game. CPU matters as much as the GPU, maybe more.
I'm sorry, what? risc, really? What's next, cpu's can only handle one instruction at a time? Maybe look into vetor processing or something. The whole point of opencl, cuda, and compute shaders is to provide the gpu with an api which can perform functions of a cpu.
There's no point in offloading the work to the GPU; you still need a CPU to retrieve the finished product and output. As far as it's been explained to me by a friend engineer, hyperthreaded CPU cores handle 2 threads at a time because to handle anymore wouldn't lead to more gains since opening more applications dumps the RAM and resets the core's functions. So the way to go is multi-core CPUs. This just means that the future as of right now is still powerful multi-core CPUs with discrete GPUs.
I want something like physx on ATI cards
You can play engines with PhysX with ATI cards. I played Alice Madness Returns last night and that game uses it.
Yes, many games use standard CPU Physx. But you can't turn on these: http://www.youtube.com/watc...
Theres no need for this. OpenCL can be used for parallel acceleration such as physics calculations and its a better alternative to the Nvidia only PhysX
Indeed it can be used. But no game HAS ever used OpenCL acceleration because no decent middleware exists. Developers usually choose betweeen Bullet, Havok and PhysX these days. Bullet and Havok are CPU only. PhysX is CPU with optional GPU-acceleration.
I thought that was gorgeous... im no expert but thats pretty cool its dynamic
Still wondering when games will really start being able to do this kind of simulation in realtime in a normal game, not a techdemo. Hardware apparently has a ways to go before processing will be wasted on making proper shadows, water, fire, smoke and other particle effects look real.
N4G is a community of gamers posting and discussing the latest game news. It’s part of NewsBoiler, a network of social news sites covering today’s pop culture.