GPUs with PhysX support
From PhysX Wiki
|
This page is incomplete. You can help PhysX Wiki project by filling it up. |
|
This page was composed by competent person. Even if you can't find source of the information, it is trustworthy. |
PhysX support or hardware PhysX acceleration feature is the ability of NVIDIA GeForce GPUs to perform certain types of physics simulation (based on NVIDIA PhysX SDK physics engine) in hardware, faster than conventional CPUs.
Minimum requirements are:[1]
- CUDA architecture (GeForce 8xxx series and above)
- Minimum 32 CUDA cores
- Minimum 256 Mb of onboard memory (512 Mb recommended for single GPUs)
Contents |
NVIDIA GPUs supporting PhysX accelleration
We added a rough and approximate rating system, to indicate PhysX computing capabilities of GPUs in several types of configurations.
Performance rating is provided with reference to GPU PhysX games, released in 2011.
| Lowest | 1 | 2 | 3 | 4 | 5 | Highest |
In general: the amount of CUDA cores and Processor/GPU clocks matters (the more the better), on-board memory on the other hand; not so much (as long as it fits minimum requirements).
GeForce GPUs
|
|
|
|
|
|
* Only available for OEM
** Dual-cards can use only one GPU for physics calculation, if utilized as dedicated PhysX cards.
GeForce for Notebooks
Tesla/Quadro Professional Graphics
Technically, GPU PhysX is supported by CUDA-capable Tesla/Quadro graphics cards, however one will have to use the WDDM GeForce driver, not the TCC (Tesla Compute Cluster) driver, to get the GPU physics acceleration.
Additional signs
In addition, GeForce GPUs featuring PhysX support can be recognized by several variations of signs on their packages, for example:
 |  |
See also
References
External links
- NVIDIA GeForce GPUs supporting NVIDIA PhysX Technology - official list from NVIDIA
| |||||||||||||||||
