Official PhysX Thread

Self explaining , if you want to play your Harold advanture with hairwork or Batman with the fancy cape and smoke without any frame drops what so ever on your AMD GPU.



Personal i think it got to do with the dx12/vulkan open agreement to multi gpu, just wild speculating here . And Micro$ofr and Khronos got the call after all from a develper point off view. AMD bin open source all the time from what I'm aware . Nvidia well .. yah well $

P.S. Who will get the little,tiny itsipitsi hint, i put in there ;)
 
Last edited:
konjini said:
Self explaining , if you want to play your Harold advanture with hairwork or Batman with the fancy cape and smoke without any frame drops what so ever on your AMD GPU.



Personal i think it got to do with the dx12/vulkan open agreement to multi gpu, just wild speculating here . And Micro$ofr and Khronos got the call after all from a develper point off view. AMD bin open source all the time from what I'm aware . Nvidia well .. yah well $

P.S. Who will get the little,tiny itsipitsi hint, i put in there ;)
Click to expand...
"Self explaining"

How about drop the cloak and dagger bullshit and tell us what you're on about? :confused:
 
SIrPauly said:
It's good to see Zogrim's site: Physxinfo still going. Nvidia still spending resources on PhysX and Flex.
Click to expand...

I honestly thought you died because you stopped posting other forums as well, and your kin wasn't aware of your online presence to even inform us.
 
konjini said:
Self explaining , if you want to play your Harold advanture with hairwork or Batman with the fancy cape and smoke without any frame drops what so ever on your AMD GPU.



Personal i think it got to do with the dx12/vulkan open agreement to multi gpu, just wild speculating here . And Micro$ofr and Khronos got the call after all from a develper point off view. AMD bin open source all the time from what I'm aware . Nvidia well .. yah well $

P.S. Who will get the little,tiny itsipitsi hint, i put in there ;)
Click to expand...

It is interesting the sorta options it's opening up. It also is enabling Freesync to work on nVidia GPU's and G-Sync to work on AMD GPU's.
 
Thanks for the feedback, Pauly started this thread and he contributes to it so I'll wait to hear from him.
 
Physx is still utilized even to now and integrated into most game engines. It's automatically installed with every new Nvidia driver, unlike stereoscopic 3d which is no longer included or supported in new drivers. But PhysX is so common place and integrated you don't need to really reference "physx" like it's something new or 3rd party anymore. Most game engines that use Physx as part of their physics engine don't even advertise it as "physx" anymore like you used to see in game logos. When physx code is run on AMD/Intel it runs on the CPU, with Nvidia you can choose CPU or GPU, but I don't think GPU makes much of a difference, if any these days.
 
Last edited:

NVIDIA Support

nvidia.custhelp.com nvidia.custhelp.com

Support plan for 32-bit CUDA​


Updated 01/17/2025 01:22 AM
Support plan for 32-bit CUDA


32-bit compilation native and cross-compilation were removed from CUDA 12.0 and later Toolkit. 32-bit CUDA applications cannot be developed or debugged using CUDA 12.0 or later toolkit for any target architecture. Use the CUDA Toolkit from earlier releases for 32-bit compilation

CUDA Driver will continue to support running 32-bit application binaries on GeForce RTX 40 (Ada), GeForce RTX 30 series (Ampere), GeForce RTX 20/GTX 16 series (Turing), GeForce GTX 10 series (Pascal) and GeForce GTX 9 series (Maxwell) GPUs. CUDA Driver will not support 32-bit CUDA applications on GeForce RTX 50 series (Blackwell) and newer architectures.

Support for running x86 32-bit applications on x86_64 Windows is limited to use with:
  • CUDA Driver
  • CUDA Runtime (cudart)
  • CUDA Math Library (math.h)
Click to expand...

Good bye H/W Physic Acceleration starting with the Blackwell 50 series.
 
Probably not a big deal as the various graphic engines probably already have their own libraries to deal with it on systems that don't have the feature, and it frees up chip space for more general purpose/rendering?
 
koralis said:
Probably not a big deal as the various graphic engines probably already have their own libraries to deal with it on systems that don't have the feature, and it frees up chip space for more general purpose/rendering?
Click to expand...

Its not a big deal. Just sux for playing older games as it will start to become a pain in the ass to access the feature. Here's hoping something like ZLUDA, or a fork, can add 32-bit CUDA support to preserve access to the feature.

*edit* Or you can always go the some what janky route of setting up an older GeForce as a dedicated PhysX accelerator :D
 
Last edited:
You must log in or register to reply here.
Back
Top