NVIDIA RTX IO #629
robertosfield
started this conversation in
General
NVIDIA RTX IO
#629
Replies: 2 comments
-
|
Hi Robert,
sounds interesting. Also of interest are unbound buffers. Does vsg support this?
Werner
…On 5. Dezember 2022 17:18:38 MEZ, Robert Osfield ***@***.***> wrote:
NVidia have announced RTX IO for transferring data to GPU memory and
uncompressing on the GPU, seems like an interesting development for
anyone who wants to rapidly stream data to the GPU.
https://developer.nvidia.com/rtx-io
Requires two new extensions
[VK_NV_memory_decompression](https://registry.khronos.org/vulkan/specs/1.3-extensions/html/vkspec.html#memory-decompression)
and VK_NV_copy_memory_indirect, there are part of
[Vulkan1-.3.236](https://registry.khronos.org/vulkan/specs/1.3-extensions/html/vkspec.html).
--
Reply to this email directly or view it on GitHub:
#629
You are receiving this because you are subscribed to this thread.
Message ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
0 replies
-
|
On Mon, 5 Dec 2022 at 17:03, Werner Modenbach ***@***.***> wrote:
sounds interesting. Also of interest are unbound buffers. Does vsg support
this?
I am not familiar with unbounded buffers. Could you provide a link to
specs/articles discussing it?
… Message ID:
***@***.***
com>
|
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
NVidia have announced RTX IO for transferring data to GPU memory and uncompressing on the GPU, seems like an interesting development for anyone who wants to rapidly stream data to the GPU.
https://developer.nvidia.com/rtx-io
Requires two new extensions VK_NV_memory_decompression and VK_NV_copy_memory_indirect, which are part of Vulkan1-.3.236.
I haven't looked into the details of the format supported or how portable it'll be, what fallback we could provide in case extension support is not available. The above article mentions being open source but haven't spotted a link to any published source yet.
One of the possibilities with being able to uncompress general data on the GPU, not mentioned in the above article is storing data compressed on the GPU and just uncompressing when required. For instance could one handle a really large CAD or Lidar dataset by having it all resident on the GPU and just uncompressing the parts that are relevant to the current view?
Beta Was this translation helpful? Give feedback.
All reactions