Search Unity

Feature Request DrawMeshInstancedIndirect API requires an args ComputeBuffer per DrawCall O.o

Discussion in 'Universal Render Pipeline' started by Opeth001, Apr 12, 2021.

  1. Opeth001

    Opeth001

    Joined:
    Jan 28, 2017
    Posts:
    1,117
    Hello Everyone,

    I'm building a hybrid render package on top of the DrawMeshInstancedIndirect API while keeping the midrange 'OpenGL ES 3+' devices in mind, so I have a very limited number of ComputeBuffers. 'AFAIK is a minimum of 4'.

    the problem with this API is that it requires creating a ComputeBuffer args per DrawCall, which is a big waste.
    Is there an alternative / workaround to this?

    Thanks in advance!
     
    Opeth001, Apr 12, 2021
    #1

  2. aleksandrk

    aleksandrk

    Unity Technologies

    Joined:
    Jul 3, 2017
    Posts:
    3,019
    Hi!
    'A minimum of 4' is per draw call, not total. Normally there's more, that's just what is guaranteed by the specification. Note: compute is available on OpenGL ES 3.1+.
     
    aleksandrk, Apr 13, 2021
    #2
    Opeth001 likes this.

  3. Opeth001

    Opeth001

    Joined:
    Jan 28, 2017
    Posts:
    1,117
    Thank you for the reply!
    what do you mean by "A minimum of 4 per draw call".
    can you please give an example of use case where it is possible and where it is not.

    Thanks again!
     
    Opeth001, Apr 13, 2021
    #3

  4. aleksandrk

    aleksandrk

    Unity Technologies

    Joined:
    Jul 3, 2017
    Posts:
    3,019
    This means that a single draw call can use up to at least 4 compute buffers in total. A compute buffer that is used for DrawMeshInstancedIndirect counts towards that limit. So if you are on a device that supports up to 4 compute buffers per draw call, you can use the remaining three in your shader for rendering.
     
    aleksandrk, Apr 13, 2021
    #4
    Opeth001 likes this.

  5. Opeth001

    Opeth001

    Joined:
    Jan 28, 2017
    Posts:
    1,117
    Thanks!
     
    Opeth001, Apr 13, 2021
    #5