You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The time it takes to transmit data to GPU is non negligible.
Right now what we do is
mmap the file to RAM
create RAM buffers for vertices & colors & the indices of the elements that are worth rendering
transfer these buffers to GPU memory
draw
repeat
Instead we could
load the file into a texture (mmap on GPU?)
create vertices & their colors with shaders (tricky part*)
draw
repeat
This would make the host code much simpler.
Thus porting to Web or Android would be much simpler.
*The tricky part is that to do something like Algo3DCubeFrebet with shaders, one needs to somehow:
read multiple bytes from the file-texture
write a color to the output texture but at different coordinates
I believe that (x1,y1) --> (x2,y2,z) transformation is hard for a GPU.
Maybe compute shaders can help here?
The time it takes to transmit data to GPU is non negligible.
Right now what we do is
Instead we could
This would make the host code much simpler.
Thus porting to Web or Android would be much simpler.
*The tricky part is that to do something like
Algo3DCubeFrebet
with shaders, one needs to somehow:I believe that (x1,y1) --> (x2,y2,z) transformation is hard for a GPU.
Maybe compute shaders can help here?
@icehess Ideas?
The text was updated successfully, but these errors were encountered: