Added Matrix Multiplication #3
Conversation
examples/CMakeLists.txt
Outdated
| @@ -2,6 +2,7 @@ set(GPGPU_GLES_EXAMPLES | |||
| array_add_float | |||
| fir_conv_float | |||
| array_add_fixed16 | |||
| mult_mat_int | |||
There was a problem hiding this comment.
why not floats? ints make little sense here
There was a problem hiding this comment.
We have changed the entire operation on ints to that on floats
There was a problem hiding this comment.
so change the name to float please :)
shaders/mult_mat_int.fs
Outdated
| float i = vTexCoord.s; | ||
| float j = vTexCoord.t; | ||
| float result = 0.0; | ||
| for (float k = 0.0; k < 4.0; ++k) |
There was a problem hiding this comment.
To reduce the value of k to lie between 0 to 1, we are dividing it by 4. This is to make sure that the coordinates of the texture element we are multiplying lies in the range 0 to 1 in the texture coordinate system.
Foe texture2D function, the second parameter is vec2 which contains floats as its elements. This is the reason for k being a floating point value.
There was a problem hiding this comment.
Hi @JDuchniewicz any update on merging this inclusion of matrix multiplication to the upstream!?
There was a problem hiding this comment.
Hi, sorry got bogged down by other work. I still do not understand why 4.0 and if 4.0 is necessary (I presume the matrix size?). If this is indeed the matrix size, it should be configurable as a uniform and not hard-coded. Please change that.
There was a problem hiding this comment.
Hi Jakub, yes its for size.
So, we are testing for a generalized notation and will update soon, can we use something like the const kspan for indicating this 4.0 as its tested for this size.
In this PR, we have added support for Matrix Multiplication and made the required changes in the relevant directories.
We have obtained the correct results and have attached the screenshot for the same in the description.