1. Avoid the need for multiple-of-four input textures. If you don't pass
a multiple of four then we'll do our best to pad the image.
2. Fix a bunch of bugs where we didn't accurately report why we couldn't
compress a texture and just crashed instead.
3. Some code refactoring to make certain if statements more readable.
There was a bug where certain blocks were not being written due to
improper iterating over non power-of-two textures depending on the
thread that got split across a newline of blocks. This should now
be fixed.
The offending command line arguments looked something like:
CLTool/tc -q 0 -t 32 -j 32 /path/to/npot-texture.png
1. Split compression parameter generation and compression parameter
packing. This gives a good performance boost, since we don't pack every
single time we compress. The error is computed each time, and only the
best parameters are packed.
2. Allow the shape selection function to specify up to ten shapes to
try for compression. We were already doing this kind of hackily where
we allowed both a three and two partition shape. This makes it a little
cleaner and exposes it to the user.
Change RGBACluster to be a class that only really persists once per block.
When we switch shapes and do operations on them, then we really only need
to change which points in the block are accessed. We don't need to do this
very often, so just change the mask whenever we need it. This brings us back
closer to our original performance, but we're still not where we were when
we started refactoring.
We suffered another performance hit. This time it comes from the fact
that we're copying around a lot of data based on what partition we're
choosing. We can get rid of this a tad by only copying the data that we
need once and then using getters/setters that selectively pull from
an array based on our shape index.
Changed the RGBAEndpoints to use the vector/matrix classes in
FasTCBase. This caused a ~20ms performance hit on an 8-core machine
which is likely due to the compiler having difficulty compiling away
some procedure call overheads. Upon profiling, the biggest bottleneck
is still by far the QuantizedError function, so any and all further
optimization should be focused on that.
