- TODO: Optimize this, we can avoid the double bswap in FIFO and then in attribute push
Not very important since nobody is doing register push in high-performance path.
- Mainly affects nvidia where x/w * w can sometimes return a value smaller than x.
In such conditions, floor(x) will return x-1 if x is an integer which is horribly wrong and exaggerates minor precision drift to great proportions.