Closes #155
Interestingly, while the Intel bug is being patched upstream, this PR works around it by interpolating between -1 and 1 instead of the opposite direction. I suspect the driver takes a different code path for this, since it's more straightforward, and that new code path isn't affected by the old bug.
Closes #155 Interestingly, while the Intel bug is being patched upstream, this PR works around it by interpolating between -1 and 1 instead of the opposite direction. I suspect the driver takes a different code path for this, since it's more straightforward, and that new code path isn't affected by the old bug.