28 Apr
2023
28 Apr
'23
2:05 p.m.
On Fri Apr 28 11:14:47 2023 +0000, Giovanni Mascellani wrote:
I would have hoped that 8 ULPs were enough for having a common idea of the first derivatives of a quadratic polynomial, but it seems video card producers decided differently. I don't think there is much we can do about, except for raising our tolerance (which is the number at the end of the `probe` command).
Messed with the tolerance values a bit, all of the tests are actually okay at 8 except for the last one, which could go no lower than _40_. That seems... really high for one pixel? If it isn't, I can push this test with that one tolerance value set higher than the others.