I've just noticed a surprising comment in D3D12_RESOURCE_FLAGS, for D3D12_RESOURCE_FLAG_DENY_SHADER_RESOURCE, which says -
Some adapter architectures experience increased bandwidth for depth stencil textures when shader resource views are precluded.
Reading it several times over, I couldn't help but interpret it in a way that's contrary to my expectations that, in pretty much all cases, more restrictive/strict resource use only ever offers optimisation possibilities. I think this quoted sentence is unfortunately written in such a way that can be interpreted equally either way; it's just as easy to interpret this as saying -
- "some architectures suffer from increased bandwidth usage as a result of allowing shader resource views" or to put it another way, "some architectures gain bandwidth capacity as a result of precluding shader resource views"
- versus, "some architectures suffer from increased bandwidth usage as a result of precluding shader resource views" / "some architectures gain bandwidth capacity as a result of allowing shader resource views".
These two completely opposite interpretations could easily fit in with the quoted comment, which I think comes down to the ambiguous interpretation of "experience increased bandwidth".
I've just noticed a surprising comment in D3D12_RESOURCE_FLAGS, for
D3D12_RESOURCE_FLAG_DENY_SHADER_RESOURCE, which says -Reading it several times over, I couldn't help but interpret it in a way that's contrary to my expectations that, in pretty much all cases, more restrictive/strict resource use only ever offers optimisation possibilities. I think this quoted sentence is unfortunately written in such a way that can be interpreted equally either way; it's just as easy to interpret this as saying -
These two completely opposite interpretations could easily fit in with the quoted comment, which I think comes down to the ambiguous interpretation of "experience increased bandwidth".