Metal fix #6413
Metal fix #6413
Conversation
tools/gfx/metal/metal-device.cpp
Outdated
| @@ -404,7 +404,7 @@ Result DeviceImpl::createTextureResource( | |||
| break; | |||
| } | |||
|
|
|||
| bool isArray = desc.arraySize > 0; | |||
| bool isArray = desc.arraySize > 1; | |||
There was a problem hiding this comment.
This was discussed when it was initially submitted.
Please see the discussion:
#4331 (comment)
There was a problem hiding this comment.
@skallweitNV for an opinion.
There was a problem hiding this comment.
I wonder why this is needed? What issue is this fixing?
Are there no 1-element texture arrays in metal?
There was a problem hiding this comment.
I fix this issue on SGL side. So no fix needed here.
There was a problem hiding this comment.
The issue I'm trying to fix is that SGL presume that desc.arraySize == 1 is a non-array texture, however, slang-gfx assume this is a texture array with array size of 1. Those assumption conflicts.
|
In the new commit, I fix an obvious bug in metal. |
FYI @skallweitNV the same issue is also in slang-rhi. |
In newTextureView, levelRange should represent the mipmap level range, while sliceRange should represent the texture layer range for texture array. But the implement inverse those two wrongly.
kaizhangNV
force-pushed
the
metal_fix
branch
from
8db65b6 to
1e22f53
Compare
Fix the issue in creating texture view in metal.
In newTextureView, levelRange should represent the mipmap level range,
while sliceRange should represent the texture layer range for texture
array. But the implementation inverses those two wrongly.
Fix the issue when emitting read only texture_buffer<>
texture_buffer doesn't support access::sample, we should emit access::read when it's read only buffer.