[GUI] [mac] Support fast_gui on macOS #1981
Conversation
Codecov Report
@@ Coverage Diff @@
## master #1981 +/- ##
==========================================
- Coverage 43.43% 43.41% -0.03%
==========================================
Files 45 45
Lines 6264 6268 +4
Branches 1108 1109 +1
==========================================
Hits 2721 2721
- Misses 3369 3373 +4
Partials 174 174
Continue to review full report at Codecov.
|
python/taichi/lang/meta.py
Outdated
| @@ -16,13 +16,24 @@ def tensor_to_ext_arr(tensor: ti.template(), arr: ti.ext_arr()): | |||
|
|
|||
|
|
|||
| @ti.kernel | |||
| def vector_to_fast_image(img: ti.template(), out: ti.ext_arr()): | |||
| def vector_to_fast_image(img: ti.template(), small_endian: ti.template(), | |||
| out: ti.ext_arr()): | |||
| # FIXME: Why is ``for i, j in img:`` slower than: | |||
| for i, j in ti.ndrange(*img.shape): | |||
| u, v, w = min(255, max(0, int(img[i, img.shape[1] - 1 - j] * 255))) | |||
There was a problem hiding this comment.
Possible my fault, use r, g, b instead of u, v, w for clarity?
python/taichi/lang/meta.py
Outdated
| # We use i32 for |out| since OpenGL and Metal doesn't support u8 types | ||
| # TODO: treat Cocoa and Big-endian machines, with XOR logic | ||
| out[j * img.shape[0] + i] = w + (v << 8) + (u << 16) | ||
| if ti.static(small_endian): |
There was a problem hiding this comment.
| if ti.static(small_endian): | |
| if ti.static(ti.get_os_name() != 'osx'): |
instead of passing a template argument?
Btw, can you confirm if all OS X machines are big-endian (and also all Win and Linux X11 are little-endian)?
There was a problem hiding this comment.
Fixed.
can you confirm if all OS X machines are big-endian
Given that the non-fast way is using big-endian and we don't hear users complain, I think this is safe. Let's just roll it out and see what happens, shall we?
Sorry for the delay! Fixes #1976
Related issue = #1976
[Click here for the format server]