diff options
author | dan sinclair <dsinclair@google.com> | 2019-09-20 13:43:22 -0400 |
---|---|---|
committer | GitHub <noreply@github.com> | 2019-09-20 13:43:22 -0400 |
commit | ef6d2fa55db2c4d01c9af2db469cce1ec07119cc (patch) | |
tree | e937888862faf426e2aac64873a5c828e6ff9238 /docs | |
parent | 1c8cc31a5bd3a7b075b7bc1c64972d2cc7b4af64 (diff) | |
download | amber-ef6d2fa55db2c4d01c9af2db469cce1ec07119cc.tar.gz |
[AmberScript] Fix parsing of matrix data into buffers. (#657)
Currently parsing of matrix data in a buffer DATA block would trigger
memory corrpution issues. This was due to several things:
* Incorrect interpretation of mat2x4 for rows and columns
* Not clamping the amount of data read from the DATA block
* Handling of padding values.
With this CL we correctly handle mat2x4 as 2 columns and 4 rows. We do
not require padding values in the DATA block for matrices, it will be
inserted automatically as needed. We error out if the number of DATA
elements is not a multiple of the amount of input needed for the number
of elements.
Fixes #616
Diffstat (limited to 'docs')
-rw-r--r-- | docs/amber_script.md | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/docs/amber_script.md b/docs/amber_script.md index 3349270..3f8dfbc 100644 --- a/docs/amber_script.md +++ b/docs/amber_script.md @@ -117,7 +117,7 @@ either image buffers or, what the target API would refer to as a buffer. * `float` * `double` * vec[2,3,4]{type} - * mat[2,3,4]x[2,3,4]{type} + * mat[2,3,4]x[2,3,4]{type} (mat<columns>x<rows>) * Any of the `Image Formats` listed below. Sized arrays and structures are not currently representable. |