b/21187932
Dependency from one kernel to another kernel -- where one reads the
other's output to a global allocation -- was not handled correctly.
This CL fixed that. Needed to make a new CTS test pass.
Change-Id: I500f9cac9d89bcaec7c186e942ba4a7d413daadb
Adds in flag CREATE_FLAG_WAIT_FOR_ATTACH with value 0x0008 to the RenderScript API
to be passed down to the C++ implementation to tell it to wait for a debugger to be
attached before executing the kernel.
Change-Id: Ibc4e903efbed1b1fb14cf378d1a8517d5c8d1f26
Signed-off-by: Stephen McGroarty <stephen@codeplay.com>
bug 20822382
This corrects a copy&paste error placing F16 into F32.
Fixes an error disallowing vectors of fp16 types.
Change-Id: I34be1aa9a323d61a7121581ab9599d2c2d32dd42
bug 20822382
This corrects a copy&paste error placing F16 into F32.
Fixes an error disallowing vectors of fp16 types.
Change-Id: I34be1aa9a323d61a7121581ab9599d2c2d32dd42
b/20728113
so that we can share them with support lib.
Change-Id: I231c0d20c41ee53342d8a975c2db5a31854b9d40
(cherry picked from commit 7b2a46fee0f95b9ae2485db36e0d279730623c4a)
- Typo for validateUplo
- Typo in ZHEMM, element should be FLOAT64_2.
- For GEMM and SYMM, SYRK, 'CONJ_TRANSPOSE' should also be handled in the
validation process.
- For SYMM, check matrix A is symmetric.
- For HERK, the dimension validation was switched for Transpose case.
Also, only Conj Trans is allowed in this case.
- FOR SYR2K, fix the dimension check for Matrix C.
- For TRMM & TRSM, fix the validation part for dimension check.
Change-Id: I559b5c5695aa82604de2955ae2327b694236d3ed
Cherry-pick fix from AOSP. Error check for kernel launch was
generating a false positive.
bug 20690242
Change-Id: Ic4c6644072a11aab9a273070be5734519136f685
b/20728113
In case the requested size for memory allocation overflows, or memory
allocation fails.
Change-Id: I8dac132dd4d0210938660ffbb82cbe44000d2a90
(cherry picked from commit 4e90b9b57cc96964a9d5c1845172a72cb51feafb)
b/20728113
In case the requested size for memory allocation overflows, or memory
allocation fails.
Change-Id: I8dac132dd4d0210938660ffbb82cbe44000d2a90
Stop assuming that a Java Bitmap has a SkBitmap* that
has some externally managed lifecycle, and instead switch
a bunch of users to accessing the bitmap by providing
their own SkBitmap* on which to set the (ref counted!)
SkPixelRef* instead
Attempt #2 to land this, original issue was in getSkBitmap
and should be fixed
Change-Id: I0fd9e193968b41e5597784140d56b4885906864a
- Typo for validateUplo
- Typo in ZHEMM, element should be FLOAT64_2.
- For GEMM and SYMM, SYRK, 'CONJ_TRANSPOSE' should also be handled in the
validation process.
- For SYMM, check matrix A is symmetric.
- For HERK, the dimension validation was switched for Transpose case.
Also, only Conj Trans is allowed in this case.
- FOR SYR2K, fix the dimension check for Matrix C.
- For TRMM & TRSM, fix the validation part for dimension check.
Change-Id: I559b5c5695aa82604de2955ae2327b694236d3ed
