v3:
Fix two typos.
Signed-off-by: Chuanbo Weng
---
docs/Beignet.mdwn | 1 +
docs/howto/video-motion-estimation-howto.mdwn | 79 +++
2 files changed, 80 insertions(+)
create mode 100644
If the CL device does not support this builtin kernel, the test returns
PASS.
Signed-off-by: Guo Yejun
---
utests/CMakeLists.txt | 1 +
.../builtin_kernel_block_motion_estimate_intel.cpp | 109 +
utests/utest_helper.hpp
v2:
1. Just upload the first vme_state.
2. Remove duplicated code in check_opt1_extension.
3. Check image format before cl_gpgpu_bind_image_for_vme.
4. Fix error of getting mv. Because we suppose this kernel run in SIMD16
mode, so dword 0 of grf 1 should be
__gen_ocl_region(8,vme_result.s0),
__gen_ocl_vme is used for hardware accelerated video motion estimation.
It gets payload values as parameters and uses MOV to pass these payload
values to VME SEND Message's payload grfs. The int8 return value is used
to store SEND Message writeback.
v2:
Remove unnecessary 5 parameters(src_grf*)
1. Add a new structure DBGInfo into context class, storing debug
information.
2. Pass debug information from llvm IR to Context, for subsequently
passing, if OCL_PROFILING is true.
Signed-off-by: Yannan Bai
Signed-off-by: Meng Lv
---
1. Add a structure DBGInfo in GenEncoder class, storing debug infomation
for subsequentlt passing to GenInsn.
2. Add a structure GenInsnDBGInfo in GenInstruction defination.
3. Add a vector in GenEncoder class, storing debug
information corresponding to vector.
4. Pass debug
1. Add a structure DBGInfo in Instruction class, storing debug infomation.
2. Pass debug information from context to GEN IR, if OCL_PROFILING is true.
Signed-off-by: Yannan Bai
Signed-off-by: Meng Lv
---
backend/src/ir/context.cpp | 1 +
Print line and column numbers with ASM, if OCL_PROFILING is true.
Signed-off-by: Yannan Bai
---
backend/src/backend/gen_context.cpp | 7 +++
1 file changed, 7 insertions(+)
diff --git a/backend/src/backend/gen_context.cpp
b/backend/src/backend/gen_context.cpp
1. Add a structure DBGInfo in Opaqueue class, storing debug infomation in
selection for subsequentlt passing to selection IR.
2. Add a structure DBGInfo in SelectionInstruction class, storing debug
infomation.
3. Pass debug information from GEN IR firstly to selection queue, then
When was this workaround done (in particular, is beignet 1.1.1
affected)? As the Khronos test suite is non-public, I can't test this
myself.
Debian have now announced an intention to remove LLVM 3.5
(https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=803643).
An alternative option (which I
Reading p+1 when p==end is an out of bounds read.
Signed-off-by: Rebecca Palmer
---
(Found by valgrind while investigating #90472; probably not the
actual cause of that crash, but still a bug.)
diff --git a/backend/src/llvm/llvm_printf_parser.cpp
Hi Rebecaa,
Nice catch! This is the root cause of a bug I was recently trying to fix, it
will cause printf randomly fail if *(p+1) is accidently is '%'.
So could you resend a patch with a start "GBE:." and some more description
about the bug fix.
Thanks
Xiuli Pan
-Original
We found that it's a miss understanding of a corner case in type conversion,
not a llvm bug.
In that case the cl kernel try to convert an overflowed float value to ulong
e.g.
ulong dst = (ulong)1.8446744073709552e+19f; // this is an overflowed ulong value
with llvm3.5 the result value is
Set the data format as 1920 * 1080 four channels and type as char,short and int.
Signed-off-by: Meng Mengmeng
---
benchmark/CMakeLists.txt| 3 ++-
benchmark/benchmark_copy_buffer.cpp | 52 +
kernels/bench_copy_buffer.cl
Set the data format as 1920 * 1080 four channels(RGBA) and type as char,short
and int.
Signed-off-by: Meng Mengmeng
---
benchmark/CMakeLists.txt | 3 +-
benchmark/benchmark_copy_image.cpp | 69 ++
15 matches
Mail list logo