Add two functions gbe_kernel_use_device_enqueue and
gbe_program_get_device_enqueue_kernel_name.
Signed-off-by: Yang Rong
---
backend/src/backend/program.cpp | 17 -
backend/src/backend/program.h | 6 ++
backend/src/gbe_bin_interpreter.cpp |
Because in flush the command queue, must check the currunt flushed
command queue has device enqueue or not, it need the cl_kernel. So store
the cl_kernel pointer to gpgpu. And add two function intel_gpgpu_set_kernel
and intel_gpgpu_get_kernel for it.
Signed-off-by: Yang Rong
if the image width and height is not aligned, the VME hardware block
could use the data out of the image, there is no clear rule defines
the behavior of this case, so do not check the MVs near the border.
Signed-off-by: Guo Yejun
---
> -Original Message-
> From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of
> xionghu@intel.com
> Sent: Friday, March 11, 2016 11:42 PM
> To: beignet@lists.freedesktop.org
> Cc: Luo, Xionghu
> Subject: [Beignet] [Patch v2 2/2] [OCL20]
Signed-off-by: Yang Rong
---
backend/src/backend/program.h | 1 +
backend/src/ir/profile.cpp| 3 ++-
backend/src/ir/profile.hpp| 3 ++-
3 files changed, 5 insertions(+), 2 deletions(-)
diff --git a/backend/src/backend/program.h b/backend/src/backend/program.h
Pushed, thanks.
> -Original Message-
> From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of
> Guo, Yejun
> Sent: Tuesday, March 8, 2016 14:52
> To: Song, Ruiling ; beignet@lists.freedesktop.org
> Subject: Re: [Beignet] [PATCH] enable
Signed-off-by: Yang Rong
---
src/cl_api.c | 5 -
1 file changed, 5 deletions(-)
diff --git a/src/cl_api.c b/src/cl_api.c
index 5697e34..19c0ec7 100644
--- a/src/cl_api.c
+++ b/src/cl_api.c
@@ -488,11 +488,6 @@ clCreateCommandQueueWithProperties(cl_context
From: Luo Xionghu
should use opencl 2.0 API to test generic atomic, the atomic_fetch_add
will return the old value of the source pointer, so should move the src
data to dst with another instruction.
---
kernels/compiler_generic_atomic.cl | 3 ++-
1 file changed, 2
Now this case could passed when previous test_printf case has multiply tests.
VME engine seems to read data out of specified image buffer which is based
on drm bo.
If this drm bo of src/ref image object reuse from previous bo with garbage
by coincidence, it will cause different MV results.
Yan
From: Grigore Lupescu
Signed-off-by: Grigore Lupescu
---
kernels/bench_math.cl | 12 ++--
1 file changed, 10 insertions(+), 2 deletions(-)
diff --git a/kernels/bench_math.cl b/kernels/bench_math.cl
index b53ba50..a6b6b94 100644
--- a/kernels/bench_math.cl
+++ b/kernels/bench_math.cl
Signed-off-by: Yang Rong
---
kernels/compiler_device_enqueue.cl | 19 +++
utests/CMakeLists.txt | 3 ++-
utests/compiler_device_enqueue.cpp | 36
3 files changed, 57 insertions(+), 1 deletion(-)
create
Src may be scaler when unpack, only check dst.
Signed-off-by: Yang Rong
---
backend/src/backend/gen8_context.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/backend/src/backend/gen8_context.cpp
b/backend/src/backend/gen8_context.cpp
index
LGTM, pushed, thanks.
> -Original Message-
> From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of
> Guo, Yejun
> Sent: Tuesday, March 8, 2016 14:54
> To: Guo, Yejun ; beignet@lists.freedesktop.org
> Subject: Re: [Beignet] [PATCH] change behavior
LGTM, pushed.
> -Original Message-
> From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of
> Xiuli Pan
> Sent: Wednesday, March 16, 2016 7:53
> To: beignet@lists.freedesktop.org
> Cc: Pan, Xiuli
> Subject: [Beignet] [PATCH 1/2] Backend: Refine
From: Luo Xionghu
use the variable NOT_BUILD_STAND_ALONE_UTEST to control the build type:
for beignet build, set it to 1; for stand alone build, do NOT need set
it.
remove all clXXXIntel extension call and such kind of tests since we
intend to provide the unit test
This functions collect all device enqueue's invoke functions and
store them in the unit, and set those functions to OpenCL kernel function.
Because it change the module's kernel functions, so must called before
link, otherwize, the built-in functions called in invoke functions
may not be
From: Luo Xionghu
should call atomic_compare_exchange_strong instead of the gen type call.
v2: still use __gen_ocl_atomic_compare_exchange_strong32 to implement
the atomic_flag_test_and_set function as the spec says:
"Returns atomically, the value of the object
17 matches
Mail list logo