From: Luo
need asign the length to a local variable.
v2: bug was a false alarm. the case passed after regenerate cmake folder.
Signed-off-by Luo
---
src/cl_gt_device.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/cl_gt_device.h b/src/cl_gt_device.h
index d7855cd..63c9047 100644
---
On Wed, Jun 25, 2014 at 09:53:18AM +0400, Nikita Orlov wrote:
> Hi!
>
> Some time ago I have asked about beignet and haswell.
> I have answer to use master of beignet and patch i915 driver in kernel
> (disable-batchbuffer-security.patch).
>
> How I can know, that patch for i915 is in upstream of
> -Original Message-
> From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of
> Nikita Orlov
> Sent: Wednesday, June 25, 2014 1:53 PM
> To: beignet@lists.freedesktop.org
> Subject: [Beignet] Haswell support and i915 driver support. Some strange bugs.
>
> Hi!
>
> Some
Hi!
Some time ago I have asked about beignet and haswell.
I have answer to use master of beignet and patch i915 driver in kernel
(disable-batchbuffer-security.patch).
How I can know, that patch for i915 is in upstream of kernel?
Are any new improvements in master of HSW support in brignet?
I am
From: Luo
need asign the length to a local variable.
Signed-off-by: Luo
---
src/cl_gt_device.h | 1 +
src/cl_program.c | 8 +---
2 files changed, 6 insertions(+), 3 deletions(-)
diff --git a/src/cl_gt_device.h b/src/cl_gt_device.h
index d7855cd..63c9047 100644
--- a/src/cl_gt_device.h
+
LGTM, pushed, thanks.
On Tue, Jun 24, 2014 at 02:23:31PM +0800, Ruiling Song wrote:
> Use native_exp() as much as possible.
>
> Signed-off-by: Ruiling Song
> ---
> backend/src/ocl_stdlib.tmpl.h | 40 +++-
> 1 file changed, 11 insertions(+), 29 deletions(-)
LGTM, pushed, thanks.
On Tue, Jun 24, 2014 at 11:33:20PM +0800, junyan...@inbox.com wrote:
> From: Junyan He
>
> Signed-off-by: Junyan He
> ---
> backend/src/ir/printf.cpp |3 +++
> backend/src/ir/printf.hpp |3 ++-
> backend/src/llvm/llvm_printf_parser.cpp
LGTM, pushed, thanks.
On Tue, Jun 24, 2014 at 04:35:58PM +0800, junyan...@inbox.com wrote:
> From: Junyan He
>
> Signed-off-by: Junyan He
> ---
> backend/src/ir/printf.cpp | 144
> +---
> backend/src/ir/printf.hpp | 11 ++-
> backend/sr
LGTM, will push latter, thanks.
On Tue, Jun 24, 2014 at 06:22:07AM +0800, Guo Yejun wrote:
> Signed-off-by: Guo Yejun
> ---
> CMakeLists.txt | 3 +++
> backend/src/CMakeLists.txt | 29 +
> src/CMakeLists.txt | 2 +-
> 3 files changed, 17 insertio
LGTM, will push latter. Thanks.
On Tue, Jun 24, 2014 at 05:36:50AM +0800, Guo Yejun wrote:
> it is needed for cross compiler
>
> Signed-off-by: Guo Yejun
> ---
> backend/CMakeLists.txt | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/backend/CMakeLists.txt b/backend/CMa
On Tue, Jun 24, 2014 at 10:09:12AM +0800, xionghu@intel.com wrote:
> From: Luo
>
> detail cases: 1Darray, 2Darray, 2Darrayto2D, 2Darrayto3D, 2Dto2Darray,
> 3Dto2Darray.
>
> 1d used gpu copy.
>
> Signed-off-by: Luo
> ---
> src/CMakeLists.txt | 4 +-
> src/cl_c
Just pushed this patch with slight change.
Thanks for the patch.
On Tue, Jun 24, 2014 at 10:09:12AM +0800, xionghu@intel.com wrote:
> From: Luo
>
> detail cases: 1Darray, 2Darray, 2Darrayto2D, 2Darrayto3D, 2Dto2Darray,
> 3Dto2Darray.
>
> 1d used gpu copy.
>
> Signed-off-by: Luo
> ---
>
Some wording comments as below:
On Tue, Jun 24, 2014 at 12:28:51AM +0800, Yang Rong wrote:
> Signed-off-by: Yang Rong
> ---
> docs/optimization-guide.mdwn | 27 +++
> 1 file changed, 27 insertions(+)
> create mode 100644 docs/optimization-guide.mdwn
>
> diff --git a/docs
From: Junyan He
Signed-off-by: Junyan He
---
backend/src/ir/printf.cpp |3 +++
backend/src/ir/printf.hpp |3 ++-
backend/src/llvm/llvm_printf_parser.cpp | 21 +
3 files changed, 22 insertions(+), 5 deletions(-)
diff --git a/backend/src/
There are two issues hided in this patch,
1. We need to enqueue a 3D task rather than a 1D task for the image filling.
2. At the fill image 1d array kernel, we should use the third dimension as the
array index
rather than using the second dimension.
And another potential issue in clEnqueueMapI
I double checked the spec and found the spec doesn't define the error code
when try to copy a non-1D image to a 1D image. The CL_IMAGE_FORMAT_NOT_SUPPORTED
is not for this purpose. IMO, the most closing error code should be
CL_IMAGE_FORMAT_MISMATCH.
On Tue, Jun 24, 2014 at 09:40:15AM +, Luo,
Image1d could only be copied to image1d right now, no need for other option,
shall I return CL_IMAGE_FORMAT_NOT_SUPPORTED for all else branches?
Luo Xionghu
Best Regards
-Original Message-
From: Zhigang Gong [mailto:zhigang.g...@linux.intel.com]
Sent: Tuesday, June 24, 2014 3:51 PM
To:
From: Junyan He
Signed-off-by: Junyan He
---
backend/src/ir/printf.cpp | 144 +---
backend/src/ir/printf.hpp | 11 ++-
backend/src/llvm/llvm_printf_parser.cpp | 83 --
kernels/test_printf.cl | 10 ++-
4
On Tue, Jun 24, 2014 at 10:09:12AM +0800, xionghu@intel.com wrote:
> From: Luo
>
> detail cases: 1Darray, 2Darray, 2Darrayto2D, 2Darrayto3D, 2Dto2Darray,
> 3Dto2Darray.
>
> 1d used gpu copy.
>
> Signed-off-by: Luo
> ---
> src/CMakeLists.txt | 4 +-
> src/cl_c
Forgot to remove the printf for debug.
Luo Xionghu
Best Regards
-Original Message-
From: Luo, Xionghu
Sent: Tuesday, June 24, 2014 10:09 AM
To: beignet@lists.freedesktop.org
Cc: Luo, Xionghu
Subject: [PATCH] add cpu copy for 1Darray and 2darray related copy APIs.
From: Luo
detail cas
From: Luo
detail cases: 1Darray, 2Darray, 2Darrayto2D, 2Darrayto3D, 2Dto2Darray,
3Dto2Darray.
1d used gpu copy.
Signed-off-by: Luo
---
src/CMakeLists.txt | 4 +-
src/cl_context.h | 1 +
src/cl_mem.c
21 matches
Mail list logo
