This patchset is OK for me.
On Tue, Jun 20, 2017 at 07:07:45PM +0800, Yang Rong wrote:
> Date: Tue, 20 Jun 2017 19:07:45 +0800
> From: Yang Rong
> To: beignet@lists.freedesktop.org
> Cc: Yang Rong
> Subject: [Beignet] [PATCH 1/3] Runtime: fix a recurrent release context
> error.
> X-Mailer: git
y and there
hundred of thoundred of them, so this is also unacceptible.
On Thu, Mar 30, 2017 at 07:35:20AM +, Yang, Rong R wrote:
> Date: Thu, 30 Mar 2017 07:35:20 +
> From: "Yang, Rong R"
> To: "junyan...@inbox.com" ,
> "beignet@lists.freedeskt
Thanks for fixing it.
On Mon, Jan 30, 2017 at 03:18:09PM +0100, Giuseppe Bilotta wrote:
> Date: Mon, 30 Jan 2017 15:18:09 +0100
> From: Giuseppe Bilotta
> To: Beignet ML
> Cc: Giuseppe Bilotta
> Subject: [Beignet] [PATCH] Typo in error message
> X-Mailer: git-send-email 2.11.0.745.g0978fb64a4
>
Thanks for catching that bug.
On Wed, Dec 28, 2016 at 06:47:01PM +0800, Yang Rong wrote:
> Date: Wed, 28 Dec 2016 18:47:01 +0800
> From: Yang Rong
> To: beignet@lists.freedesktop.org
> Cc: Meng Mengmeng , Yang Rong
>
> Subject: [Beignet] [PATCH] Runtime: return CL_INVALID_EVENT_WAIT_LIST if
>
gnet] [PATCH] Refine mem.h and improve the related macro
> defination.
>
> Hi,
>
> On 20.12.2016 04:23, He Junyan wrote:
>
> >>> +if (!CL_OBJECT_IS_BUFFER(memobj)) {
>
> >> That would match pipes as well, is that intended?
>
> > Already
I notice this should be caused by event timestamp recording sequence.
We submit the NDRange and then record the queued timestamp, this is
wrong. I have already sent an another patch to "Improve event execute function"
to modify this. You can have a try, apply that patch based on this patch.
thanks.
On Mon, Dec 19, 2016 at 06:25:26PM +0100, Simon Richter wrote:
> Date: Mon, 19 Dec 2016 18:25:26 +0100
> From: Simon Richter
> To: beignet@lists.freedesktop.org
> Subject: Re: [Beignet] [PATCH] Refine mem.h and improve the related macro
> defination.
>
> Hi,
>
> On 19.12.2016 10:21, junyan...@i
The whole patch set should be push to dev/runtime first.
When it is stable, we can merge them back.
On Tue, Jul 19, 2016 at 07:25:47PM +0800, junyan...@inbox.com wrote:
> Date: Tue, 19 Jul 2016 19:25:47 +0800
> From: junyan...@inbox.com
> To: beignet@lists.freedesktop.org
> Subject: [Beignet] [PAT
On Fri, Jul 15, 2016 at 11:50:06AM +0200, Simon Richter wrote:
> Date: Fri, 15 Jul 2016 11:50:06 +0200
> From: Simon Richter
> To: beignet@lists.freedesktop.org
> Subject: Re: [Beignet] [PATCH] Runtime: Add CL base object for all cl
> objects.
>
> Hi,
>
> On 14.07.2016 10:15, junyan...@inbox.co
Hi Mircea,
Thanks a lot for you contribution.
I have several questions here.
According to my understanding, your debug feature strongly depends on
the libigfxdbgxchg64.so lib, which I think is a binary and I can not
find it in your patchset. I notice you get a SIP of some system routine
and allocat
patch of 06 and 07 have the same title?
I think it is a typo here.
Please correct it.
All the other things are OK, just rename this one and
the whole patchset can be pushed later.
Also can push my patch about printf test cases together.
On Mon, Feb 01, 2016 at 03:42:16PM +0800, yan.w...@linux.int
Ping for review
On Thu, Dec 24, 2015 at 07:01:52PM +0800, junyan...@inbox.com wrote:
> Date: Thu, 24 Dec 2015 19:01:52 +0800
> From: junyan...@inbox.com
> To: beignet@lists.freedesktop.org
> Subject: [PATCH 00/18] Enable profiling by line number.
> X-Mailer: git-send-email 1.7.9.5
>
> From: Junya
On Thu, Jan 21, 2016 at 11:29:41AM +0800, Yan Wang wrote:
> Date: Thu, 21 Jan 2016 11:29:41 +0800
> From: Yan Wang
> To: beignet@lists.freedesktop.org
> Cc: Yan Wang
> Subject: [Beignet] [Printf][PATCH 04/11] Add the implementation of printf
> ir instruction.
> X-Mailer: git-send-email 2.5.0
>
After applied the printf patch set, I find the last test still
failed, please help to check.
On Thu, Jan 28, 2016 at 12:33:05PM +0800, He Junyan wrote:
> Date: Thu, 28 Jan 2016 12:33:05 +0800
> From: He Junyan
> To: beignet@lists.freedesktop.org
> Subject: Re: [Beignet] [Printf]
On Thu, Jan 21, 2016 at 11:29:24AM +0800, Yan Wang wrote:
> Date: Thu, 21 Jan 2016 11:29:24 +0800
> From: Yan Wang
> To: beignet@lists.freedesktop.org
> Cc: Yan Wang
> Subject: [Beignet] [Printf][PATCH 03/11] Reconstruct printf parser.
> X-Mailer: git-send-email 2.5.0
>
> Contributor: Junyan He
On Thu, Jan 21, 2016 at 11:30:21AM +0800, Yan Wang wrote:
> Date: Thu, 21 Jan 2016 11:30:21 +0800
> From: Yan Wang
> To: beignet@lists.freedesktop.org
> Cc: Yan Wang
> Subject: [Beignet] [Printf][PATCH 06/11] Implement emision of printf
> instruction.
> X-Mailer: git-send-email 2.5.0
>
> Contri
Date: Thu, 21 Jan 2016 06:50:36 +
> From: "Song, Ruiling"
> To: Paulo Dias , "He, Junyan"
>
> Cc: "beignet@lists.freedesktop.org"
> Subject: Re: [Beignet] beignet git with llvm 3.7.1 + haswell (gen 7.5)
> failing
>
> Yes, I agree with
V2 just fix some typo, please just ignore the previous one.
This patch should be merget to master and release1.1
On Wed, Jan 20, 2016 at 05:57:20PM +0800, junyan...@inbox.com wrote:
> Date: Wed, 20 Jan 2016 17:57:20 +0800
> From: junyan...@inbox.com
> To: beignet@lists.freedesktop.org
> Subject: [
LGTM, thanks.
On Fri, Jan 15, 2016 at 06:58:43AM +0800, Guo Yejun wrote:
> Date: Fri, 15 Jan 2016 06:58:43 +0800
> From: Guo Yejun
> To: beignet@lists.freedesktop.org
> Cc: Guo Yejun
> Subject: [Beignet] [PATCH] fix typo for DEBUGP to avoid print extra empty
> line
> X-Mailer: git-send-email 1.
I think the basic idea is OK and the result is als OK.
Please pay attention to the code format, we prefer spaces rather
than TABs to begin the line.
And some comments below.
On Wed, Dec 23, 2015 at 05:32:19PM +0200, Grigore Lupescu wrote:
> Date: Wed, 23 Dec 2015 17:32:19 +0200
> From: Grigore Lu
Ok, That's good for me.
Thanks for enabling BTX.
On Fri, Dec 04, 2015 at 03:22:20AM +0800, Guo Yejun wrote:
> Date: Fri, 4 Dec 2015 03:22:20 +0800
> From: Guo Yejun
> To: beignet@lists.freedesktop.org
> Cc: Guo Yejun
> Subject: [Beignet] [PATCH] add Broxton support
> X-Mailer: git-send-email 1.
On Wed, Dec 09, 2015 at 08:18:29AM +, Yang, Rong R wrote:
> Date: Wed, 9 Dec 2015 08:18:29 +
> From: "Yang, Rong R"
> To: "junyan...@inbox.com" ,
> "beignet@lists.freedesktop.org"
> Subject: Re: [Beignet] [PATCH 07/13] Backend: Add WORKGROUP_OP instruction
> selection.
>
>
>
> >
Hi Grigore,
I notice that you just reuse the kernel in utest as the benchmark kernel.
In this kernel, we just call the workgroup function once, while the time
diff calculated by your benchmark here includes the whole process of exec
a kernel on GPU.
The OCL_NDRANGE itself and the LOAD and STORE in
I think that it can also be merged to master.
On Thu, Dec 03, 2015 at 12:34:26PM +0800, He Junyan wrote:
> Date: Thu, 3 Dec 2015 12:34:26 +0800
> From: He Junyan
> To: beignet@lists.freedesktop.org
> Subject: Re: [Beignet] [PATCH OCL2.0 1/2] libocl: Add three work-item
> bu
LGTM, thanks.
On Wed, Dec 02, 2015 at 04:57:38PM +0800, Pan Xiuli wrote:
> Date: Wed, 2 Dec 2015 16:57:38 +0800
> From: Pan Xiuli
> To: beignet@lists.freedesktop.org
> Cc: xiuli@intel.com
> Subject: [Beignet] [PATCH OCL2.0 1/2] libocl: Add three work-item built-in
> function
> X-Mailer: git
This patch is OK.
I appreciate that you can also add a utest case for it.
On Wed, Dec 02, 2015 at 01:37:41PM +0800, Pan Xiuli wrote:
> Date: Wed, 2 Dec 2015 13:37:41 +0800
> From: Pan Xiuli
> To: beignet@lists.freedesktop.org
> Cc: xiuli@intel.com
> Subject: [Beignet] [PATCH OCL2.0] libocl:
This is V2.
V2:
Just rebase to master and ping for review.
On Tue, Dec 01, 2015 at 04:10:28PM +0800, junyan...@inbox.com wrote:
> Date: Tue, 1 Dec 2015 16:10:28 +0800
> From: junyan...@inbox.com
> To: beignet@lists.freedesktop.org
> Subject: [Beignet] [PATCH 01/13] Backend: Add sr0 reg helpe
This patch implement reduce max/min/add for uint and float.
It needs to base on the previous workgroup broadcast patchset(5 patches).
On Mon, Nov 23, 2015 at 04:48:38PM +0800, junyan...@inbox.com wrote:
> Date: Mon, 23 Nov 2015 16:48:38 +0800
> From: junyan...@inbox.com
> To: beignet@lists.freedes
Sorry, this patch should not belong to this patch set, please ignore it.
On Tue, Nov 17, 2015 at 07:40:23AM +0800, junyan...@inbox.com wrote:
> Date: Tue, 17 Nov 2015 07:40:23 +0800
> From: junyan...@inbox.com
> To: beignet@lists.freedesktop.org
> Subject: [Beignet] [PATCH 21/21 V3] CMake: Add -lr
V2:
Fix uniform bug in conversion.
Delete verbose printf in utests.
Fix a bug for BSW when convert half to double.
On Thu, Nov 05, 2015 at 04:15:41PM +0800, junyan...@inbox.com wrote:
> Date: Thu, 5 Nov 2015 16:15:41 +0800
> From: junyan...@inbox.com
> To: beignet@lists.freedesktop.org
> Subject
Ping for review.
On Tue, Sep 22, 2015 at 06:29:23PM +0800, junyan...@inbox.com wrote:
> Date: Tue, 22 Sep 2015 18:29:23 +0800
> From: junyan...@inbox.com
> To: beignet@lists.freedesktop.org
> Subject: [Beignet] [PATCH] Backend: Refine ConvertInstruction logic in
> insn_selection
> X-Mailer: git-s
Ping for review.
On Fri, Sep 18, 2015 at 05:58:11PM +0800, junyan...@inbox.com wrote:
> Date: Fri, 18 Sep 2015 17:58:11 +0800
> From: junyan...@inbox.com
> To: beignet@lists.freedesktop.org
> Subject: [Beignet] [V2 PATCH 0/8] Implement double division on BDW
> X-Mailer: git-send-email 1.7.9.5
>
>
LGTM, thanks
On Sat, Oct 10, 2015 at 06:55:45AM -0400, xionghu@intel.com wrote:
> Date: Sat, 10 Oct 2015 06:55:45 -0400
> From: xionghu@intel.com
> To: beignet@lists.freedesktop.org
> Cc: xionghu@intel.com
> Subject: [Beignet] [PATCH] fix __kernel function called in __kernel issue.
> X
On Fri, Sep 18, 2015 at 05:01:57PM +0800, Bai Yannan wrote:
> Date: Fri, 18 Sep 2015 17:01:57 +0800
> From: Bai Yannan
> To: beignet@lists.freedesktop.org
> Cc: Bai Yannan , Lv Meng
> Subject: [Beignet] [PATCH 7/7] refine pass debug info from llvm ir to gen
> insn
> X-Mailer: git-send-email 1.9.
On Fri, Sep 18, 2015 at 05:01:54PM +0800, Bai Yannan wrote:
> Date: Fri, 18 Sep 2015 17:01:54 +0800
> From: Bai Yannan
> To: beignet@lists.freedesktop.org
> Cc: Bai Yannan
> Subject: [Beignet] [PATCH 4/7] fix bug that LOADI cannot inherits debug info
> X-Mailer: git-send-email 1.9.1
>
> Signed-o
On Fri, Sep 18, 2015 at 05:01:53PM +0800, Bai Yannan wrote:
> Date: Fri, 18 Sep 2015 17:01:53 +0800
> From: Bai Yannan
> To: beignet@lists.freedesktop.org
> Cc: Bai Yannan
> Subject: [Beignet] [PATCH 3/7] pass dbginfo from gen ir to geninsn
> X-Mailer: git-send-email 1.9.1
>
> 1, pass debug
On Fri, Sep 18, 2015 at 05:01:52PM +0800, Bai Yannan wrote:
> Date: Fri, 18 Sep 2015 17:01:52 +0800
> From: Bai Yannan
> To: beignet@lists.freedesktop.org
> Cc: Bai Yannan
> Subject: [Beignet] [PATCH 2/7] transfer debugloc from llvm ir to gen ir
> X-Mailer: git-send-email 1.9.1
>
> Signed-off-by
This patch set can basically work, but has some problems.
Common issue:
1. Please pay attention to the code format, the line should not
begin with TAB and should not have trailing SPACEs.
2. Some tmp comments should be deleted, and no need for author
name in code.
3. I think the reorganise t
On Tue, Sep 15, 2015 at 05:57:13AM -0700, Matt Turner wrote:
> Date: Tue, 15 Sep 2015 05:57:13 -0700
> From: Matt Turner
> To: "junyan.he"
> Cc: "beignet@lists.freedesktop.org"
> Subject: Re: [Beignet] [PATCH 5/8] Backend: Add the MADM function to gen8
> encoder.
>
> On Tue, Sep 15, 2015 at 4:
On Tue, Sep 15, 2015 at 06:00:57AM -0700, Matt Turner wrote:
> Date: Tue, 15 Sep 2015 06:00:57 -0700
> From: Matt Turner
> To: "junyan.he"
> Cc: "beignet@lists.freedesktop.org"
> Subject: Re: [Beignet] [PATCH 6/8] Backend: Implement FDIV64 on BDW.
>
> On Tue, Sep 15, 2015 at 4:15 AM, wrote:
>
I think it would be better after we integrate the binary_to_soure
feature. Then I think it is easy and clear to describe how to use
the profiling feature.
On Thu, Sep 10, 2015 at 05:56:17AM +, Zou, Nanhai wrote:
> Date: Thu, 10 Sep 2015 05:56:17 +
> From: "Zou, Nanhai"
> To: "junyan...@in
The half float can work for BSW,
I will send a standalone patch to enable it later.
On 2015年06月19日 15:18, Yang, Rong R wrote:
One concern: Does cherryview support half?
The other part of the patchset LGTM.
-Original Message-
From: Beignet [mailto:beignet-boun...@lists.freedesktop.org
: Wednesday, July 01, 2015 10:40 AM
To: Gerald Baier; beignet@lists.freedesktop.org
Cc: He, Junyan
Subject: RE: [Beignet] thread safety and OpenMP
> -Original Message-
> From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of
> Gerald Baier
> Sent: Sunday, Jun
.
According to this, I still prefer to use the software imitation
for half float. It's verbose, but it will not have any side effect
for runtime performance.
On 2015年05月22日 14:28, He Junyan wrote:
Thanks for your information.
I will do some research for it.
On 2015年05月22日 05:51, Matt Turner
Some comments,
On 2015年04月03日 13:39, xionghu@intel.com wrote:
From: Luo Xionghu
this patch allows create 2d image with a cl buffer.
Signed-off-by: Luo Xionghu
---
src/cl_api.c | 3 ++-
src/cl_mem.c | 67 +++-
2 files changed, 5
Thanks for your information.
I will do some research for it.
On 2015年05月22日 05:51, Matt Turner wrote:
On Thu, May 21, 2015 at 1:25 AM, wrote:
From: Junyan He
Because the CPU of X86 does not support half float
instructions, there is no support for half float operations.
So we introduce the
I have modified it in the new patch set.
Just ignore this one. Thanks.
On 2015年04月30日 13:49, Zhigang Gong wrote:
Junyan,
I haven't found any new response to address this comment from Rong and me.
Do you miss this comment or do I miss your new patch?
Thanks,
Zhigang Gong.
On Thu, Apr 02, 2015
That's OK
On 2015年03月23日 17:09, Yang Rong wrote:
HADD is equal to (src0>>1) + (src1>>1) + ((src0&0x1) & (src1&0x1)),
and RHADD is equal to (src0>>1) + (src1>>1) + ((src0&0x1) | (src1&0x1)).
V2: Save 1 register for both, and 1 AND for I64HADD.
Signed-off-by: Yang Rong
---
backend/src/backend
OK, it's a better way to avid the usage of addc.
I think tmp_dst can also be avoided here to save
one tmp register.
On 2015年03月23日 15:44, Song, Ruiling wrote:
Good idea, the patch LGTM.
-Original Message-
From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of
Yang R
On 2015年03月09日 09:11, Zhigang Gong wrote:
On Fri, Mar 06, 2015 at 03:24:00PM +0800, junyan...@inbox.com wrote:
From: Junyan He
The swap for short will be like:
mov(1) a0<1>:UD0xe600e61UD{ align1 WE_all };
mov(1) a0.1<1>:UD 0xe620e63UD{ align1 WE_all };
It's really a bug.
Because of the HW limitation, vertical stride is at least aligned to 2.
For 1D array image, the data has interval. The size calculated in memset
is right,
but the image size is just twice as big as it.
Use clEnqueueWriteImage is safe and I will fix it later.
On 2015年02月04日 07
Except some format problem, this patchset LGTM and can pass all the
utest cases on my platform.
On 2015年01月30日 10:59, Yang Rong wrote:
Skl's LD message payload order is changed from u, lod, v, w to u, v, lod, w.
Add the Gen9Context and Selection9 to handle it.
Skl Still use Gen8Encoder.
Signed
Hi,
It seems that gen9_context.hpp and gen9_context.cpp are lost and can not
compile.
On 2015年01月29日 16:16, Yang Rong wrote:
SKL add the new GT4 type device.
Signed-off-by: Yang Rong
---
src/cl_device_data.h | 45 +++
src/cl_device_id.c | 122 +
On 四, 2015-01-08 at 13:14 +0800, Zhigang Gong wrote:
> On Tue, Jan 06, 2015 at 06:01:54PM +0800, junyan...@inbox.com wrote:
> > From: Junyan He
> >
> > The conversion logic is too complicated.
> > We split it more clearly for each case.
> > Notice: For I64 to I8, the conversion can not be compl
Sorry to add V2 comment, just ignore this.
On 二, 2015-01-13 at 15:40 +0800, junyan...@inbox.com wrote:
> From: Junyan He
>
> We can not know the accurate size of the printf buffer
> size before run the kernel. Sometimes, especially when
> the global work items size is huge, the output buffer
>
The Print imm logic is added in the [PATCH 10/27]
of the new patchset.
On 二, 2015-01-06 at 00:50 +, Yang, Rong R wrote:
> Also need add long/ulong imm print.
>
> > -Original Message-
> > From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of
> > junyan...@inbox.co
OK, LGTM, thanks
On 二, 2014-12-16 at 16:37 +0800, Yan Wang wrote:
> PrintfState include std::string object and shouldn't be copied by
> malloc/memcpy.
>
> Signed-off-by: Yan Wang
> ---
> backend/src/ir/printf.hpp | 33 +
> 1 file changed, 29 insertions(+), 4 dele
On 二, 2014-12-09 at 12:41 +0800, Yan Wang wrote:
> PrintfState includes std::string object and shouldn't be copied by
> malloc/memcpy.
>
> Signed-off-by: Yan Wang
> ---
> backend/src/ir/printf.hpp | 23 +++
> 1 file changed, 19 insertions(+), 4 deletions(-)
>
> diff --git
It's good for me
On 三, 2014-11-12 at 14:12 +0800, Zhigang Gong wrote:
> As we still have the image 1d array workaround, we need to
> fix it for BDW as well.
>
> Signed-off-by: Zhigang Gong
> ---
> src/intel/intel_gpgpu.c | 6 --
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --
This patchset LGTM
On 一, 2014-09-29 at 13:37 +0800, Yang Rong wrote:
> From: Junyan He
>
> Must call cl_bind_buf instead of intel_gpgpu_bind_buf directly in intel_gpgpu.
>
> Signed-off-by: Junyan He
> ---
> src/intel/intel_gpgpu.c | 36 +++-
> 1 file changed, 2
`gpgpu->binded_buf[i]->offset != 0' failed.
Segmentation fault (core dumped)
Sometimes this kind of assert may have relationship with multi-thread
bug.
On 四, 2014-10-23 at 06:12 -0700, yan.w...@linux.intel.com wrote:
> Update:
> Under root account, it will crash.
> Under user account(e.g. yanwa
OK, I have tried the issue you mentioned,
and it really causes some problems.
Thanks for your reporting and please
re-send a patch to include all the places you
have mentioned.
On 日, 2014-10-19 at 17:30 +0100, Steven Newbury wrote:
> On Sun, 2014-10-19 at 17:13 +0100, Steven Newbury wrote:
> > On
Yes, we have found this problem.
New version beignet requires libdrm at least 2.4.52
Old version of libdrm will cause the build fail.
We have already add the libdrm version check
On 四, 2014-10-16 at 21:53 -0700, Andi Kleen wrote:
> I tried the current beignet master on my OpenSUSE 13.1 HSW and
>
OK, LGTM
On 四, 2014-10-16 at 15:11 +0800, Yang Rong wrote:
> Signed-off-by: Yang Rong
> ---
> src/intel/intel_gpgpu.c | 6 ++
> 1 file changed, 6 insertions(+)
>
> diff --git a/src/intel/intel_gpgpu.c b/src/intel/intel_gpgpu.c
> index 259882a..167d8d9 100644
> --- a/src/intel/intel_gpgpu.c
This patch set will cause displacement_map_element case hang every time.
But no regression found on previous platforms.
We can find the bug later and fix it.
On 一, 2014-09-29 at 13:38 +0800, Yang Rong wrote:
> BDW's scratch buffer change to power 2 alignment from 1024.
>
> Signed-off-by: Yang Ro
This patchset is OK and will not cause regression on previous platform.
In this patch set, the GenEncoder will be a pure virtual class and
all platform encoders will derive from it.
But the GenContext still represents the Gen7 context. I think it is
better to follow the same way as the encoder to m
LGTM
On 四, 2014-09-18 at 08:35 +0800, Zhigang Gong wrote:
> Signed-off-by: Zhigang Gong
> ---
> backend/src/libocl/CMakeLists.txt | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/backend/src/libocl/CMakeLists.txt
> b/backend/src/libocl/CMakeLists.txt
> index f015eec
Sorry, this if V2
V2:
Replace all the long and ulong to int64_t
On 四, 2014-09-18 at 12:39 +0800, junyan...@inbox.com wrote:
> From: Junyan He
>
> Signed-off-by: Junyan He
> ---
> backend/src/ir/printf.cpp | 25 -
> backend/src/llvm/llvm_printf_parse
LGTM
thanks
On 三, 2014-09-17 at 11:33 +0800, Ruiling Song wrote:
> Signed-off-by: Ruiling Song
> ---
> backend/src/llvm/llvm_bitcode_link.cpp |5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/backend/src/llvm/llvm_bitcode_link.cpp
> b/backend/src/llvm/llvm_bitcod
LGTM, thanks
On 二, 2014-09-16 at 09:57 +0800, Zhigang Gong wrote:
> Ping for review.
>
> On Fri, Sep 12, 2014 at 05:18:16PM +0800, Zhigang Gong wrote:
> > Signed-off-by: Zhigang Gong
> > ---
> > backend/src/libocl/script/ocl_math.def | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
LGTM, thanks
On 二, 2014-09-16 at 09:57 +0800, Zhigang Gong wrote:
> Ping for review.
>
> On Fri, Sep 12, 2014 at 05:38:06PM +0800, Zhigang Gong wrote:
> > Signed-off-by: Zhigang Gong
> > ---
> > backend/src/libocl/tmpl/ocl_math.tmpl.cl | 8
> > 1 file changed, 4 insertions(+), 4 deleti
Hi:
I found after apply the patch
4a61eef93401c247eb9d2819da9fa9d7106d5eba
Enable ICC and CLANG compiler for beignet
the 'COMPILER' is to choose the detail compiler,the default is GCC.
Signed-off-by: Lv Meng
Reviewed-by: Zhigang Gong
some linkage problems occur.
I do not install my clang/ll
I suggest that you delete the PCH file in the system install dir
and try again.
On 五, 2014-09-12 at 07:52 -0400, Yichao Yu wrote:
> On Fri, Sep 12, 2014 at 7:32 AM, Yichao Yu wrote:
> > Hi,
> >
> > I've got an error when generating src/kernels/cl_internal_*_str.c's.
> >
> > The error message read
The patchset is OK
On 五, 2014-09-12 at 14:33 +0800, Zhigang Gong wrote:
> If we want to link multiple files together, and one kernel
> function need refer other kernel functions in other files,
> we must not set those functions as linked once attribute.
>
> Signed-off-by: Zhigang Gong
> ---
>
It's my typo, thanks for your fixing
On 三, 2014-09-10 at 16:23 +0800, Zhigang Gong wrote:
> Signed-off-by: Zhigang Gong
> ---
> backend/src/libocl/tmpl/ocl_math.tmpl.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/backend/src/libocl/tmpl/ocl_math.tmpl.h
> b/backend/
OK, LGTM
On 二, 2014-09-09 at 20:31 -0400, Yichao Yu wrote:
> Signed-off-by: Yichao Yu
> ---
> backend/src/libocl/CMakeLists.txt | 8
> 1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/backend/src/libocl/CMakeLists.txt
> b/backend/src/libocl/CMakeLists.txt
> index fb93
LGTM,
thanks.
On 四, 2014-09-04 at 13:59 +0800, Zhigang Gong wrote:
> This is not a predefined macro according to the spec. Let's not
> define it by default. This patch also disable the fp64 when enter
> user kernels.
>
> v2:
> Some internal .cl files require cl_khr_fp64 enabled. Fixed that issue
That OK
On 二, 2014-09-02 at 10:36 +0800, Zhigang Gong wrote:
> Similar as the bug found by junyan, some events are
> accessed before assigned.
>
> Signed-off-by: Zhigang Gong
> ---
> utests/runtime_event.cpp | 2 +-
> utests/runtime_marker_list.cpp | 2 +-
> 2 files changed, 2 insertions(
That's better than memset,
It's OK
On 一, 2014-09-01 at 09:27 +0800, Zhigang Gong wrote:
> I just checked the test case. This may not the best fix. The issue should be
> the first time to access all the events which also inclues the uninitialized
> event 3,4,5, And the following patch should be b
The dependencies of these files are complex here and may cause some
inconvenience because the CMake lacks of the flexibility.
After discussion with Zhigang, I notice that it is important to keep it
complying with the whole project and I will rewrite the all the building
files using CMake.
On 三, 2
On 四, 2014-08-28 at 02:51 +, Song, Ruiling wrote:
>
> Suppose you cannot remove below lines of code, it is for clCompileProgram()
> -//for clCompilerProgram usage.
> -if(temp_header_path){
> - clOpt += " -I ";
> - clOpt += temp_header_path;
> - clOpt += " ";
> -}
>
OK,
Thanks for finding this bug.
On 二, 2014-08-26 at 15:39 +0800, Ruiling Song wrote:
> It should be cleared, to prevent invalid pointers staying there
> when processing next Function.
>
> Signed-off-by: Ruiling Song
> ---
> backend/src/llvm/llvm_printf_parser.cpp |1 +
> 1 file changed, 1
Some comments,
On 五, 2014-07-04 at 12:00 +0800, Lv Meng wrote:
> Signed-off-by: Lv Meng
> ---
> src/CMakeLists.txt | 3 ++-
> src/cl_context.h| 1 +
> src/cl_mem.c| 27
> +++--
That's OK
On 四, 2014-07-03 at 12:53 +0800, Zhigang Gong wrote:
> To comply with the full profile.
>
> Signed-off-by: Zhigang Gong
> ---
> src/cl_gt_device.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/src/cl_gt_device.h b/src/cl_gt_device.h
> index 97ba7e2..63c9047
OK, that's LGTM
On 四, 2014-07-03 at 14:14 +0800, Zhigang Gong wrote:
> When pending a command queue, we need to record the whole gpgpu
> structure not just the batch buffer. For the following reason:
>
> 1. We need to keep those private buffer, for example those printf buffers.
> 2. We need to ma
V2 just do the rebase and warning killing
On 五, 2014-06-20 at 17:41 +0800, junyan...@inbox.com wrote:
> From: Junyan He
>
> Signed-off-by: Junyan He
> ---
> kernels/test_printf.cl | 19 +++
> 1 file changed, 19 insertions(+)
>
> diff --git a/kernels/test_printf.cl b/kernels/te
Spec says:
For a 1D image buffer
object, the image pixels are taken from the buffer object’s data store.
When the contents of a
buffer object’s data store are modified, those changes are reflected in
the contents of the 1D
image buffer object and vice-versa at corresponding sychronization
points.
On 五, 2014-06-20 at 07:18 +, Yang, Rong R wrote:
> Two comments.
>
> -Original Message-
> From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of
> junyan...@inbox.com
> Sent: Wednesday, June 18, 2014 2:42 PM
> To: beignet@lists.freedesktop.org
> Cc: Junyan He
> Su
The hw limitation cause us to add a lot of
tricks into our code. Maybe we need to find
some place to declare details for others
to understand.
The whole patch set is OK for me.
On 四, 2014-06-19 at 15:44 +0800, Zhigang Gong wrote:
> As sample LD message doesn't support array index, we have
> to cr
Thanks for find the problem.
I think the key point is that for array image it always uses the
slice_pitch rather than the image_row_pitch.
This patch set is good for me.
As you mentioned, I will improve my utest case later, and you can
push my first 2 patches firstly to make your patch set work.
Really has risk here.
Thanks for fixing it
On Tue, 2014-06-17 at 11:18 +0800, Zhigang Gong wrote:
> cl_gpgpu_flush may be called when the batch buffer has been
> released. We need to check whether there is a valid buffer
> before we really take the following actions.
>
> Signed-off-by: Zhigang
for this patch set, the compare_image_2d_and_1d_array output:
## x y is (30, 15), color1 is (2 1 30 15), color2 is (2 1 30 30)
## x y is (31, 0), color1 is (2 1 31 0), color2 is (2 1 31 0)
## x y is (31, 1), color1 is (2 1 31 1), color2 is (2 1 31 2)
## x y is (31,
Sorry, this patch is for opencl-1.2
On Thu, 2014-06-12 at 14:55 +0800, junyan...@inbox.com wrote:
> From: Junyan He
>
> Signed-off-by: Junyan He
> ---
> utests/compiler_fill_image_3d.cpp | 4
> utests/compiler_fill_image_3d_2.cpp | 4
> 2 files changed, 8 insertions(+)
>
> diff --
Tested on my HSW platform,
no obvious regression found.
On Thu, 2014-06-12 at 19:42 +0800, Yang Rong wrote:
> GT3 have 4 half slice, so should shift left 2 bits, and also should enlarge
> the stack buffer size,
> otherwize, if thread generate is non-balance, may out of bound.
> Per bspec, scrat
I verified this patch on both my 32bits and 64bits IVB platforms.
Both results are OK and pass the profiling_exec test case.
On Tue, 2014-06-10 at 16:17 +0800, Yang Rong wrote:
> In x86_64 system, the low 32bits of timestamp count are stored in the high 32
> bits of result which
> got from drm_i
al Message-
> > From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of
> He
> > Junyan
> > Sent: Tuesday, June 10, 2014 12:33 PM
> > To: Zhigang Gong
> > Cc: beignet@lists.freedesktop.org
> > Subject: Re: [Beignet] [[OpenCL-1.2]] GBE: E
That's OK
On Fri, 2014-06-06 at 05:07 +0800, xionghu@intel.com wrote:
> From: Luo
>
> creates an array of sub-devices that each reference a non-intersecting
> set of compute units within in_device, according to a partition scheme
> given by properties.
> ---
> src/cl_api.c | 10 +
hi,
I want to add a fake stub function here, printf the warning of not
implement, and return CL_SOME_ERRORXX, the null function always cause
the program crash.
On Tue, 2014-06-10 at 09:01 +0800, Zhigang Gong wrote:
> Signed-off-by: Zhigang Gong
> ---
> src/cl_khr_icd.c | 14 +++---
> 1
Hi
I find the
drm_intel_gem_context_create, which will call
DRM_IOCTL_I915_GEM_CONTEXT_CREATE IOCtrl of kernel.
This is implemented after kernel 3.10 version.
So if the kernel version is before 3.10,
assert(driver->ctx);
will happen.
So our beignet will not support old kernel version now,
I thi
So, you apply the same logic in the function
intel_gpgpu_event_get_exec_timestamp here.
BSpec really have some mistakes.
But for most of the IVB platform,
>-result = result & 0xF000;
>-result = result >> 28;
works very well.
So I think if you want to correct this, you should
1 - 100 of 128 matches
Mail list logo
