Re: [Beignet] [PATCH 2/2] GBE: optimize unaligned char and short data vector's load.

2014-09-01 Thread Song, Ruiling
The patch LGTM -Original Message- From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of Zhigang Gong Sent: Wednesday, August 27, 2014 12:13 PM To: beignet@lists.freedesktop.org Cc: Gong, Zhigang Subject: [Beignet] [PATCH 2/2] GBE: optimize unaligned char and short data

Re: [Beignet] [PATCH] utests: fix two utest bugs.

2014-09-01 Thread He Junyan
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(

Re: [Beignet] [PATCH] GBE: fix error in the rootn fastpath function for some special input.

2014-09-01 Thread Song, Ruiling
The patch LGTM -Original Message- From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of Zhigang Gong Sent: Tuesday, September 02, 2014 12:37 PM To: Gong, Zhigang Cc: beignet@lists.freedesktop.org Subject: Re: [Beignet] [PATCH] GBE: fix error in the rootn fastpath funct

Re: [Beignet] [PATCH] GBE: fix error in the rootn fastpath function for some special input.

2014-09-01 Thread Zhigang Gong
Ping for review. On Fri, Aug 29, 2014 at 10:13:57AM +0800, Zhigang Gong wrote: > The fastpath is to lose some accuracy but get fast speed. It is not > to generate error result. The rootn has many special input and need > to be taken care before we call the native pow directly. > This patch fix all

Re: [Beignet] [PATCH 2/2] GBE: optimize unaligned char and short data vector's load.

2014-09-01 Thread Zhigang Gong
On Wed, Aug 27, 2014 at 12:12:44PM +0800, Zhigang Gong wrote: > The gather the contiguous short/char loads into a single load instruction > could give us a good pportunity to use untyped load to optimize them. > > This patch enable the short/char load gathering at the load store optimize > pass. T

Re: [Beignet] [PATCH 1/2] GBE: relax the batch byte/short load vector size restrication.

2014-09-01 Thread Zhigang Gong
Ping for review including this 2 patches and the previous 2 patches. Thanks. On Thu, Aug 28, 2014 at 10:46:03AM +0800, Zhigang Gong wrote: > Previous restrication is that the vector size must be multiple > of DWORD. This restrication prevent the vload2/3 of char or > vload3 of ushort to be optimiz

[Beignet] [PATCH] utests: fix two utest bugs.

2014-09-01 Thread Zhigang Gong
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(+), 2 deletions(-) diff --git a/utests/runtime_event.cpp b/utests/runtime_eve

Re: [Beignet] [PATCH] Fix the global string bug for printf.

2014-09-01 Thread Zhigang Gong
LGTM, pushed, thanks. On Mon, Sep 01, 2014 at 04:18:45PM +0800, junyan...@inbox.com wrote: > From: Junyan He > > When there are multi printf statements in multi kernel > fucntions within the same translate unit, if they have > the same sting parameter, the Clang will just generate > one global s

Re: [Beignet] [PATCH V2] Fix a bug for runtime_barrier_list.cpp, event array out of bound

2014-09-01 Thread Zhigang Gong
LGTM, pushed, thanks. On Tue, Sep 02, 2014 at 10:37:02AM +0800, junyan...@inbox.com wrote: > From: Junyan He > > Signed-off-by: Junyan He > --- > utests/runtime_barrier_list.cpp | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/utests/runtime_barrier_list.cpp b/utests/r

Re: [Beignet] [Patch V2 1/2] Two minor fix.

2014-09-01 Thread Zhigang Gong
LGTM, pushed, thanks. On Mon, Sep 01, 2014 at 01:05:05PM +0800, Yang Rong wrote: > 1. Some systems don't define ulong type, use unsigned long instead of.. > 2. Use sA, sB... instead of sa, sb... to access vector 16, because sometimes > sa, sb will cause clang error. > > Signed-off-by: Yang Rong

Re: [Beignet] [hpc12/tools] build of 'sum' on 'Intel(R) HD Graphics IvyBridge M GT2' failed

2014-09-01 Thread Yang, Rong R
Hi, David It may be caused by drm's command parser. Can you disable the command parser simple and try again? The command parser disable by following command with root: echo 0 > /sys/module/i915/parameters/enable_cmd_parser Thanks, Yang Rong > -Original Message- > From: Zhigang Gong [m

[Beignet] [PATCH V2] Fix a bug for runtime_barrier_list.cpp, event array out of bound

2014-09-01 Thread junyan . he
From: Junyan He Signed-off-by: Junyan He --- utests/runtime_barrier_list.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/utests/runtime_barrier_list.cpp b/utests/runtime_barrier_list.cpp index 6987d5e..135996f 100644 --- a/utests/runtime_barrier_list.cpp +++ b/utests/run

Re: [Beignet] [PATCH] Fix a bug for runtime_barrier_list.cpp, local var not inited.

2014-09-01 Thread He Junyan
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

Re: [Beignet] [hpc12/tools] build of 'sum' on 'Intel(R) HD Graphics IvyBridge M GT2' failed

2014-09-01 Thread Zhigang Gong
It may be another kernel driver issue. CC to Rong, could you help to check whether this is a kernel related issue. The kernel version is 3.16.1-301.fc21.x86_64. Thanks, Zhigang Gong. On Mon, Sep 1, 2014 at 8:59 PM, David Liebman wrote: > I now have the patched version installed on my laptop. I r

Re: [Beignet] [hpc12/tools] build of 'sum' on 'Intel(R) HD Graphics IvyBridge M GT2' failed

2014-09-01 Thread Igor Gnatenko
I'll check it tonight. -- -Igor Gnatenko On Sep 1, 2014 4:59 PM, "David Liebman" wrote: > I now have the patched version installed on my laptop. I ran the 'sum' > program and it didn't work. I will include the output of the 'sum' program > below, but it is the same as in previous emails. One thin

Re: [Beignet] [hpc12/tools] build of 'sum' on 'Intel(R) HD Graphics IvyBridge M GT2' failed

2014-09-01 Thread David Liebman
I now have the patched version installed on my laptop. I ran the 'sum' program and it didn't work. I will include the output of the 'sum' program below, but it is the same as in previous emails. One thing to note about the 'sum' program is that it does ask for the user to input which platform y

[Beignet] [PATCH 19/22 V3] Add the bit code linker into the module pass.

2014-09-01 Thread junyan . he
From: Junyan He The bit code linker will load the beignet.bc as a lib module and link the module of the kernel together. Then we will filter out all the dead bit code by create an InternalizePass for the module. After this stage, the ir will include the bitcode just used by the cl kernel. Signed

[Beignet] [PATCH] Fix the global string bug for printf.

2014-09-01 Thread junyan . he
From: Junyan He When there are multi printf statements in multi kernel fucntions within the same translate unit, if they have the same sting parameter, the Clang will just generate one global string named .strXXX to represent that string. So when translating the kernel to gen, we can not unref th

[Beignet] [PATCH 21/22 V3] Use the PCH to accelerate the parsing speed of the ocl.h

2014-09-01 Thread junyan . he
From: Junyan He We disable the valid check for the PCH to avoid path and modified time check, which brings us some trouble. Signed-off-by: Junyan He --- backend/src/backend/program.cpp | 22 -- 1 file changed, 20 insertions(+), 2 deletions(-) diff --git a/backend/src/bac

[Beignet] [PATCH 20/22 V3] Enable libocl and disable the usage of the old huge header.

2014-09-01 Thread junyan . he
From: Junyan He The llvm ir print out is modified. >From the OCL_OUTPUT_LLVM_BEFORE_EXTRA_PASS and OCL_OUTPUT_LLVM, we change to OCL_OUTPUT_LLVM_BEFORE_LINK OCL_OUTPUT_LLVM_AFTER_LINK OCL_OUTPUT_LLVM_AFTER_GEN The first one print out the IR before link the bitcode lib. The second one print out th