Re: [Beignet] [PATCH v2] GetEventProfilingInfo could query all event status.

2015-08-03 Thread Zhigang Gong
The spec only says if event is not in CL_COMPLETE status, it will get CL_PROFILING_INFO_NOT_AVAILABLE. It doesn't mean a clFinish() should update all events status. According to the spec, clFinish() only need to make sure the previously enqueued task has been processed and completed. Form my poin

Re: [Beignet] [PATCH] output float immediate value with %f instead of %g

2015-08-03 Thread Guo, Yejun
Got it, thanks. The ASM is outputted for human reading, it is expected that a correct and human readable format is used. -Original Message- From: Song, Ruiling Sent: Tuesday, August 04, 2015 12:41 PM To: Guo, Yejun; beignet@lists.freedesktop.org Subject: RE: [Beignet] [PATCH] output flo

[Beignet] [PATCH v2] GetEventProfilingInfo could query all event status.

2015-08-03 Thread xionghu . luo
From: Luo Xionghu this could fix the shoc project reported CL_PROFILING_INFO_NOT_AVAILABLE issue. https://github.com/vetter/shoc/issues/47 v2: per spec, the GetEventProfilingInfo need return CL_PROFILING_INFO_NOT_AVAILABLE if the event is not CL_COMPLETE, so the event should be updated in clFinis

Re: [Beignet] [PATCH] output float immediate value with %f instead of %g

2015-08-03 Thread Song, Ruiling
> -Original Message- > From: Guo, Yejun > Sent: Tuesday, August 4, 2015 10:50 AM > To: Song, Ruiling; beignet@lists.freedesktop.org > Subject: RE: [Beignet] [PATCH] output float immediate value with %f instead > of %g > > Yes, it would be better, not sure why the current code generates "

Re: [Beignet] [PATCH] backend/src/backend: Handle -dump-opt-llvm=[PATH]

2015-08-03 Thread Song, Ruiling
LGTM Thanks! Ruiling > -Original Message- > From: Navare, Manasi D > Sent: Monday, August 3, 2015 6:37 PM > To: Song, Ruiling; beignet@lists.freedesktop.org > Cc: Ekstrand, Laura D; Navare, Manasi D > Subject: [PATCH] backend/src/backend: Handle -dump-opt-llvm=[PATH] > > From: Laura Ekst

Re: [Beignet] [PATCH] output float immediate value with %f instead of %g

2015-08-03 Thread Guo, Yejun
Yes, it would be better, not sure why the current code generates "F" for the mov instruction, I'll take a look. -Original Message- From: Song, Ruiling Sent: Tuesday, August 04, 2015 10:45 AM To: Guo, Yejun; beignet@lists.freedesktop.org Subject: RE: [Beignet] [PATCH] output float immedia

Re: [Beignet] [PATCH] output float immediate value with %f instead of %g

2015-08-03 Thread Song, Ruiling
> -Original Message- > From: Guo, Yejun > Sent: Sunday, August 2, 2015 10:42 AM > To: Song, Ruiling; beignet@lists.freedesktop.org > Subject: RE: [Beignet] [PATCH] output float immediate value with %f instead > of %g > > For the kernel > > uint aaa = 0xfffefffe; > ... > > It is finally

Re: [Beignet] [PATCH] backend/src/backend: Handle -dump-opt-llvm=[PATH]

2015-08-03 Thread Navare, Manasi D
Hi Ruiling, Below is the patch I have resubmitted with the requested changes to support LLVM-3.6. This patch below enables a feature (LLVM dump) for the Intel tool OpenCL Code Builder. Kindly review the patch below. Regards Manasi -Original Message- From: Navare, Manasi D Sent: Monday

[Beignet] [PATCH] backend/src/backend: Handle -dump-opt-llvm=[PATH]

2015-08-03 Thread Manasi Navare
From: Laura Ekstrand Allows the user to request a dump of the LLVM-generated IR to the file specified in [PATH]. Signed-off-by: Manasi Navare --- backend/src/backend/program.cpp | 37 ++--- 1 file changed, 34 insertions(+), 3 deletions(-) diff --git a/backend/s

Re: [Beignet] [PATCH] runtime: fix event status check bug.

2015-08-03 Thread Yang, Rong R
Although clGetEventProfilingInfo could get CL_PROFILING_COMMAND_QUEUED and other times, but them seems only available after the event finished. CL_PROFILING_INFO_NOT_AVAILABLE if the CL_QUEUE_PROFILING_ENABLE flag is not set for the command-queue, if the execution status of the command identifie