- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
In my application I run my kernels several times (1 to 50000 times). All the steps use the same kernel and approximatively do the same work. But, after a few time (40 steps) the application in some context crash !
But where !!??
Is there a way to know where an OpenCL application crash ?
Thanks
Link Copied
7 Replies
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
If you use Visual Studio you can look in output window and see where do you have the exception.
In linux, you can use GDB and then load and analyze core dump.
If you use Visual Studio you can look in output window and see where do you have the exception.
In linux, you can use GDB and then load and analyze core dump.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Thanks Evgeny,
I use Visual Studio 2008 and only got the following message :
I use Visual Studio 2008 and only got the following message :
First-chance exception at 0x0fb2c160 in pureStudio.exe: 0xC0000005: Access violation writing location 0x1dd4fffc.
Unhandled exception at 0x0fb2c160 in pureStudio.exe: 0xC0000005: Access violation writing location 0x1dd4fffc.
But I don't know where (in my OpenCL kernel) the application crash :-(
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
What I have notice is that if I change my "build options" to "-g -cl-opt-disable", then I have NO crash !
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
The issue might be in some compiler optimization.
Can you provide small reprodusor?
Thanks,
Evgeny
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Evgeny,
Unfortunately no ! The software is a big application I'm working on it since 2 years now !
It is difficult to know "where" the crash occur, I have no information in the output, just that VS2008 show me a 'dissassemble' window with some assembly code !
My current task is to stabilize the software, and first, with the Intel SDK.
What I can do is to provide to someone of Intel an access to the code repository ! Or simply provide a "debug" version of the software too ?
In fact I hope to release the software soon, but first I must be sure that it work with at least one SDK. So, do you think that we can find an agreement with Intel to insure that the application work (before the release of the SDK) ? and why not... my application can serve as a "showcase" ?
Krys
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
So,
How can I help you to fix the problem ?
BTW: it will be fine to have a tool to debug and find memory leaks that occurs in the OpenCL kernel. Today, when a crash occur... we have no idea of the reason of this crash (and where it occur).
Thanks
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
I'm sorry, but the only way we can help if you can provide small reprodosor for the issue.
OpenCL products are emerging and maybe you can find a tool that can help you.
I'm sorry, but the only way we can help if you can provide small reprodosor for the issue.
OpenCL products are emerging and maybe you can find a tool that can help you.

Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page