- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
My application with Qt UI uses the OpenCL backend of the MNN Framework to perform concurrent inference on multiple neural networks.
It crashes on this machine:
- PC: HP ProBook 440 G1
- CPU: Intel(R) Core(TM) Ultra 5 125H
- GPU: Intel(R) Arc(TM) Graphics
- GPU Driver: 32.0.101.6651
The crash dump info and stack is as follows:
- 0x00007FFB3BB687B3 (igxelpgicd64.dll) (a963fad9-49c1-4644-8316-a3be326cc8e4.dmp 中)处有未经处理的异常: 0xC0000005: 读取位置 0x0000000000000018 时发生访问冲突。
- > igxelpgicd64.dll!00007ffb3bb687b3() 未知
igxelpgicd64.dll!00007ffb3bb61ab4() 未知
igxelpgicd64.dll!00007ffb3bb5fd76() 未知
igxelpgicd64.dll!00007ffb3bb60088() 未知
igxelpgicd64.dll!00007ffb3bb69ce7() 未知
igxelpgicd64.dll!00007ffb3bafbde5() 未知
igxelpgicd64.dll!00007ffb3bbb6228() 未知
igxelpgicd64.dll!00007ffb3bbb7ce7() 未知
igxelpgicd64.dll!00007ffb3bf41b3e() 未知
Qt6Gui.dll!executeCommandBuffer() 行 2788 C++
Qt6Gui.dll!endFrame() 行 1989 C++
Qt6Gui.dll!endFrame() 行 7622 C++
Qt6Quick.dll!syncAndRender() 行 744 C++
Qt6Quick.dll!run() 行 942 C++
Qt6Core.dll!start() 行 292 C++
[外部代码]
At the same time I observed that MNN print some error logs when calling the OpenCL API enqueueWriteBuffer() with size 150528 or 196608:
- OpenCL enqueue write error:-5 (CL_OUT_OF_RESOURCES)
This was alleviated when I upgraded the driver to version 32.0.101.6989.
I want to confirm, what are the known issues in the driver version in question? What is the reason for the crash? How to properly handle or walkaround errors instead of crashing?
Thanks!
Link Copied
0 Replies

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