Closed lazyranma closed 11 years ago
I think Intel has figured out a new way to break OpenCL, they seem to be very good at this. In the meantime start DiabloMiner with -D 1, it should skip the crash and keep going.
I'm closing this bug until someone can confirm Intel continues to not comply with the OpenCL specification.
With Intel OpenCl SDK installed I get following exception: [22.04.12 17:50:13] Started(DiabloMiner.java:1218)
at com.diablominer.DiabloMiner.DiabloMiner.execute(DiabloMiner.java:572)
at com.diablominer.DiabloMiner.DiabloMiner.main(DiabloMiner.java:140)
[22.04.12 17:50:13] Connecting to: XXX [22.04.12 17:50:13] P2Pool no delay mode enabled
[22.04.12 17:50:13] Using NVIDIA CUDA OpenCL 1.1 CUDA 4.2.1
[22.04.12 17:50:13] Added GeForce GTX 460 (#1) (7 CU, local work size of 1024) [22.04.12 17:50:13] Using Intel(R) OpenCL OpenCL 1.1 LINUX
Exception in thread "main" java.lang.NullPointerException at com.diablominer.DiabloMiner.DiabloMiner$DeviceState.
Corresponding line from the source file (if line number is correct): deviceName = device.getInfoString(CL10.CL_DEVICE_NAME).trim() + " (#" + count + ")";
It could be that ether device is null or device.getInfoString() returns null. But when I use clGetDeviceInfo() from C API to query device name, it properly returns string. It could mean that device is null or there are some problems with java bindings.
Linux x86_64, Java(TM) SE Runtime Environment (build 1.6.0_26-b03), latest Intel OpenCL SDK 1.5.