"aclopRegisterCompileFunc failed" Is Reported in Logs Due to On-Chip memory Stress Test Failure

Symptom

The on-chip memory stress test fails.

The version information about the device is as follows:

  • Driver: 23.0.5.1
  • Firmware: 7.1.0.8.220
  • Toolkit: 8.0.RC2
  • MindCluster ToolBox: 5.0.1
[root@***** ~]# ascend-dmi -c
|================================================================================================|
|                              System Information                                                |
|================================================================================================|
|  Architecture      |  aarch64                                                                  |
+------------------------------------------------------------------------------------------------+
|  Type              |  xxxxxxx                                                                  |
|================================================================================================|
|                       Compatibility Check Result: Compatible                                   |
|================================================================================================|
|  Package           |  Version          |  Status        |  Innerversion    |  Dependencies     |
+------------------------------------------------------------------------------------------------+
|  npu-driver        |  23.0.5.1         |  OK            |  xxxxxxx         |  NA               |
+------------------------------------------------------------------------------------------------+
|  npu-firmware      |  7.1.0.8.220      |  OK            |  NA              |  NA               |
+------------------------------------------------------------------------------------------------+
|  toolkit           |  8.0.RC2          |  OK            |  xxxxxxx         |  NA               |
+------------------------------------------------------------------------------------------------+
|  toolbox           |  5.0.1            |  OK            |  NA              |  NA               |
|================================================================================================|

Error information:

[ERROR]...[hbm_stress.cpp xxxxxx] aclopRegisterCompileFunc failed, error code: 100030, device: 4

Possible Causes

In the CANN package of version 8.0.RC2, the operator for the on-chip memory stress test has been upgraded, but it does not match the MindCluster ToolBox input of version 6.0.RC1 or earlier. As a result, the on-chip memory stress test fails.

Solution

Use CANN and MindCluster ToolBox of the matching versions, such as CANN 8.0.RC2 and MindCluster ToolBox 6.0.RC2 or later, to perform the on-chip memory stress test.