Cannot run Opal example

[nhatdangncsu]

New Issue for sst-elements

1 - When I run opal example in SST element directory sst-elements/src/sst/elements/Opal/tests/basic_1node_1smp.py, it 's giving me the following error:

MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD with errorcode -1.

NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes. You may or may not see output from other processes, depending on exactly when Open MPI kills them.

2 - Command to reproduce: sst -v sst-elements/src/sst/elements/Opal/tests/basic_1node_1smp.py

3 - Operating system: Ubuntu 22.04

4 - MPI version 4.0.5 Can you help me check this issue? Thank you.

[nhatdangncsu]

Is merlin.Bridge removed? This is the beginning of my output.

WARNING: Component name 'Bridge:node0_internal_network-node0_Bridge' is not valid                                                                                                                         
 WARNING: Link name 'B0-node0_internal_network-node0_Bridge' is not valid                                                                                                                                  
 WARNING: Link name 'B0-node0_internal_network-node0_Bridge' is not valid                                                                                                                                  
 WARNING: Link name 'B1-node0_internal_network-node0_Bridge' is not valid                                                                                                                                  
 WARNING: Link name 'B1-node0_internal_network-node0_Bridge' is not valid                                                                                                                                  
 WARNING: Component name 'Bridge:Ext_Mem_Net-node0_Bridge' is not valid                                                                                                                                    
 WARNING: Link name 'B0-Ext_Mem_Net-node0_Bridge' is not valid                                                                                                                                             
 WARNING: Link name 'B0-Ext_Mem_Net-node0_Bridge' is not valid                                                                                                                                             
 WARNING: Link name 'B1-Ext_Mem_Net-node0_Bridge' is not valid                                                                                                                                             
 WARNING: Link name 'B1-Ext_Mem_Net-node0_Bridge' is not valid                                                                                                                                             
 WARNING: Component name 'Bridge:node1_internal_network-node1_Bridge' is not valid                                                                                                                         
 WARNING: Link name 'B0-node1_internal_network-node1_Bridge' is not valid                                                                                                                                  
 WARNING: Link name 'B0-node1_internal_network-node1_Bridge' is not valid                                                                                                                                  
 WARNING: Number of invalid link names exceeds limit of 10, no more messages will be printed                                                                                                               
 WARNING: Component name 'Bridge:Ext_Mem_Net-node1_Bridge' is not valid

[feldergast]

merlin.Bridge is still available. These are warning that the names being used are invalid. We instituted a naming convention, which essentially requires names to be valid python names (with a few tweaks) and that input file must not have been updated with new names that were compliant. None of those warnings would cause functional issues in the simulation.

[nhatdangncsu]

Ok. Did you run the above example file successfully? I still don't understand why it keeps giving me the above error.

[feldergast]

I'm not able to run that file on my development machine because I don't have PIN installed. This is not my area of expertise, but if I had to guess, I would guess that there is an issue with you PIN installation or version. I know that there have been some issues with PIN support recently, but I don't know the details.

[nhatdangncsu]

I don't think it 's a PIN problem because I can run stream example in ariel directory.

  496 LLSC: N Lock: (0,0)
   497 LLSC: N Lock: (0,0)
   498 LLSC: N Lock: (0,0)
   499 LLSC: N Lock: (0,0)
   500 LLSC: N Lock: (0,0)
   501 LLSC: N Lock: (0,0)
   502 LLSC: N Lock: (0,0)
   503 LLSC: N Lock: (0,0)
   504 LLSC: N Lock: (0,0)
   505 LLSC: N Lock: (0,0)
   506 LLSC: N Lock: (0,0)
   507 LLSC: N Lock: (0,0)
   508 LLSC: N Lock: (0,0)
   509 LLSC: N Lock: (0,0)
   510 LLSC: N Lock: (0,0)
   511 LLSC: N Lock: (0,0)
End MemHierarchy::Cache

  Checking for unreceived events on up link: 
  Checking for unreceived events on down link: 
--------------------------------------------------------------------------
MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD
with errorcode -1.

NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.
You may or may not see output from other processes, depending on
exactly when Open MPI kills them.
--------------------------------------------------------------------------

[nhatdangncsu]

I have been able to identify the line that cause the problem Line 240: PTWMemLink.connect((mmu, "ptw_to_mem%d"%(next_core-cores//2), "300ps"), (l1_cpulink, "port", "300ps")) I want to ask if page table walker in Samba needs to use ptw_to_mem port or not.