[Gems-users] FILE_SPECIFIED Topology

[Edward Lee <edwl202@xxxxxxxxx>]

Hi,

I am trying to use the FILE_SPECIFIED topology for my experiments with the MOESI_SMP_directory protocol. If I try to use the sample network file "NUCA_Procs-16_ProcsPerChip-1_L2Banks-16_Memories-16.txt" for my 16 processor simulation, I am getting the following error:

failed assertion 'atoi((string_split(varStr, ':')).c_str()) == g_endpoint_bandwidth' at fn void Topology::makeFileSpecified() in network/simple/Topology.C:540
failed assertion 'atoi((string_split(varStr, ':')).c_str()) == g_endpoint_bandwidth' at fn void Topology::makeFileSpecified() in network/simple/Topology.C:540
At this point you might want to attach a debug to the running and get to the
crash site; otherwise press enter to continue
PID: 13257

Abort (SIGABRT) in main thread
The simulation state has been corrupted. Simulation cannot continue.
Please restart Simics.

If I try to use Garnet, and generate a topology file using the python script and changing the g_CACHE_DESIGN to MESH, accordingly using the "MESH_Procs-16_ProcsPerChip-1_L2Banks-16_Memories-16.txt" file I am getting the following error:

Warning: in fn MachineType string_to_MachineType(const std::string&) in generated/MOESI_SMP_directory/MachineType.C:36: str is L2Cache
Warning: in fn MachineType string_to_MachineType(const std::string&) in generated/MOESI_SMP_directory/MachineType.C:36: str is L2Cache
Fatal Error: in fn MachineType string_to_MachineType(const std::string&) in generated/MOESI_SMP_directory/MachineType.C:37: Invalid string conversion for type MachineType
Fatal Error: in fn MachineType string_to_MachineType(const std::string&) in generated/MOESI_SMP_directory/MachineType.C:37: Invalid string conversion for type MachineType
Abort (SIGABRT) in main thread
The simulation state has been corrupted. Simulation cannot continue.
Please restart Simics.

I saw an earlier thread (https://lists.cs.wisc.edu/archive/gems-users/2006-May/msg00011.shtml  ) similar to my problem here suggesting to use NUCA but apparently it didn't help as I tried that in the first case. So, is there anything not compatible with the SMP protocols in this topology files or topology file creation scripts?

And here is my script for the second case: (first one is similar rather than MESH --> NUCA, and comment out the Garnet part)

###################
istc-disable
dstc-disable
instruction-fetch-mode instruction-fetch-trace
magic-break-enable
break-hap "Core_Magic_Instruction"
cpu-switch-time 1

load-module ruby
ruby0.setparam g_NUM_PROCESSORS 16
ruby0.setparam g_MEMORY_SIZE_BYTES 4294967296
ruby0.setparam g_PROCS_PER_CHIP 1
ruby0.setparam g_NUM_L2_BANKS 16
ruby0.setparam L2_CACHE_NUM_SETS_BITS 12

# Network Parameters

ruby0.setparam_str g_NETWORK_TOPOLOGY FILE_SPECIFIED
ruby0.setparam_str g_CACHE_DESIGN MESH
ruby0.setparam_str g_GARNET_NETWORK true
ruby0.setparam_str g_DETAIL_NETWORK true
ruby0.setparam_str g_NETWORK_TESTING false
ruby0.setparam g_FLIT_SIZE 16
ruby0.setparam g_VCS_PER_CLASS 4
ruby0.setparam g_BUFFER_SIZE 4

ruby0.init

##################

Can somebody help me with this issue?

Regards,

Ed