Re: [Gems-users] Network topology for TOKEN_CMP

["Mike Marty" <mikem@xxxxxxxxxxx>]

Trying using "NUCA" as the g_CACHE_DESIGN.  

The files you are referring to should have been removed (they slipped past
me), as the version of MOESI_CMP_token in the release does not use the
PersistentArbiters as you pointed out.

--Mike


> -----Original Message-----
> From: gems-users-bounces@xxxxxxxxxxx 
> [mailto:gems-users-bounces@xxxxxxxxxxx] On Behalf Of Ricardo 
> Fernández Pascual
> Sent: Tuesday, May 09, 2006 12:55 PM
> To: Gems Users
> Subject: [Gems-users] Network topology for TOKEN_CMP
> 
> 
> Hello,
> 
>   I am trying to simulate a 16 processors CMP with the 
> TokenCMP protocol using the supplied 
> "TOKEN_CMP_Procs-16_ProcsPerChip-16_L2Banks-16_Memories-16.txt
> " file. I get some assertion failures that confuse me. Using 
> the following
> configuration:
> 
> g_NETWORK_TOPOLOGY: "FILE_SPECIFIED"
> g_CACHE_DESIGN: "TOKEN_CMP"
> 
> I get:
> 
> failed assertion 'atoi((string_split(varStr, ':')).c_str()) 
> == g_endpoint_bandwidth' at fn void 
> Topology::makeFileSpecified() in network/simple/Topology.C:540
> 
> So, I decided to try adding the following (although I am not sure if I
> should):
> 
> g_endpoint_bandwidth": 1000
> 
> But, after that, I get a different assertion failure:
> 
> Warning: in fn MachineType string_to_MachineType(const 
> std::string&) in generated/MOESI_CMP_token/MachineType.C:40: 
> str is PersistentArbiter Fatal Error: in fn MachineType 
> string_to_MachineType(const std::string&) in 
> generated/MOESI_CMP_token/MachineType.C:41: Invalid string 
> conversion for type MachineType
> 
> The MOESI_CMP_token does not use a PersistentArbiter AFAIK, 
> but a PersistentTable. So, it seem that the topology file is 
> outdated, am I right? How could I fix it?
> 
> Thank you for your help.
> 
> Best regards,
> Ricardo.
> 
> -- 
> Ricardo Fernández Pascual
> r.fernandez@xxxxxxxxxxx
> Murcia. España.
> 
> _______________________________________________
> Gems-users mailing list
> Gems-users@xxxxxxxxxxx 
> https://lists.cs.wisc.edu/mailman/listinfo/gems-users
>