The code for the current interconnect is in $GEMS/ruby/network/simple
The existing interconnect model is at a higher level than you imply.
There is no flit-level transmission. Each link uses a basic
latency/bandwidth queuing model. The switch is not modeled in detail.
Future versions of GEMS may contain a more detailed interconnect model.
pana wrote:
Hi,
I use Gems/Ruby with the MOESI_CMP_NUCA protocol. I have 8 cpu on a
single chip and a shared D-NUCA cache memory (tetris like clustered)
Where can i find in the code the low level features of the network
components?
I need for example to know which kind of links are used to connect the
switches, (i.e. how many wire for every link) , the kind of switch the
network is using, how the packet are managed on trasmission ( i.e. are
they separated in flit? how many packet for trasmission?)
thanks a lot for your help
greetings
francesco
------------------------------------------------------------------------
_______________________________________________
Gems-users mailing list
Gems-users@xxxxxxxxxxx
https://lists.cs.wisc.edu/mailman/listinfo/gems-users
Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/" to your search.
|