Mailing List Archives

Authenticated access

 		[Computer Systems Lab]

[← Prev]  [Date]  [Next →] [← Prev]  [Thread]  [Next →]

Re: [Gems-users] RING Topolgy

Date: Wed, 08 Apr 2009 01:02:53 -0400
From: niket <niketa@xxxxxxxxxxxxx>
Subject: Re: [Gems-users] RING Topolgy
GARNET currently uses deterministic routing. Adaptive routing can be readily modeled with a bit of effort. For example, the routing table at routers contain multiple output ports as choices for the same destination, if both are on the shortest path. Minimal adaptive routing can be implemented by choosing the output port based on some metric, e.g. link utilization. However, some mechanism would have to be implemented to take care of deadlocks.

Cheers,
Niket

Shoaib Altaf wrote:
 
Thanks & one more thing do GEMS support adaptive routing for the GARNET?
 I believe it supports adaptive routing for the Simple Network in the file... $GEMS/ruby/network /simple /perfectswitch.c & uses deterministic routing based on minimum path algorithm for the GARNET.
 
 
Regards
Muhammad Shoaib Bin Altaf

From: niket
Sent: Tuesday, April 07, 2009 11:36 PM
Subject: Re: [Gems-users] RING Topolgy

If you take a look at the file ruby/network/simple/Topology.C and in the function makeFileSpecified(), for every link, the otherDirectionNodes are also inserted in the nodePairs list. Removing the insertion of otherDirectionNodes should make the links uni-directional.

Cheers,
Niket

Shoaib Altaf wrote: 
But I am unable to locate the Files in which I can make such changes. Can 
you guide me in this regard?

Regards
Muhammad Shoaib


--------------------------------------------------
From: "Niket Agarwal" <niketa@xxxxxxxxxxxxx>
Sent: Monday, April 06, 2009 2:56 PM
To: "Gems Users" <gems-users@xxxxxxxxxxx>
Subject: Re: [Gems-users] RING Topolgy

  
Whenever a link is created, two corresponding links are created in
both directions. I think you can remove the reverse direction links
and things should work fine.

One thing you might want to take a look at is deadlocks. A ring can
have deadlocks if you simulate limited VCs and buffers. You might want
to implement some standard deadlock avoidance mechanism.

Cheers,
Niket

On Apr 6, 2009, at 3:12 PM, Muhammad Shoaib wrote:

    
Thanks Greg for the reply.

But can you guide me where should I look for if I want to make my
links to be unidirectional. Does the Routing algorithm needs to be
changed or some parametric changes???

Muhammad Shoaib


----- Original Message -----
From: Greg Byrd <gbyrd@xxxxxxxx>
Date: Saturday, April 4, 2009 8:53 pm
Subject: Re: [Gems-users] RING Topolgy
To: Gems Users <gems-users@xxxxxxxxxxx>


      
This is certainly a ring, but it won't be a unidirectional ring
unless
you change how routing is done.  The links are bidirectional, and the
routing algorithm generates a shortest-path table.  So a packet will
go in the direction that is the shortest path to its destination.

I don't know if you want a unidirectional ring.  If not, I think
you're done!

...Greg


Muhammad Shoaib wrote:
        
Hi All,

Will the following be a valid RING Topology, specified in a file
          
"RING_Procs-16_ProcsPerChip-16_L2Banks-16_Memories-16.txt"   ???
        
processors:16
procs_per_chip:16
L2banks:16
memories:16

ext_node:L1Cache:0 int_node:0 link_latency:1 bw_multiplier:72
ext_node:L1Cache:1 int_node:1 link_latency:1 bw_multiplier:72
ext_node:L1Cache:2 int_node:2 link_latency:1 bw_multiplier:72
ext_node:L1Cache:3 int_node:3 link_latency:1 bw_multiplier:72
ext_node:L1Cache:4 int_node:4 link_latency:1 bw_multiplier:72
ext_node:L1Cache:5 int_node:5 link_latency:1 bw_multiplier:72
ext_node:L1Cache:6 int_node:6 link_latency:1 bw_multiplier:72
ext_node:L1Cache:7 int_node:7 link_latency:1 bw_multiplier:72
ext_node:L1Cache:8 int_node:8 link_latency:1 bw_multiplier:72
ext_node:L1Cache:9 int_node:9 link_latency:1 bw_multiplier:72
ext_node:L1Cache:10 int_node:10 link_latency:1 bw_multiplier:72
ext_node:L1Cache:11 int_node:11 link_latency:1 bw_multiplier:72
ext_node:L1Cache:12 int_node:12 link_latency:1 bw_multiplier:72
ext_node:L1Cache:13 int_node:13 link_latency:1 bw_multiplier:72
ext_node:L1Cache:14 int_node:14 link_latency:1 bw_multiplier:72
ext_node:L1Cache:15 int_node:15 link_latency:1 bw_multiplier:72
ext_node:L2Cache:0 int_node:0 link_latency:1 bw_multiplier:72
ext_node:L2Cache:1 int_node:1 link_latency:1 bw_multiplier:72
ext_node:L2Cache:2 int_node:2 link_latency:1 bw_multiplier:72
ext_node:L2Cache:3 int_node:3 link_latency:1 bw_multiplier:72
ext_node:L2Cache:4 int_node:4 link_latency:1 bw_multiplier:72
ext_node:L2Cache:5 int_node:5 link_latency:1 bw_multiplier:72
ext_node:L2Cache:6 int_node:6 link_latency:1 bw_multiplier:72
ext_node:L2Cache:7 int_node:7 link_latency:1 bw_multiplier:72
ext_node:L2Cache:8 int_node:8 link_latency:1 bw_multiplier:72
ext_node:L2Cache:9 int_node:9 link_latency:1 bw_multiplier:72
ext_node:L2Cache:10 int_node:10 link_latency:1 bw_multiplier:72
ext_node:L2Cache:11 int_node:11 link_latency:1 bw_multiplier:72
ext_node:L2Cache:12 int_node:12 link_latency:1 bw_multiplier:72
ext_node:L2Cache:13 int_node:13 link_latency:1 bw_multiplier:72
ext_node:L2Cache:14 int_node:14 link_latency:1 bw_multiplier:72
ext_node:L2Cache:15 int_node:15 link_latency:1 bw_multiplier:72
ext_node:Directory:0 int_node:0 link_latency:20 bw_multiplier:80
ext_node:Directory:1 int_node:1 link_latency:20 bw_multiplier:80
ext_node:Directory:2 int_node:2 link_latency:20 bw_multiplier:80
ext_node:Directory:3 int_node:3 link_latency:20 bw_multiplier:80
ext_node:Directory:4 int_node:4 link_latency:20 bw_multiplier:80
ext_node:Directory:5 int_node:5 link_latency:20 bw_multiplier:80
ext_node:Directory:6 int_node:6 link_latency:20 bw_multiplier:80
ext_node:Directory:7 int_node:7 link_latency:20 bw_multiplier:80
ext_node:Directory:8 int_node:8 link_latency:20 bw_multiplier:80
ext_node:Directory:9 int_node:9 link_latency:20 bw_multiplier:80
ext_node:Directory:10 int_node:10 link_latency:20 bw_multiplier:80
ext_node:Directory:11 int_node:11 link_latency:20 bw_multiplier:80
ext_node:Directory:12 int_node:12 link_latency:20 bw_multiplier:80
ext_node:Directory:13 int_node:13 link_latency:20 bw_multiplier:80
ext_node:Directory:14 int_node:14 link_latency:20 bw_multiplier:80
ext_node:Directory:15 int_node:15 link_latency:20 bw_multiplier:80


int_node:0 int_node:1 link_latency:1 bw_multiplier:72
int_node:1 int_node:2 link_latency:1 bw_multiplier:72
int_node:2 int_node:3 link_latency:1 bw_multiplier:72
int_node:3 int_node:4 link_latency:1 bw_multiplier:72
int_node:4 int_node:5 link_latency:1 bw_multiplier:72
int_node:5 int_node:6 link_latency:1 bw_multiplier:72
int_node:6 int_node:7 link_latency:1 bw_multiplier:72
int_node:7 int_node:8 link_latency:1 bw_multiplier:72
int_node:8 int_node:9 link_latency:1 bw_multiplier:72
int_node:9 int_node:10 link_latency:1 bw_multiplier:72
int_node:10 int_node:11 link_latency:1 bw_multiplier:72
int_node:11 int_node:12 link_latency:1 bw_multiplier:72
int_node:12 int_node:13 link_latency:1 bw_multiplier:72
int_node:13 int_node:14 link_latency:1 bw_multiplier:72
int_node:14 int_node:15 link_latency:1 bw_multiplier:72
int_node:15 int_node:0 link_latency:1 bw_multiplier:72


Regards


_______________________________________________
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.
        
_______________________________________________
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.

        
_______________________________________________
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.

      
_______________________________________________
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.


    
_______________________________________________
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.

_______________________________________________
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.


_______________________________________________
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.

[← Prev in Thread] Current Thread [Next in Thread→]
Previous by Date:  Re: [Gems-users] RING Topolgy, Shoaib Altaf
Next by Date:  [Gems-users] Ruby Question, Mark Anderson
Previous by Thread:  Re: [Gems-users] RING Topolgy, Shoaib Altaf
Next by Thread:  [Gems-users] Ruby and TM system failing on 64 core serengethi, BYONG WU CHONG
Indexes:  [Date] [Thread]