Running something like a ring might induce corner cases, since most of
the Orion code was written for something like a 5 port router. I would
suggest printing out the different components of the router power,
like buffer, crossbar, etc in NetworkPower.C to see which component is
causing the trouble.
Also, do check whether the technology point, Vdd, frequency parameters
are what you want and are reasonable.
Cheers,
Niket
On Apr 16, 2009, at 12:35 AM, Shoaib Altaf wrote:
Niket,
Thanks fro the reply, but I forgot to mention that I was running the
simulation after initializing the variables in NetworkPower.C as well.
So, in that case, is there any other place to look for???
Regards
Muhammad Shoaib
--------------------------------------------------
From: "Niket Agarwal" <niketa@xxxxxxxxxxxxx>
Sent: Wednesday, April 15, 2009 11:20 PM
To: "Gems Users" <gems-users@xxxxxxxxxxx>
Subject: Re: [Gems-users] High Router power
Hi Shoaib,
Please also initialize the E_in_buf, E_xbar, etc variables in line
no. 352 of NetworkPower.C? These couple of errors have got in to the
release. Correcting this should solve your problem.
Our group is working on the release of Orion 2.0 which has further
corrections to the original Orion distribution. We will also plug it
into GEMS soon.
Cheers,
Niket
On Apr 16, 2009, at 12:13 AM, Muhammad Shoaib wrote:
Hi,
I am trying to simulate a CMP having a RING topology with 16
processors. The stats give me very high powers like
Router # = 0
Power = 581939 W
Router # = 1
Power = 634745 W
Router # = 2
Power = 645353 W
Router # = 3
Power = 773772 W
Router # = 4
Power = 711782 W
Router # = 5
Power = 801895 W
Router # = 6
Power = 657095 W
Router # = 7
Power = 612172 W
Router # = 8
Power = 610153 W
Router # = 9
Power = 577075 W
Router # = 10
Power = 544089 W
Router # = 11
Power = 548418 W
Router # = 12
Power = 549896 W
Router # = 13
Power = 535504 W
Router # = 14
Power = 559912 W
Router # = 15
Power = 542870 W
Total Link Power = 2.02815 W
Total Router Power = 4.94334e+06 W
I have already made the following changes in SIM_port.h that Niket
suggested in a thread to cater this sort of problem:
#define PARM_vc_in_arb_model RR_ARBITER
#define PARM_vc_out_arb_model RR_ARBITER
#define PARM_vc_in_arb_ff_model NEG_DFF
#define PARM_vc_out_arb_ff_model NEG_DFF
#define PARM_sw_in_arb_model RR_ARBITER
#define PARM_sw_out_arb_model RR_ARBITER
#define PARM_sw_in_arb_ff_model NEG_DFF
#define PARM_sw_out_arb_ff_model NEG_DFF
#define PARM_VC_per_MC 4
Any idea what can be the possible reason???
Muhammad Shoaib
_______________________________________________
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.
|