Re: [Gems-users] why router power is negative

[Niket Agarwal <niketa@xxxxxxxxxxxxx>]

Interesting .. that could be a case of some variable overflow. I haven't seen it in my runs. May be you can nail down the exact performance counter that is getting overflown and correct it.

-Niket

On 04/15/2010 10:47 AM, VLIW wrote:

Thanks Niket, but I have changed the code.

I think it maybe something wrong in 32-bit system.

because the result in 64-bit system is correct.

 

------------------

Fu GuiTao

 

 

 

------------------ Original ------------------

From:  "Niket Agarwal"<niketa@xxxxxxxxxxxxx>;

Date:  Thu, Apr 15, 2010 10:00 PM

To:  "gems-users"<gems-users@xxxxxxxxxxx>;

Subject:  Re: [Gems-users] why router power is negative

 

There are a couple of posts on why Orion 1.0 gives power in 1000s of W. You can search on the list. Try the suggestion below and see if it works for you.

https://lists.cs.wisc.edu/archive/gems-users/2008-November/msg00013.shtml

-Niket

On 04/15/2010 12:59 AM, VLIW wrote:

I reinstall the gems,  the result in amdx64 is normal, but the result in x86 is very high.

Router Power is thousands W. it's not correct.

the configuration is same. how to explain this phenomenon?

 

------------------

Fu GuiTao

 

 

 

------------------ Original ------------------

From:  "gt f"<fgtnudt@xxxxxxxxx>;

Date:  Tue, Apr 13, 2010 10:02 PM

To:  "Gems Users"<gems-users@xxxxxxxxxxx>;

Subject:  Re: [Gems-users] why router power is negative number£¿£¿

 

I found the mistake before in the wrong place. I get the SWITCHING_FATOR in correct value. and I found the nagative number cased by : arr->data_bitline_pre.e_charge, arr->data_amp.e_access and arr->row_dec.e_chg_l1. I don't know how to resolve this problem.

 

------------------

Fu GuiTao

 

 

 

------------------ Original ------------------

From:  "gt f"<fgtnudt@xxxxxxxxx>;

Date:  Tue, Apr 13, 2010 07:56 PM

To:  "Gems Users"<gems-users@xxxxxxxxxxx>;

Subject:  [Gems-users] why router power is negative number£¿£¿

 

Hi list:

I use the GEMS for a CMP system with 16 core TILED topolgy. I can get the power of link. but the power of router is nagative number (Total Router Power = -4.13954e+17W). I have changed the code in SIM_port.h file ( QUEUE_ARBITER repalced by RR_ARBITER), and initialized the E_in_buf, E_xbar, etc variables in line no. 352 in the NetworkPower.C file. I print the power of buffer, crossbar arbiter and static, I found that the nagative number form buffer (E_in_buf<0).

then  print the variables in SIM_reg_energy(power_array_info *info, power_array *arr, double n_read, double n write) in NetworkPower.C file. I think it maybe casuse by SWITCHING_FATOR. Because the resault of SWITCHING_FACTOR is not 0.5 but very big (about 10^9).

So can anyone help me, why does this happen ? my host system is AMD x64 and OS is DEBIN 5.

 

 

------------------

Fu GuiTao

 

 

 

_______________________________________________
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.