From ee09b004 at iith.ac.in  Thu Aug 16 14:02:53 2012
From: ee09b004 at iith.ac.in (AFTAB ALAM KHAN)
Date: Thu, 16 Aug 2012 14:02:53 -0700
Subject: [Hotspot] hotspot
Message-ID: <CANzAW2dKtz67P=o=Jm56nH1f+mkt1rXXjSA6zi0V=YtWPW0xKw@mail.gmail.com>

Hello

I was about to simulate DSP (digital signal processor) in hotspot to learn
its temperature profile nature,
but before that I need average power values of the individual blocks used
in the making of DSP.
So sir it would be great help if you can tell me how to get these average
power values.


thanks
Aftab Alam Khan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20120816/d2fbf8ba/attachment.html 

From skadron at cs.virginia.edu  Thu Aug 16 14:17:33 2012
From: skadron at cs.virginia.edu (Kevin Skadron)
Date: Thu, 16 Aug 2012 17:17:33 -0400
Subject: [Hotspot] hotspot
In-Reply-To: <CANzAW2dKtz67P=o=Jm56nH1f+mkt1rXXjSA6zi0V=YtWPW0xKw@mail.gmail.com>
References: <CANzAW2dKtz67P=o=Jm56nH1f+mkt1rXXjSA6zi0V=YtWPW0xKw@mail.gmail.com>
Message-ID: <502D636D.2060009@cs.virginia.edu>

Unfortunately, I don't know of an existing power model for any DSPs, but 
this is probably the easiest way.  An alternative is to synthesize a DSP 
core for your desired technology node and extract power.  Lacking these, 
you might be able to find a power breakdown for some DSP of interest in 
the literature.
--Kevin

On 8/16/2012 5:02 PM, AFTAB ALAM KHAN wrote:
> Hello
>
> I was about to simulate DSP (digital signal processor) in hotspot to
> learn its temperature profile nature,
> but before that I need average power values of the individual blocks
> used in the making of DSP.
> So sir it would be great help if you can tell me how to get these
> average power values.
>
>
> thanks
> Aftab Alam Khan
>
>
> _______________________________________________
> HotSpot mailing list
> HotSpot at mail.cs.virginia.edu
> http://www.cs.virginia.edu/mailman/listinfo/hotspot

From smilingsantanu at gmail.com  Fri Aug 17 00:45:45 2012
From: smilingsantanu at gmail.com (Santanu Sarma)
Date: Fri, 17 Aug 2012 00:45:45 -0700
Subject: [Hotspot] hotspot
In-Reply-To: <502D636D.2060009@cs.virginia.edu>
References: <CANzAW2dKtz67P=o=Jm56nH1f+mkt1rXXjSA6zi0V=YtWPW0xKw@mail.gmail.com>
	<502D636D.2060009@cs.virginia.edu>
Message-ID: <CANq=v5ccUdEV0kpTrTLRFEd0Z6efCNDU5jA=G7c4P-j2f97R2A@mail.gmail.com>

Dear Prof. Skadron and HotSpot Group,
  I am new to the Hotspot simulator and would like to use and extend it for
multi/many core architectures . I am trying to understand the underlying
LTI State Space model (ignoring the nonlinear temperature dependent leakage
power) and have few queries regarding the dimensions of the matrices in the
LTI model. Specifically, when the hostspot LTI state space model is
represented as
d
-- (T_i) = A*T_i + B*P_i
dt

I would like to know if the following dimensions for the ev6 floor plan in
Hotspot 5.02 is correct:
N=30 ( no of locks in the die as in the ev6.flp file)
M=30 (No of Blocks/ Nodes in the  thermal interface material (TIM))
L= 5 (No of Blocks/ Nodes in the  Heat Spreader(HS))
O=9 (No of Blocks/ Nodes in the  Heat Sink )

So the dimension of A in the above LTI State space model will be 74x74
(30+30+5+9=74 ) and B is 74x74. T_i , the temperature of each block/nodes
is 74x1 and the power P_i consumed at each node is 74x1. P_i has only N
non-zero elements i.e. 30 power producing blocks in the die.


Thanks
Santanu

On Thu, Aug 16, 2012 at 2:17 PM, Kevin Skadron <skadron at cs.virginia.edu>wrote:

> Unfortunately, I don't know of an existing power model for any DSPs, but
> this is probably the easiest way.  An alternative is to synthesize a DSP
> core for your desired technology node and extract power.  Lacking these,
> you might be able to find a power breakdown for some DSP of interest in
> the literature.
> --Kevin
>
> On 8/16/2012 5:02 PM, AFTAB ALAM KHAN wrote:
> > Hello
> >
> > I was about to simulate DSP (digital signal processor) in hotspot to
> > learn its temperature profile nature,
> > but before that I need average power values of the individual blocks
> > used in the making of DSP.
> > So sir it would be great help if you can tell me how to get these
> > average power values.
> >
> >
> > thanks
> > Aftab Alam Khan
> >
> >
> > _______________________________________________
> > HotSpot mailing list
> > HotSpot at mail.cs.virginia.edu
> > http://www.cs.virginia.edu/mailman/listinfo/hotspot
> _______________________________________________
> HotSpot mailing list
> HotSpot at mail.cs.virginia.edu
> http://www.cs.virginia.edu/mailman/listinfo/hotspot
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20120817/8b6fc2d4/attachment-0001.html 

From wh6p at virginia.edu  Fri Aug 17 06:58:55 2012
From: wh6p at virginia.edu (Wei Huang)
Date: Fri, 17 Aug 2012 08:58:55 -0500
Subject: [Hotspot] hotspot
In-Reply-To: <CANq=v5ccUdEV0kpTrTLRFEd0Z6efCNDU5jA=G7c4P-j2f97R2A@mail.gmail.com>
References: <CANzAW2dKtz67P=o=Jm56nH1f+mkt1rXXjSA6zi0V=YtWPW0xKw@mail.gmail.com>
	<502D636D.2060009@cs.virginia.edu>
	<CANq=v5ccUdEV0kpTrTLRFEd0Z6efCNDU5jA=G7c4P-j2f97R2A@mail.gmail.com>
Message-ID: <CANYeXBorCA2SR0Ucbbw4BZkiREqXJMBXOEPDTXZAA=3-VCO07w@mail.gmail.com>

Hi, I believe that heat spreader and heat sink nodes are N+4 and N+8,
respectively, in latest HotSpot release. N is the number of blocks in your
silicon floorplan.

-Wei

On Fri, Aug 17, 2012 at 2:45 AM, Santanu Sarma <smilingsantanu at gmail.com>wrote:

> Dear Prof. Skadron and HotSpot Group,
>   I am new to the Hotspot simulator and would like to use and extend it
> for multi/many core architectures . I am trying to understand the
> underlying LTI State Space model (ignoring the nonlinear temperature
> dependent leakage power) and have few queries regarding the dimensions of
> the matrices in the LTI model. Specifically, when the hostspot LTI state
> space model is represented as
> d
> -- (T_i) = A*T_i + B*P_i
> dt
>
> I would like to know if the following dimensions for the ev6 floor plan in
> Hotspot 5.02 is correct:
> N=30 ( no of locks in the die as in the ev6.flp file)
> M=30 (No of Blocks/ Nodes in the  thermal interface material (TIM))
> L= 5 (No of Blocks/ Nodes in the  Heat Spreader(HS))
> O=9 (No of Blocks/ Nodes in the  Heat Sink )
>
> So the dimension of A in the above LTI State space model will be 74x74
> (30+30+5+9=74 ) and B is 74x74. T_i , the temperature of each block/nodes
> is 74x1 and the power P_i consumed at each node is 74x1. P_i has only N
> non-zero elements i.e. 30 power producing blocks in the die.
>
>
> Thanks
> Santanu
>
>
> On Thu, Aug 16, 2012 at 2:17 PM, Kevin Skadron <skadron at cs.virginia.edu>wrote:
>
>> Unfortunately, I don't know of an existing power model for any DSPs, but
>> this is probably the easiest way.  An alternative is to synthesize a DSP
>> core for your desired technology node and extract power.  Lacking these,
>> you might be able to find a power breakdown for some DSP of interest in
>> the literature.
>> --Kevin
>>
>> On 8/16/2012 5:02 PM, AFTAB ALAM KHAN wrote:
>> > Hello
>> >
>> > I was about to simulate DSP (digital signal processor) in hotspot to
>> > learn its temperature profile nature,
>> > but before that I need average power values of the individual blocks
>> > used in the making of DSP.
>> > So sir it would be great help if you can tell me how to get these
>> > average power values.
>> >
>> >
>> > thanks
>> > Aftab Alam Khan
>> >
>> >
>> > _______________________________________________
>> > HotSpot mailing list
>> > HotSpot at mail.cs.virginia.edu
>> > http://www.cs.virginia.edu/mailman/listinfo/hotspot
>> _______________________________________________
>> HotSpot mailing list
>> HotSpot at mail.cs.virginia.edu
>> http://www.cs.virginia.edu/mailman/listinfo/hotspot
>>
>
>
> _______________________________________________
> HotSpot mailing list
> HotSpot at mail.cs.virginia.edu
> http://www.cs.virginia.edu/mailman/listinfo/hotspot
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20120817/09fd62a0/attachment.html 

From waqasch86 at gmail.com  Wed Aug 22 06:25:47 2012
From: waqasch86 at gmail.com (waqas ch)
Date: Wed, 22 Aug 2012 15:25:47 +0200
Subject: [Hotspot] Thermal Vias
Message-ID: <CAAr8icKXtzFpNVj_Wsq7GxRd-Mi_Zb3Syu+5nzmCG4VL3+6Vig@mail.gmail.com>

Hi

I am trying to run a 3D simulation and i need some hint on how can i change
individually the thermal resistances of Copper Vias. Any help in this
regard would be appreciated.

Regards

Waqas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20120822/bfa6fc66/attachment.html 

From skadron at cs.virginia.edu  Mon Aug 27 06:08:35 2012
From: skadron at cs.virginia.edu (Kevin Skadron)
Date: Mon, 27 Aug 2012 09:08:35 -0400
Subject: [Hotspot] Thermal Vias
In-Reply-To: <CAAr8icKXtzFpNVj_Wsq7GxRd-Mi_Zb3Syu+5nzmCG4VL3+6Vig@mail.gmail.com>
References: <CAAr8icKXtzFpNVj_Wsq7GxRd-Mi_Zb3Syu+5nzmCG4VL3+6Vig@mail.gmail.com>
Message-ID: <503B7153.8090807@cs.virginia.edu>

HotSpot can model 3D stacks of different materials, but assumes that 
each layer of material is homogeneous, and hence cannot model vias. 
However, a 3D extension by Prof. Coskun, linked from the HotSpot 
website, can help with this.
--Kevin

On 8/22/12 9:25 AM, waqas ch wrote:
> Hi
>
> I am trying to run a 3D simulation and i need some hint on how can i
> change individually the thermal resistances of Copper Vias. Any help in
> this regard would be appreciated.
>
> Regards
>
> Waqas
>
>
> _______________________________________________
> HotSpot mailing list
> HotSpot at mail.cs.virginia.edu
> http://www.cs.virginia.edu/mailman/listinfo/hotspot
>

From sthannirmalai at hotmail.com  Wed Aug 29 19:37:53 2012
From: sthannirmalai at hotmail.com (thannir malai)
Date: Thu, 30 Aug 2012 02:37:53 +0000
Subject: [Hotspot] steady state temperature for heat sink
Message-ID: <BAY163-W39366E64AEFE8F16EAA521B2A70@phx.gbl>


Hi all,
Just my initializing the simulation with ambient temperature, will the heat sink and other components will ever attain the steady state temperature calculated by the hotspot (compute_steady_state)?
Thanks   		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20120830/e60ce45b/attachment.html 

From skadron at cs.virginia.edu  Thu Aug 30 05:51:48 2012
From: skadron at cs.virginia.edu (Kevin Skadron)
Date: Thu, 30 Aug 2012 08:51:48 -0400
Subject: [Hotspot] steady state temperature for heat sink
In-Reply-To: <BAY163-W39366E64AEFE8F16EAA521B2A70@phx.gbl>
References: <BAY163-W39366E64AEFE8F16EAA521B2A70@phx.gbl>
Message-ID: <503F61E4.2080804@cs.virginia.edu>

Yes, assuming steady power dissipation commensurate with what you used 
in the steady-state calculation -- but reaching the steady state from 
ambient initial conditions may take an extremely long time, probably 
much longer than you can afford to wait.  It is better to initialize 
with temperatures close to the values you would see during sustained 
operation with an expected workload--initializing with the steady state 
is one way.
/K

On 8/29/12 10:37 PM, thannir malai wrote:
> Hi all,
>
> Just my initializing the simulation with ambient temperature, will the
> heat sink and other components will ever attain the steady state
> temperature calculated by the hotspot (compute_steady_state)?
>
> Thanks
>
>
> _______________________________________________
> HotSpot mailing list
> HotSpot at mail.cs.virginia.edu
> http://www.cs.virginia.edu/mailman/listinfo/hotspot
>

From indirarawat at gmail.com  Thu Aug 30 21:19:18 2012
From: indirarawat at gmail.com (Indira Rawat)
Date: Fri, 31 Aug 2012 09:49:18 +0530
Subject: [Hotspot] (no subject)
Message-ID: <CAB6-a68Qcg4Cq0T86BEsscVh9c73Ynf1bPo2PHa=pGWSGrWMoQ@mail.gmail.com>

can we find out the transient temperatures using hotspot or does it
give only steady state values?
Indira

From Runjie at virginia.edu  Thu Aug 30 21:44:25 2012
From: Runjie at virginia.edu (Runjie Zhang)
Date: Fri, 31 Aug 2012 00:44:25 -0400
Subject: [Hotspot] (no subject)
In-Reply-To: <CAB6-a68Qcg4Cq0T86BEsscVh9c73Ynf1bPo2PHa=pGWSGrWMoQ@mail.gmail.com>
References: <CAB6-a68Qcg4Cq0T86BEsscVh9c73Ynf1bPo2PHa=pGWSGrWMoQ@mail.gmail.com>
Message-ID: <CAHxZ6YDPOYe7VcRVFhG38ZjjZUx0n3Eg1ZcH=E6nNNoJNn2WYg@mail.gmail.com>

Hi, Indira

  HotSpot is capable of both transient and steady state simulation.

Runjie

On Fri, Aug 31, 2012 at 12:19 AM, Indira Rawat <indirarawat at gmail.com>wrote:

> can we find out the transient temperatures using hotspot or does it
> give only steady state values?
> Indira
> _______________________________________________
> HotSpot mailing list
> HotSpot at mail.cs.virginia.edu
> http://www.cs.virginia.edu/mailman/listinfo/hotspot
>



-- 
Runjie Zhang
Computer Engineering
University of Virginia
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20120831/da592212/attachment.html 

From sthannirmalai at hotmail.com  Thu Aug 30 23:13:32 2012
From: sthannirmalai at hotmail.com (thannir malai)
Date: Fri, 31 Aug 2012 06:13:32 +0000
Subject: [Hotspot] steady state temperature for heat sink
In-Reply-To: <503F61E4.2080804@cs.virginia.edu>
References: <BAY163-W39366E64AEFE8F16EAA521B2A70@phx.gbl>,
	<503F61E4.2080804@cs.virginia.edu>
Message-ID: <BAY163-W4953D9B5EFDCA473ED4C32B2A60@phx.gbl>


Thanks for your response. 
How long is extremely long time? I tried running for couple of hours with a constant workload. The steady state temperature was never reached.

> Date: Thu, 30 Aug 2012 08:51:48 -0400
> From: skadron at cs.virginia.edu
> To: sthannirmalai at hotmail.com
> CC: hotspot at mail.cs.virginia.edu
> Subject: Re: [Hotspot] steady state temperature for heat sink
> 
> Yes, assuming steady power dissipation commensurate with what you used 
> in the steady-state calculation -- but reaching the steady state from 
> ambient initial conditions may take an extremely long time, probably 
> much longer than you can afford to wait.  It is better to initialize 
> with temperatures close to the values you would see during sustained 
> operation with an expected workload--initializing with the steady state 
> is one way.
> /K
> 
> On 8/29/12 10:37 PM, thannir malai wrote:
> > Hi all,
> >
> > Just my initializing the simulation with ambient temperature, will the
> > heat sink and other components will ever attain the steady state
> > temperature calculated by the hotspot (compute_steady_state)?
> >
> > Thanks
> >
> >
> > _______________________________________________
> > HotSpot mailing list
> > HotSpot at mail.cs.virginia.edu
> > http://www.cs.virginia.edu/mailman/listinfo/hotspot
> >
 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20120831/ed61a010/attachment-0001.html 

From wh6p at virginia.edu  Fri Aug 31 08:41:01 2012
From: wh6p at virginia.edu (Wei Huang)
Date: Fri, 31 Aug 2012 10:41:01 -0500
Subject: [Hotspot] steady state temperature for heat sink
In-Reply-To: <BAY163-W4953D9B5EFDCA473ED4C32B2A60@phx.gbl>
References: <BAY163-W39366E64AEFE8F16EAA521B2A70@phx.gbl>
	<503F61E4.2080804@cs.virginia.edu>
	<BAY163-W4953D9B5EFDCA473ED4C32B2A60@phx.gbl>
Message-ID: <CANYeXBq-6wqjOexd676ZFv-dz_ESy6k=7k1cLSvLTpNmMn7ncw@mail.gmail.com>

The actual run time to reach steady state temperature depends on the
sampling interval you set in the hotspot.config file. Imagine it takes at
least 10 minutes (600seconds) for the heat sink to physically heat up, and
you set your sampling interval to 100us, that means HotSpot needs to 6
million iterations. The time to execute each iteration also depends on
whether you are running in block mode or grid mode. It usually takes much
more than 2 hours, if not days.

Can you tell us why do you prefer such a run?

-Wei

On Fri, Aug 31, 2012 at 1:13 AM, thannir malai <sthannirmalai at hotmail.com>wrote:

>  Thanks for your response.
> How long is extremely long time? I tried running for couple of hours with
> a constant workload. The steady state temperature was never reached.
>
> > Date: Thu, 30 Aug 2012 08:51:48 -0400
> > From: skadron at cs.virginia.edu
> > To: sthannirmalai at hotmail.com
> > CC: hotspot at mail.cs.virginia.edu
> > Subject: Re: [Hotspot] steady state temperature for heat sink
>
> >
> > Yes, assuming steady power dissipation commensurate with what you used
> > in the steady-state calculation -- but reaching the steady state from
> > ambient initial conditions may take an extremely long time, probably
> > much longer than you can afford to wait. It is better to initialize
> > with temperatures close to the values you would see during sustained
> > operation with an expected workload--initializing with the steady state
> > is one way.
> > /K
> >
> > On 8/29! /12 10:37 PM, thannir malai wrote:
> > > Hi all,
> > >
> > > Just my initializing the simulation with ambient temperature, will the
> > > heat sink and other components will ever attain the steady state
> > > temperature calculated by the hotspot (compute_steady_state)?
> > >
> > > Thanks
> > >
> > >
> > > _______________________________________________
> > > HotSpot mailing list
> > > HotSpot at mail.cs.virginia.edu
> > > http://www.cs.virginia.edu/mailman/listinfo/hotspot
> > >
>
> _______________________________________________
> HotSpot mailing list
> HotSpot at mail.cs.virginia.edu
> http://www.cs.virginia.edu/mailman/listinfo/hotspot
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20120831/ee5d849e/attachment.html