From slogan at ucsc.edu  Thu Sep  3 19:31:41 2009
From: slogan at ucsc.edu (Sheldon Logan)
Date: Thu, 03 Sep 2009 19:31:41 -0700
Subject: [Hotspot] including power in C4 bumps and Solder Balls in Hotspot
	v5.0
Message-ID: <4AA07C0D.3070807@ucsc.edu>

I was wondering if it would be possible to assign power values to C-4 
bumps/solder balls in Hotspot v5.0. Our research is in chip package 
co-design, and we want to accurately simulate thermal effects at the 
package level.
                                                                         
                                    -Sheldon Logan

From wh6p at virginia.edu  Fri Sep  4 10:21:19 2009
From: wh6p at virginia.edu (Wei Huang)
Date: Fri, 04 Sep 2009 13:21:19 -0400
Subject: [Hotspot] including power in C4 bumps and Solder Balls in
 Hotspot	v5.0
In-Reply-To: <4AA07C0D.3070807@ucsc.edu>
References: <4AA07C0D.3070807@ucsc.edu>
Message-ID: <web-7124334@neon.mail.virginia.edu>

Sheldon,

There is a way to do that in HotSpot grid model with 3D chip configuration. 
Bascially you need to add a new layer in the layer config file to represent 
the C4 pads, and assign power in the power trace file. You also have to make 
a "floorplan" for the pads layer.

(I am also cc'ing our new group member, Runjie, who just did a similar 
experiment on HotSpot. Runjie, please comment if I missed anything.)

Regards,
-Wei

On Thu, 03 Sep 2009 19:31:41 -0700
  Sheldon Logan <slogan at ucsc.edu> wrote:
> I was wondering if it would be possible to assign power values to C-4 
> bumps/solder balls in Hotspot v5.0. Our research is in chip package 
> co-design, and we want to accurately simulate thermal effects at the 
> package level.
>                                                                         
>                                    -Sheldon Logan
> _______________________________________________
> HotSpot mailing list
> HotSpot at mail.cs.virginia.edu
> http://www.cs.virginia.edu/mailman/listinfo/hotspot


From zhwang at cise.ufl.edu  Fri Sep  4 12:26:50 2009
From: zhwang at cise.ufl.edu (Zhe Wang)
Date: Fri, 04 Sep 2009 15:26:50 -0400
Subject: [Hotspot] How does HotSpot compute the steady-state temperature?
Message-ID: <4AA169FA.4050103@cise.ufl.edu>

Hi,
     HotSpot is very useful to my research, great job. But I lack some 
background and want to know the physics equations to compute the 
steady-state. Can anyone give me some references?
     Thank you.

Zhe

From wh6p at virginia.edu  Fri Sep  4 13:12:33 2009
From: wh6p at virginia.edu (Wei Huang)
Date: Fri, 04 Sep 2009 16:12:33 -0400
Subject: [Hotspot] How does HotSpot compute the steady-state temperature?
In-Reply-To: <4AA169FA.4050103@cise.ufl.edu>
References: <4AA169FA.4050103@cise.ufl.edu>
Message-ID: <web-7105998@tetra.mail.virginia.edu>

Hi Zhe,

The physics behind HotSpot thermal modeling is based on two things: 
Fourier's Law for heat conduction within solids,  and some semi-analytical 
equations for convection heat transfer. We abstract those equations into a 
lumped model with thermal resistors and capacitors. You may want to refer to 
an introductory heat transfer book (I found the book by Yunus Cengel is very 
helpful).

Hope this helps.

-Wei

On Fri, 04 Sep 2009 15:26:50 -0400
  Zhe Wang <zhwang at cise.ufl.edu> wrote:
> Hi,
>     HotSpot is very useful to my research, great job. But I lack some 
> background and want to know the physics equations to compute the 
> steady-state. Can anyone give me some references?
>     Thank you.
> 
> Zhe
> _______________________________________________
> HotSpot mailing list
> HotSpot at mail.cs.virginia.edu
> http://www.cs.virginia.edu/mailman/listinfo/hotspot


From augusto.vega at bsc.es  Mon Sep  7 01:15:15 2009
From: augusto.vega at bsc.es (Augusto Vega)
Date: Mon, 07 Sep 2009 10:15:15 +0200
Subject: [Hotspot] Power model (Wattch or Cacti?)
Message-ID: <4AA4C113.9060004@bsc.es>

Hi all,

I'm performing some research using Hotspot and I've a basic question 
about the Wattch power model (I know it's the Hotspot mailing list, but 
because both power and thermal models are close related, maybe somebody 
can help me).

Wattch uses Cacti in order to get some cache (geometry) parameters which 
are passed to the power model. But I've also seen Cacti computes power 
dissipation (at least last versions 4.x and 5.x). Can I directly use 
Cacti to compute power values instead of using Wattch?

Thank you in advance.

Augusto Vega