From aziabari at soe.ucsc.edu  Wed Mar  9 17:36:52 2011
From: aziabari at soe.ucsc.edu (Amirkoushyar Ziabari)
Date: Wed, 9 Mar 2011 14:36:52 -0800 (PST)
Subject: [Hotspot] Leakage model
In-Reply-To: <AANLkTi=wt0os9fC=dwoR_2SYuJiZGLUCsL1rg4VADvnR@mail.gmail.com>
Message-ID: <146637792.49486.1299710212701.JavaMail.root@mail-01.cse.ucsc.edu>

Dear all,

I was wondering if HotSpot is able to model 3D chips with thermal vias? 

Best,

Amirkoushyar Ziabari
Graduate Research Assistant
Electrical Engineering Department
Baskin School of Engineering
University of California, Santa Cruz

Mail Stop:
SOE3-Grads
Baskin School of Engineering
University of California, Santa Cruz
1156 High Street, Santa Cruz
CA95064

Telephone: +1 (831) 459-1292
Email: aziabari at soe.ucsc.edu

----- Original Message -----
From: "Wei Huang" <wh6p at virginia.edu>
To: "haha haha" <baikeina2 at gmail.com>
Cc: hotspot at mail.cs.virginia.edu
Sent: Tuesday, February 22, 2011 7:51:10 PM
Subject: Re: [Hotspot] Leakage model


Hi, We don't have up to date numbers for the mentioned leakage model. But you can try to use ITRS projection to estimate alpha and beta for 45nm. 


-Wei 


On Tue, Feb 22, 2011 at 8:34 PM, haha haha < baikeina2 at gmail.com > wrote: 


Hi all, 
I'm a Ph.D student in Zhejiang University. 
In the Hotspot there is a leakage power model, Pl = Area* alpha *exp(beta*(T-Tbase)). The values of parameter alpha and beta are for 180nm. Does anyone have the values for 45nm ? Or if anyone has another leakage power model, could you please provide the values of the parameters ? 
We don't have a SPICE simulator yet. we just wanna perform some high level estimations about the temperature dependent leakage power. 
Thanks very much 
Best 
Xiaohang 
_______________________________________________ 
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

From klmacdonald at gmail.com  Sun Mar 13 13:53:14 2011
From: klmacdonald at gmail.com (Kevin Macdonald)
Date: Sun, 13 Mar 2011 11:53:14 -0700
Subject: [Hotspot] possible bug in temperature_grid.c
Message-ID: <AANLkTi=tLDEvDfiKmdKjWCnnPK_jUWrMBB5nbS8ZksgX@mail.gmail.com>

In the function slope_fn_pack() on line 2686 of temperature_grid.c, I
believe the line
    psum += (A3D(v,subidx,nr-1,j,nl,nr,nc) - x[SOLDER_S]);
should be
    psum += (A3D(v,subidx,nr-1,j,nl,nr,nc) - x[SUB_S]);
I'm just making this determination based upon pattern matching rather
than a deep understanding of what's going on in the code, so I could
be wrong.
-Kevin Macdonald

From wh6p at virginia.edu  Thu Mar 17 09:28:48 2011
From: wh6p at virginia.edu (Wei Huang)
Date: Thu, 17 Mar 2011 09:28:48 -0500
Subject: [Hotspot] Leakage model
In-Reply-To: <146637792.49486.1299710212701.JavaMail.root@mail-01.cse.ucsc.edu>
References: <AANLkTi=wt0os9fC=dwoR_2SYuJiZGLUCsL1rg4VADvnR@mail.gmail.com>
	<146637792.49486.1299710212701.JavaMail.root@mail-01.cse.ucsc.edu>
Message-ID: <AANLkTi=3_=ev_x1rGXwkSEJ2vyseVv5Hh6oC2_y2v5fn@mail.gmail.com>

Hi, Right now, HotSpot doesn't model different materials inside one layer.
If you are interested in the average effect of adding thermal vias (or TSVs
in general), you may want to try average thermal properties. On the other
hand, if you are interested in modeling individual vias, the tool can't do
that at this stage.

Hope this helps.

-Wei

On Wed, Mar 9, 2011 at 4:36 PM, Amirkoushyar Ziabari
<aziabari at soe.ucsc.edu>wrote:

> Dear all,
>
> I was wondering if HotSpot is able to model 3D chips with thermal vias?
>
> Best,
>
> Amirkoushyar Ziabari
> Graduate Research Assistant
> Electrical Engineering Department
> Baskin School of Engineering
> University of California, Santa Cruz
>
> Mail Stop:
> SOE3-Grads
> Baskin School of Engineering
> University of California, Santa Cruz
> 1156 High Street, Santa Cruz
> CA95064
>
> Telephone: +1 (831) 459-1292
> Email: aziabari at soe.ucsc.edu
>
> ----- Original Message -----
> From: "Wei Huang" <wh6p at virginia.edu>
> To: "haha haha" <baikeina2 at gmail.com>
> Cc: hotspot at mail.cs.virginia.edu
> Sent: Tuesday, February 22, 2011 7:51:10 PM
> Subject: Re: [Hotspot] Leakage model
>
>
> Hi, We don't have up to date numbers for the mentioned leakage model. But
> you can try to use ITRS projection to estimate alpha and beta for 45nm.
>
>
> -Wei
>
>
> On Tue, Feb 22, 2011 at 8:34 PM, haha haha < baikeina2 at gmail.com > wrote:
>
>
> Hi all,
> I'm a Ph.D student in Zhejiang University.
> In the Hotspot there is a leakage power model, Pl = Area* alpha
> *exp(beta*(T-Tbase)). The values of parameter alpha and beta are for 180nm.
> Does anyone have the values for 45nm ? Or if anyone has another leakage
> power model, could you please provide the values of the parameters ?
> We don't have a SPICE simulator yet. we just wanna perform some high level
> estimations about the temperature dependent leakage power.
> Thanks very much
> Best
> Xiaohang
> _______________________________________________
> 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/20110317/21494c80/attachment.html 

From wh6p at virginia.edu  Thu Mar 17 13:13:52 2011
From: wh6p at virginia.edu (Wei Huang)
Date: Thu, 17 Mar 2011 13:13:52 -0500
Subject: [Hotspot] possible bug in temperature_grid.c
In-Reply-To: <AANLkTi=tLDEvDfiKmdKjWCnnPK_jUWrMBB5nbS8ZksgX@mail.gmail.com>
References: <AANLkTi=tLDEvDfiKmdKjWCnnPK_jUWrMBB5nbS8ZksgX@mail.gmail.com>
Message-ID: <AANLkTin4mT+tnu5dt6OmtnF46qPgZ5ZTnjPMADEEhkMW@mail.gmail.com>

Hi Kevin,

I think you are right. That was a good catch. Thank you. We are fixing it
and will update the website.

-Wei

On Sun, Mar 13, 2011 at 1:53 PM, Kevin Macdonald <klmacdonald at gmail.com>wrote:

> In the function slope_fn_pack() on line 2686 of temperature_grid.c, I
> believe the line
>    psum += (A3D(v,subidx,nr-1,j,nl,nr,nc) - x[SOLDER_S]);
> should be
>    psum += (A3D(v,subidx,nr-1,j,nl,nr,nc) - x[SUB_S]);
> I'm just making this determination based upon pattern matching rather
> than a deep understanding of what's going on in the code, so I could
> be wrong.
> -Kevin Macdonald
> _______________________________________________
> 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/20110317/17220744/attachment.html 

From mamsadegh at googoolia.com  Thu Mar 17 18:15:03 2011
From: mamsadegh at googoolia.com (MohammadSadegh Sadri)
Date: Thu, 17 Mar 2011 19:15:03 -0400
Subject: [Hotspot] Different Grid Resolutions for Each Block
Message-ID: <4D8295F7.4080802@googoolia.com>

Hi,

I'm new to HotSpot,

In HotSpot documentation it is mentioned that we can have different grid 
resolution for each block of a layer.

now I'm wondering how can i do this when performing the simulation in 
grid mode?

Can I have different grid resolution for blocks of one single layer or 
different resolution between multiple layers?

here is the portion of code related to this topic I think,
As we can see, grid rows and cols, are defined as properties of a layer, 
and not a block.

(temperature_grid.h)
typedef struct grid_model_t_st
{
     /* configuration    */
     thermal_config_t config;

     /* layer information    */
     layer_t *layers;
     int n_layers;

     /* grid resolution    */
     int rows;
     int cols;
     /* dimensions    */
     double width;
     double height;
     ....

Best Regards,
MohammadSadegh Sadri.


From mamsadegh at googoolia.com  Sun Mar 20 06:39:10 2011
From: mamsadegh at googoolia.com (MohammadSadegh Sadri)
Date: Sun, 20 Mar 2011 07:39:10 -0400
Subject: [Hotspot] HotSpot Block Mode vs. Grid Mode
Message-ID: <4D85E75E.7070807@googoolia.com>

Hi,

I have integrated HotSpot into my power simulation engine.
HotSpot is called regularly with power trace passed to it to perform 
temperature calculations of the simple floorplan that i have made.

I use default HotSpot config structures and I just defined the floorplan 
myself.

initial temperature of all units is 300 and ambient temperature is the 
default 273+45.

Performing simulations in block mode, I achieve reasonable results for 
temperatures (after 1000 simulation iterations) :
P_0    316.09
L2_0    314.37
L1D_0    321.50
L1I_0    300.87
... (P is the cpu core and L is the cache ... )

Now, I just change the simulation mode from BLOCK_MODE to GRID_MODE, 
every thing else is the same (power trance, floorplan and initial temp.) 
, this time the temperature values are completely unusual and I believe 
they are wrong:

P_0    269.72
L2_0    269.87
L1D_0    287.98
L1I_0    280.90
...

This shows the chip is getting colder! and the interesting point is that 
the ambient temperature is 318 so even if I pass all zeros as power 
consumption of different units of the chip, again the temperatures 
should go up toward the ambient temperature.

Am I right?
Can you guess what is wrong in the simulations?

Regards,
MohammadSadegh Sadri.



From wh6p at virginia.edu  Mon Mar 21 15:29:49 2011
From: wh6p at virginia.edu (Wei Huang)
Date: Mon, 21 Mar 2011 13:29:49 -0700
Subject: [Hotspot] HotSpot Block Mode vs. Grid Mode
In-Reply-To: <4D85E75E.7070807@googoolia.com>
References: <4D85E75E.7070807@googoolia.com>
Message-ID: <AANLkTikU_-JBmccNVRkjCp1X2oFhiuoLktbPba1jqKgJ@mail.gmail.com>

I would recommend to do the following:
1) try the steps in "How-To" session of the HotSpot website with default
files/traces, to make sure you have a clean copy of HotSpot.
2) try to check if your floorplan file has any white spaces and/or
overlapping blocks. HotSpot requires all spaces are defined in the floorplan
file without overlap. Otherwise, it will give unexpected results. These are
mentioned in the documentation.
3) Initial temperature is better to be higher than ambient, which is the
most common case.

Hope this helps.

-Wei

On Sun, Mar 20, 2011 at 4:39 AM, MohammadSadegh Sadri <
mamsadegh at googoolia.com> wrote:

> Hi,
>
> I have integrated HotSpot into my power simulation engine.
> HotSpot is called regularly with power trace passed to it to perform
> temperature calculations of the simple floorplan that i have made.
>
> I use default HotSpot config structures and I just defined the floorplan
> myself.
>
> initial temperature of all units is 300 and ambient temperature is the
> default 273+45.
>
> Performing simulations in block mode, I achieve reasonable results for
> temperatures (after 1000 simulation iterations) :
> P_0    316.09
> L2_0    314.37
> L1D_0    321.50
> L1I_0    300.87
> ... (P is the cpu core and L is the cache ... )
>
> Now, I just change the simulation mode from BLOCK_MODE to GRID_MODE,
> every thing else is the same (power trance, floorplan and initial temp.)
> , this time the temperature values are completely unusual and I believe
> they are wrong:
>
> P_0    269.72
> L2_0    269.87
> L1D_0    287.98
> L1I_0    280.90
> ...
>
> This shows the chip is getting colder! and the interesting point is that
> the ambient temperature is 318 so even if I pass all zeros as power
> consumption of different units of the chip, again the temperatures
> should go up toward the ambient temperature.
>
> Am I right?
> Can you guess what is wrong in the simulations?
>
> Regards,
> MohammadSadegh Sadri.
>
>
> _______________________________________________
> 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/20110321/5b2e087e/attachment.html 

From Runjie at virginia.edu  Mon Mar 21 15:45:25 2011
From: Runjie at virginia.edu (Runjie Zhang)
Date: Mon, 21 Mar 2011 16:45:25 -0400
Subject: [Hotspot] Different Grid Resolutions for Each Block
In-Reply-To: <4D8295F7.4080802@googoolia.com>
References: <4D8295F7.4080802@googoolia.com>
Message-ID: <AANLkTinr_ZxQzBYO4a7PrA1DcS9C1tqqBGhGj71FOPYn@mail.gmail.com>

So far, HotSpot grid mode requires uniform resolution across all the layers.
Non-uniform grid resolution is one of our proposed ideas and might be
released in future versions.

Thanks

On Thu, Mar 17, 2011 at 7:15 PM, MohammadSadegh Sadri <
mamsadegh at googoolia.com> wrote:

> Hi,
>
> I'm new to HotSpot,
>
> In HotSpot documentation it is mentioned that we can have different grid
> resolution for each block of a layer.
>
> now I'm wondering how can i do this when performing the simulation in
> grid mode?
>
> Can I have different grid resolution for blocks of one single layer or
> different resolution between multiple layers?
>
> here is the portion of code related to this topic I think,
> As we can see, grid rows and cols, are defined as properties of a layer,
> and not a block.
>
> (temperature_grid.h)
> typedef struct grid_model_t_st
> {
>     /* configuration    */
>     thermal_config_t config;
>
>     /* layer information    */
>     layer_t *layers;
>     int n_layers;
>
>     /* grid resolution    */
>     int rows;
>     int cols;
>     /* dimensions    */
>     double width;
>     double height;
>     ....
>
> Best Regards,
> MohammadSadegh Sadri.
>
> _______________________________________________
> 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/20110321/461e3142/attachment.html 

From a9051224 at st2.fju.edu.tw  Tue Mar 22 11:11:14 2011
From: a9051224 at st2.fju.edu.tw (Joan Chang)
Date: Wed, 23 Mar 2011 00:11:14 +0800
Subject: [Hotspot] thermal conductance
Message-ID: <AANLkTinDE9oHVDbjpYMtNmkZUqXzPLarC4vyeLfY-YVE@mail.gmail.com>

Hi,

In some papers, they mentioned that thermal conductance related with
temperature. But, I didn't find the equation to model the relationship
between thermal conductance and temperature  while I traced source code.
Thank you!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20110323/9521fd64/attachment.html 

From skadron at cs.virginia.edu  Tue Mar 22 11:18:50 2011
From: skadron at cs.virginia.edu (Kevin Skadron)
Date: Tue, 22 Mar 2011 12:18:50 -0400
Subject: [Hotspot] thermal conductance
In-Reply-To: <AANLkTinDE9oHVDbjpYMtNmkZUqXzPLarC4vyeLfY-YVE@mail.gmail.com>
References: <AANLkTinDE9oHVDbjpYMtNmkZUqXzPLarC4vyeLfY-YVE@mail.gmail.com>
Message-ID: <4D88CBEA.6090507@cs.virginia.edu>

You're right; this is a feature that's on our to-do list but hasn't yet 
been implemented.

--Kevin

On 3/22/11 12:11 PM, Joan Chang wrote:
> Hi,
>
> In some papers, they mentioned that thermal conductance related with
> temperature. But, I didn't find the equation to model the relationship
> between thermal conductance and temperature  while I traced source code.
> Thank you!
>
>
>
> _______________________________________________
> HotSpot mailing list
> HotSpot at mail.cs.virginia.edu
> http://www.cs.virginia.edu/mailman/listinfo/hotspot

From bixiuyuan at gmail.com  Tue Mar 29 17:58:51 2011
From: bixiuyuan at gmail.com (Xiuyuan Bi)
Date: Tue, 29 Mar 2011 18:58:51 -0400
Subject: [Hotspot] About the LCF file in 3-D stacking chip
Message-ID: <7633CBBB-B7E2-427D-8A1F-A2665B4AD5E3@gmail.com>

Hi, everyone

	I am trying to use HotSpot to simulate a 3-D chip.
	I would like to know in the LCF files, like "example.lcf", which layer is the top layer? Layer 0 or Layer 3?
	Thanks a lot. 

Xiuyuan Bi

From Runjie at virginia.edu  Tue Mar 29 18:05:53 2011
From: Runjie at virginia.edu (Runjie Zhang)
Date: Tue, 29 Mar 2011 19:05:53 -0400
Subject: [Hotspot] About the LCF file in 3-D stacking chip
In-Reply-To: <7633CBBB-B7E2-427D-8A1F-A2665B4AD5E3@gmail.com>
References: <7633CBBB-B7E2-427D-8A1F-A2665B4AD5E3@gmail.com>
Message-ID: <AANLkTik7ir467ehevZH205vB2tbdKY7K9mgkRfJMkd7b@mail.gmail.com>

In that file, layer 3 is the top layer.

Runjie

On Tue, Mar 29, 2011 at 6:58 PM, Xiuyuan Bi <bixiuyuan at gmail.com> wrote:

> Hi, everyone
>
>        I am trying to use HotSpot to simulate a 3-D chip.
>        I would like to know in the LCF files, like "example.lcf", which
> layer is the top layer? Layer 0 or Layer 3?
>        Thanks a lot.
>
> Xiuyuan Bi
> _______________________________________________
> 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/20110329/1b674a48/attachment.html