Hi there

Just to recap:

Amy and Stacey are both identical Dell Optiplex GX270 Tower Systems with same BIOS, MoBo, Std air cooling, 512Mb RAM, 80Gb HDD, video etc apart from:

Amy - 3.2GHz P4 H/T (with H/T on and continually running 2nd unrelated CPDN model concurrently), 1Mb cache, 800MHz FSB, WinXP (Home) SP2.

Stacey - 3.0Ghz P4 H/T (with H/T on but not running any other DC programs during this experiment), 512kb cache, 800MHz FSB, WinXP (Pro) SP1.

Jayne - the name (for this experiment) given to geophi's AMD64 machine, further description will need to come from geophi.

Phase 3 Temperature/Time Trace
<img src="http://cpdn.tuxie.org/Pete_B/Combined_Data/Amy_Stacey_Jayne_Ph3_Temp_Time_Trace.png">

Not much to say about that, it's all too close to call.


Amy-Stacey Summer 2051 Global Temp Difference
<img src="http://cpdn.tuxie.org/Pete_B/Combined_Data/Amy_Stacey_2051_Global_Temp_Diff.png">


Amy-Jayne Summer 2051 Global Temp Difference
<img src="http://cpdn.tuxie.org/Pete_B/Combined_Data/Amy_Jayne_2051_Global_Temp_Diff.png">

In both cases, warmer and colder areas seem evenly distributed.


Amy-Stacey Summer 2065 Global Temp Difference
<img src="http://cpdn.tuxie.org/Pete_B/Combined_Data/Amy_Stacey_2065_Global_Temp_Diff.png">

Warmer and colder areas for Amy's World seem fairly evenly distributed.


Amy-Jayne Summer 2065 Global Temp Difference
<img src="http://cpdn.tuxie.org/Pete_B/Combined_Data/Amy_Jayne_2065_Global_Temp_Diff.png">

This time, much greater -ve (and a smaller area of +ve) temperature anomalies are present in Amy's Antarctic region. I don't know what the regional differences mean though in terms of different computation.

It would be interesting to see how a heavily overclocked machine handled this model compared to these 3 machines.

Pete

Hi Pete, thanks for all the work and throwing those graphics up there. Looks like Amy was about 1/2 degree cooler than the other two at the end of the model run. Not too much compared to how warm they all got. Stacy and Jayne were amazingly close on the time series. From your other graphics, it looks very much like the weather differed significantly on the seasonal time scale in some places, but the overall earth average was very close.

The specs for my Athlon64 are 3200+ Clawhammer (2.0 GHz, 1 MB L2 cache), 512 MB PC3200 DDR at 200/400 MHz, WinXP (Home) SP2.

This is really interesting experiment.
Comparing Intel P4 Northwood (512kb cache) with Intel Prescott (1MB cache) and AMD64 Clawhammer gives slightly different results on average temperature; seasonal temperature maps shows larger difference.
There are a lot of other CPU types to run (AMD XP, AMD64 Winchester, mobile Pnetium etc.) and yes, some overclocked machines as well.
It would be also interesting to re-run models if we get same resuls - and i, somehow, doubt that results will be the same.

Some guru programmer feeback may also put light to there results...

&gt; This is really interesting experiment.
&gt; Comparing Intel P4 Northwood (512kb cache) with Intel Prescott (1MB cache) and
&gt; AMD64 Clawhammer gives slightly different results on average temperature;
&gt; seasonal temperature maps shows larger difference.
&gt; There are a lot of other CPU types to run (AMD XP, AMD64 Winchester, mobile
&gt; Pnetium etc.) and yes, some overclocked machines as well.
&gt; It would be also interesting to re-run models if we get same resuls - and i,
&gt; somehow, doubt that results will be the same.
&gt;
&gt; Some guru programmer feeback may also put light to there results...
&gt;

Pete B has done a great job of running this comparison and including geophi's results. Honza suggests that we might look at a whole host of other cross machine comparisons. My question is this: Have we already run any of these comparisons as part of the standard experiment? If so, could the Oxford team release results from a few of them to volunteers to look at like Pete B has done in this case?

---

&gt; Pete B has done a great job of running this comparison and including geophi's
&gt; results. Honza suggests that we might look at a whole host of other cross
&gt; machine comparisons. My question is this: Have we already run any of these
&gt; comparisons as part of the standard experiment? If so, could the Oxford team
&gt; release results from a few of them to volunteers to look at like Pete B has
&gt; done in this case?
&gt;
I think they are actually going to do a publication on this...but knowing how long it takes to review and publish things, it might be awhile. In order for us to do this type of comparison though, one needs to use CPView, and in order to use CPView, one needs the *pe* files. Those aren't uploaded at the end of the model.

The investiators are going to use whatever files we do upload to do the comparison. The question is, what type of software are they using/going to use to create the difference fields/graphics?

Where next with this theme?

What I want to see next is a graph of how the sum of the differences squared evolves over time.

I am trying to contact Bruno Moretti who is doing the same sulphur cycle model in alpha testing as I am. I do not know if his Athlon XP will produce different or identical results to my Athlon64. Anyone else doing a model where they know another person doing the same model?

---

Visit BOINC WIKI for help

And join BOINC Synergy for all the news in one place.

Hi Crandles,

I'm crunching a model which has already been completed. It seems to be a cold run on my machine and it's a cold run on the other user machine.
It's <a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/workunit.php?wuid=230914">2j5o_100139336</a>
I'm currently at 75% and Boincview tells me it will be completed in 1 month (I know it's long but I have another classic model and 1 alpha model running at the same time)
Let me know if you're interested: Cold runs are not very representative

Thanks Arnaud.

I would be interested if both you and the user 'Team Jolt Cola' would be willing to upload their pe11 pe12 etc files to Lorvija's ftp pages.

---

Visit BOINC WIKI for help

And join BOINC Synergy for all the news in one place.

It's OK for me.
I don't have a broadband connexion but I will upload the whole folder when the wu is finished via a internet public access.

I don't know for 'Team Jolt Cola'. You'll have to try and contact him.
He has completed his model yesterday and I hope he won't erase his model to free his hard disc space.

Edit: I'm going to crunch the Wu on priority thanks to CC4.20.
If everything goes well, the Wu will be finished in 8 days :o)

&gt; Hi Pete, thanks for all the work and throwing those graphics up there. Looks
&gt; like Amy was about 1/2 degree cooler than the other two at the end of the
&gt; model run. Not too much compared to how warm they all got. Stacy and Jayne
&gt; were amazingly close on the time series. From your other graphics, it looks
&gt; very much like the weather differed significantly on the seasonal time scale
&gt; in some places, but the overall earth average was very close.
&gt;

Hi there

The results machine to machine are all pretty close, much closer than I expected.

I also wanted to add these Phase 3 warming results for NH winter and Summer but ran out of time last night.

Amy Phase 3 Winter 2050 to 2064
<img src="http://cpdn.tuxie.org/Pete_B/Amy_Data/WU248363/Phase_3/Amy_Ph3_NHWinter_Warming.png">

Amy Phase 3 Summer 2051 to 2065
<img src="http://cpdn.tuxie.org/Pete_B/Amy_Data/WU248363/Phase_3/Amy_Ph3_NHSummer_Warming.png">

Stacey Phase 3 Winter 2050 to 2064
<img src="http://cpdn.tuxie.org/Pete_B/Stacey_Data/WU248363/Phase_3/Stacey_Ph3_NHWinter_Warming.png">

Stacey Phase 3 Summer 2051 to 2065
<img src="http://cpdn.tuxie.org/Pete_B/Stacey_Data/WU248363/Phase_3/Stacey_Ph3_NHSummer_Warming.png">

Jayne Phase 3 winter 2050 to 2064
<img src="http://cpdn.tuxie.org/Pete_B/Jayne_Data/WU248363/Phase_3/Jayne_Ph3_NHWinter_Warming.png">

Jayne Phase 3 Summer 2051 to 2065
<img src="http://cpdn.tuxie.org/Pete_B/Jayne_Data/WU248363/Phase_3/Jayne_Ph3_NHSummer_Warming.png">

Nothing much to add to what the CO2 warming theory already says i.e. most of the warming occurs in the high latitude regions, especially in the Arctic. The NH summers show a fair amount of land warmimg also.

Pete