climateprediction.net home page
New region?
New region?
log in

Advanced search

Message boards : Number crunching : New region?

1 · 2 · 3 · 4 · Next
Author Message
Profile Iain Inglis
Volunteer moderator
Send message
Joined: 16 Jan 10
Posts: 959
Credit: 1,964,466
RAC: 9,794
Message 54659 - Posted: 18 Aug 2016, 9:00:20 UTC

wah2_dub25_braj_200412_13_430_010663792_1

Les Bayliss
Volunteer moderator
Send message
Joined: 5 Sep 04
Posts: 6781
Credit: 20,534,340
RAC: 8,056
Message 54660 - Posted: 18 Aug 2016, 9:43:26 UTC - in response to Message 54659.

Yes, with several more to come.

KWSN - Sir Frank of the Wood
Send message
Joined: 3 Nov 10
Posts: 39
Credit: 2,257,850
RAC: 0
Message 54661 - Posted: 18 Aug 2016, 10:31:13 UTC

Dublin ??? Dubuque ???

frank

Profile geophi
Volunteer moderator
Send message
Joined: 7 Aug 04
Posts: 1756
Credit: 34,931,570
RAC: 6,467
Message 54662 - Posted: 18 Aug 2016, 12:10:51 UTC
Last modified: 18 Aug 2016, 12:11:30 UTC

Dubai Area outlined in pink on map.

Profile Iain Inglis
Volunteer moderator
Send message
Joined: 16 Jan 10
Posts: 959
Credit: 1,964,466
RAC: 9,794
Message 54663 - Posted: 18 Aug 2016, 13:04:32 UTC

Aha, thanks. Might work for the 2022 World Cup in Qatar as well ...

Profile Iain Inglis
Volunteer moderator
Send message
Joined: 16 Jan 10
Posts: 959
Credit: 1,964,466
RAC: 9,794
Message 54667 - Posted: 18 Aug 2016, 19:34:34 UTC
Last modified: 22 Mar 2017, 10:22:02 UTC

This question shouldn't need to be asked again by me as:

(a) I have downloaded geophi's region map, and

(b) I have written a script to generate an image of the region from the namelist file downloaded with the model, as below.



... which is the same as that identified by geophi, though Dubai is not exactly central!

Les Bayliss
Volunteer moderator
Send message
Joined: 5 Sep 04
Posts: 6781
Credit: 20,534,340
RAC: 8,056
Message 54669 - Posted: 18 Aug 2016, 20:05:49 UTC

It's possible that parts of that large area can/may affect the weather in Dubai, so it's included as a "area of interest".
The area that actually gets used in any one experiment can be varied to suit "current thinking".

Profile Iain Inglis
Volunteer moderator
Send message
Joined: 16 Jan 10
Posts: 959
Credit: 1,964,466
RAC: 9,794
Message 54680 - Posted: 21 Aug 2016, 19:18:19 UTC
Last modified: 22 Mar 2017, 10:22:32 UTC

I've had a look at some of the other areas, Europe (50 km) for example:

... or it makes more sense viewed from space:


I haven't got the CPDN regions quite right yet, though the CORDEX regions do work - no doubt it will become clear as more regions are processed.

However, I only have:

AFR50, ANZ50, DUB25, EU25, EU50, PNW25, SAS50

Other regions would be appreciated, a MEX50 or anything else you can think of. I just need the small download data Zip (e.g. wah2_eu25_a6vc_198812_12_385_010461237.zip) via Dropbox, e-mail etc. - PM me with instructions.

[Edit: Google Earth file replaced by satellite view, as KML renders as XML - on Edge, anyway.]

Les Bayliss
Volunteer moderator
Send message
Joined: 5 Sep 04
Posts: 6781
Credit: 20,534,340
RAC: 8,056
Message 54682 - Posted: 21 Aug 2016, 21:32:18 UTC

Part of the size may be to eliminate boundary effects, but I've sent an email to ask about the size.

Les Bayliss
Volunteer moderator
Send message
Joined: 5 Sep 04
Posts: 6781
Credit: 20,534,340
RAC: 8,056
Message 54684 - Posted: 22 Aug 2016, 2:33:16 UTC

Information is forthcoming, but I don't know when.
Some for now: Some of "our" regions are CORDEX, and some aren't.

And, to make sure we're "on the same page", this is the page that we should be on: WCRP CORDEX

Profile Iain Inglis
Volunteer moderator
Send message
Joined: 16 Jan 10
Posts: 959
Credit: 1,964,466
RAC: 9,794
Message 54686 - Posted: 22 Aug 2016, 10:18:38 UTC
Last modified: 22 Aug 2016, 12:37:48 UTC

As far as I can see, none of the CPDN regions matches any CORDEX region exactly. Sometimes the rotated pole matches (EU, ANZ) or almost matches (SAS50, CPDN=236.68 vs CORDEX=236.66 pole longitude, typo?) but the grid extent is always different. None of that matters, as I'm simply trying to understand the general areas out of interest and as a means of crudely explaining the run times.

Profile Iain Inglis
Volunteer moderator
Send message
Joined: 16 Jan 10
Posts: 959
Credit: 1,964,466
RAC: 9,794
Message 54689 - Posted: 23 Aug 2016, 0:06:34 UTC

I now have a MEX50, which I think completes the set ...

Profile JIM
Send message
Joined: 31 Dec 07
Posts: 1071
Credit: 18,636,429
RAC: 5,368
Message 54690 - Posted: 23 Aug 2016, 3:57:09 UTC - in response to Message 54689.

I now have a MEX50, which I think completes the set ...


Good luck. May all the zips upload for you.
____________

Profile Iain Inglis
Volunteer moderator
Send message
Joined: 16 Jan 10
Posts: 959
Credit: 1,964,466
RAC: 9,794
Message 54704 - Posted: 24 Aug 2016, 20:07:36 UTC
Last modified: 24 Aug 2016, 20:10:00 UTC

BTW The upload Zip from a DUB25 is a whopping 67.5 MB (Zip #1 of mine, at any rate) - which for Mac users is going to require 13 x 67.5 MB = 877.5 MB, plus the restart Zip, to be uploaded at the end. [If you are a Mac user:] It might be prudent to suspend/quit/restart BOINC Manager at suitable points, which will cause any waiting Zips to be uploaded before the model completes.

Profile Iain Inglis
Volunteer moderator
Send message
Joined: 16 Jan 10
Posts: 959
Credit: 1,964,466
RAC: 9,794
Message 54716 - Posted: 27 Aug 2016, 18:13:45 UTC
Last modified: 22 Mar 2017, 10:23:29 UTC

Here is a gallery of region images as viewed from space (in order to give an area indication):















The boundary colours are intended to match the existing region map (supplied by geophi) plus the animation on the main site. However, EU50 doesn't appear to match any of the regions on the map; all the others do. Of course, regions might change for all sorts of reasons and I might have made an error.

Profile Iain Inglis
Volunteer moderator
Send message
Joined: 16 Jan 10
Posts: 959
Credit: 1,964,466
RAC: 9,794
Message 54761 - Posted: 7 Sep 2016, 15:08:55 UTC
Last modified: 22 Mar 2017, 10:25:24 UTC

Some new WAH2 models today for South America.

Profile Iain Inglis
Volunteer moderator
Send message
Joined: 16 Jan 10
Posts: 959
Credit: 1,964,466
RAC: 9,794
Message 54822 - Posted: 22 Sep 2016, 13:21:34 UTC
Last modified: 22 Sep 2016, 14:49:04 UTC

It looks like the new SAM50 model (at least on my machines and for the batch analysed) has a rather pessimistic GFLOPS estimate:

Region..Resolution..Months..GFLOPS........Normalised GFLOPS/sec
DUB.......25...............13..........3,327,828......0.92
EU.........25...............12..........1,903,330......0.85
EU.........25...............13..........2,061,502......0.86
MEX.......50...............12..........784,376.........0.86
MEX.......50...............13..........849,559.........0.88
MEX.......25...............13..........4,172,902......n/a
PNW......25...............16..........1,277,597.......n/a
PNW......25...............24..........1,915,068.......1.17
SAM......50...............13..........2,147,398.......1.68
SAS.......50...............13..........1,495,731.......1.08

[NB If all GFLOPS estimates were perfect for the machines and batches considered then the normalised GFLOPS/sec would be 1.00.]

Profile Iain Inglis
Volunteer moderator
Send message
Joined: 16 Jan 10
Posts: 959
Credit: 1,964,466
RAC: 9,794
Message 54839 - Posted: 26 Sep 2016, 18:16:51 UTC

Some new WAH2 models at 50 km resolution for Africa today. They cover the same area as for the HADAM3P example earlier.

Profile Iain Inglis
Volunteer moderator
Send message
Joined: 16 Jan 10
Posts: 959
Credit: 1,964,466
RAC: 9,794
Message 54882 - Posted: 5 Oct 2016, 14:47:21 UTC
Last modified: 5 Oct 2016, 14:48:35 UTC

The new WAH2 AFR50 model also seems to have a pessimistic GFLOPS estimate:

Region..Resolution..Months..GFLOPS........Normalised GFLOPS/sec
AFR.......50...............13..........1,924,645......1.62
DUB.......25...............13..........3,327,828......0.96
EU.........25...............12..........1,903,330......0.91
EU.........25...............13..........2,061,502......0.91
MEX.......50...............12..........784,376.........0.91
MEX.......50...............13..........849,559.........0.94
MEX.......25...............13..........4,172,902......n/a
PNW......25...............16..........1,277,597.......n/a
PNW......25...............24..........1,915,068.......1.22
PNW......25...............49..........n/a.................n/a
SAM......50...............13..........2,147,398.......1.67
SAS.......50...............13..........1,495,731.......1.14

In a mix of models, a model with a "normalised GFLOPS/sec" greater than 1.00 will finish earlier than expected.

The table has been updated with some more completed models, adding Windows runs to earlier Mac and Linux runs. There's also a new model configuration PNW25/49 that's come out but I haven't received any models of that type yet (and don't have the GFLOPS estimate - though the PNW series are usually scaled).

Profile Byron Leigh Hatch @ team Carl Sagan
Avatar
Send message
Joined: 17 Aug 04
Posts: 279
Credit: 43,022,075
RAC: 0
Message 54883 - Posted: 5 Oct 2016, 16:24:14 UTC - in response to Message 54882.



Hi Iain and everyone ... I'm now crunching two of the Weather At Home2 (wah2) 8.19

wah2_pnw25_c0eh_200312_49_452_010736499_1

does the _49_ signify 49 months and 49 Zip files?

just curiosity.

Weather At Home2 (wah2) 8.19

wah2_pnw25_c0eh_200312_49_452_010736499_1


1 · 2 · 3 · 4 · Next

Message boards : Number crunching : New region?


Main page · Your account · Message boards


Copyright © 2019 climateprediction.net