ARGOMENTO:

E' appena partito un nuovo progetto di ClimatePrediction.net, riguardante il...

Nuovo progetto CPND

L'applicativo potrebbe essere questo: "WRF 4.6.1 smp urban configuration".
Ma non vedo ancora lavoro in coda...

Be advised, these are going to be LONG RUNNING work units. Maybe as much as 90 days on some systems.
Update: The model runs will be simulating 90 days. The actual runtime for the WUs seem to be predicted at 30 days. The description says 10+ days so I guess we will have to run a few and see how the timing works out. This is an OPT-IN project so you have to select it in your CPDN preferences to get any work.

About CPDN preferences

cpdnboinc preferences

Enable OpenIFS multicore app.This app requires ~26Gb of memory and 4 cpus.
Only enable this option if you can provide the required resources.
Enable WRF long-running multicore app.This app requires at least 10 days compute time
if run 24/7 on recent hardware and 4 cpus.
It requires ~6Gb of memory.
Only enable this option if you can provide the required resources.

Computing status
Work
Tasks ready to send 0
Tasks in progress 2
Total tasks computed 9224492

Non ho capito 26gb si intende per 4 core? quindi su 8 core raddoppia dico bene?
Mi esploderebbe il pc XD

Memory requirement is not related to CPU cores. For the OIFS project, each WU uses about 26GB and defaults to 4 CPU threads. If you change the thread count through the app_config.xml file, up to a maximum of 8 threads, the memory requirement will still be 26GB.

The WRF project uses 6GB and a default of 4 CPU threads. Once again, you can use up to 8 CPU threads by changing/using the app_config.xml file. The memory requirement stays a 6GB.

Like other CPDN projects, WRF will be using the trickle capability to send data back to the server periodically. It should be noted that the amount of data sent is not trivial for those who might be on slow internet or metered internet connections. It will probably be in the 10s of MBs each trickle. Supposedly Glenn is still working on optimizing the compression the output files.

Latest Update (20:30 UTC):
App: WRF 4.6.1 urban configuration, linux only.

Batches 1113 & 1114.

Description: London - urban now scenario. Year 2006.
- 33 workunits per batch. Subsequent batches will be multi-year.
- For more information about this project please see: main.cpdn.org/active_projects.php

Technical requirements
- multicore, default 4 threads.
- max memory 6Gb
- runtime: ~ 10 real days running 24/7 on a modern cpu at 4 cores.

Data
- Download. A 2Gb data file has to be downloaded for each workunit.
- Uploads. Total of 15 uploads, each upload approx 260Mb.

Notes:
- Because of the long runtimes, you must opt-in for these tasks by enabling 'run WRF' in your project preferences.
- Only 1 workunit per host at present. This will change for the forthcoming larger batches.

Thanks entity. I enabled and opted all in but did not get anything. Tried on Ubuntu in a Proxmox server, 32 GB RAM available, 24 processors, Intel XEON Silver 4510. Is this a supported config ? What could I try ? I didn't touch any xml file.
Thanks ! ciao

The latest project just started and Glenn has only released 33 WUs to start with to make sure nothing strange falls out. He will be releasing some more after he gets the data from the researchers and these initial WUs have run a few days. What I do with CPDN is set the project to accept work and then implement a cron job to ask for work about every hour or so. That way it doesn't matter when Glenn releases the work, I will probably get a few during one of the updates. The project is supposed to have 2200 WUs total.

Thanks.
If I leave CPDN active in Boinc Manager, isn't it going to ask the CPDN server automatically every 60 minutes if there are some tasks and alternating data crunching from one project to another every 60 minutes?
Thanks, ciao

No, it will only ask for work every 60 minutes. If a WU is sent, the client will determine when to run it based on project priorities and deadlines. It may start right away and it may not? As CPDN is configured right now, you can only get one WU per host and it will use 4 CPUs by default. Other work can run alongside of it if there are processors available.