climateprediction.net home page
Posts by MarkJ

Posts by MarkJ

1) Message boards : Number crunching : Applications page (Message 60440)
Posted 25 Jun 2019 by MarkJ
Post:
Thanks but I was trying to convey that it doesn’t work so the project staff could look at fixing it.

It seems the intent was to have stats by application, or maybe it came as a standard template and the project haven’t customised it to CPDN requirements yet.

77,742 gigaflops is quite low for a BOINC project, but I guess the long running models and/or the size of the uploads is off putting for many users.
2) Message boards : Number crunching : Applications page (Message 60423)
Posted 24 Jun 2019 by MarkJ
Post:
On the Applications page it shows 0 gigaflops for all the apps. Most projects put that bit of information on their Server status page. According to the server status page there are some 78k wah2 in-progress and another 4k of the region independent wah2 in-progress so surely there should be some flops figures.
3) Message boards : climateprediction.net Science : Project announcements discussion (Message 60417)
Posted 23 Jun 2019 by MarkJ
Post:
Just OpenIFS.

Thanks, better update my preferences then as I don’t have any 32 bit libs installed.
4) Message boards : climateprediction.net Science : Project announcements discussion (Message 60415)
Posted 23 Jun 2019 by MarkJ
Post:
A quick question on the HadAM4 N144 and N216 programs. Are they 64 bit or was it just OpenIFS that’s 64 bit? On the Applications page they’re listed as Linux/x86.

Thanks
5) Message boards : climateprediction.net Science : Project announcements discussion (Message 60135)
Posted 12 May 2019 by MarkJ
Post:
making use of the larger memory and processor power available in modern computers.

<snip>

For now, these models are only going to be run on Linux systems.

Does that mean they'll finally be 64 bit apps?
6) Questions and Answers : Unix/Linux : 32 bit support on the way out? (Message 59226)
Posted 23 Dec 2018 by MarkJ
Post:
It seems the Lubuntu folks are following and dropping support of 32 bit as well. They pointed out that a number of upstream packages don't support 32 bit so its getting too hard for them to offer 32 bit.

OSX has already gone to 64 bit only.

Nvidia has gone down this path too, not that it effects this project.

I hope the project can look at switching to 64 bit apps, initially for the Linux and OSX people and then Windows. I am sure the user stats would point out that 95% or more users are running a 64 bit OS.
7) Message boards : Number crunching : Latest BOINC Client Development Status (Message 57425)
Posted 2 Dec 2017 by MarkJ
Post:
Apple have said that High Sierra will be the last to support 32 bit apps.

Apple announced more information during their Platforms State of the Union presentation at WWDC 2017.

The transition to 64-bit-only Mac apps is going to take place very similarly to how we handled it on iOS. Most importantly, High Sierra is going to be the last macOS release to fully support 32-bit apps without compromises. In the next major release after High Sierra, you're going to aggressively start warning users if apps are not compatible for 64-bit. You're updating the Mac App Store rules accordingly.

In about six months, you'll require that all new apps submitted to the Mac App Store will be 64-bit capable and, in a year from now, will require all new apps and updates to existing apps to be 64-bit only when submitted to the Mac App Store.
8) Message boards : Number crunching : Latest BOINC Client Development Status (Message 57236)
Posted 28 Oct 2017 by MarkJ
Post:
What would be nice is if they could build 64 bit Linux apps rather than sticking with 32 bit. Most of the world has moved on from 32 bit and Linux isn't the best at 32 bit compatibility.


The trouble with that is, the code belongs to the Met Office and isn't open source. Even if Oxford did have a licence to alter it, at over a million lines of Fortran it isn't a trivial task.

They should be able to build a 64 bit version without any code changes. Fortran being a high level language doesn’t have any bit-ness.

I wouldn’t expect any speed advantage, just the ability to run it without having to fiddle with 32 bit libraries.
9) Message boards : Number crunching : Latest BOINC Client Development Status (Message 57218)
Posted 27 Oct 2017 by MarkJ
Post:
I don't know if the BOINC server code allows task allocation to only a certain client version or above? This would seem a logical way around the dearth of work for Linux/Mac clients.

I believe there is an option to specify a minimum (and maximum) client version server side.

What would be nice is if they could build 64 bit Linux apps rather than sticking with 32 bit. Most of the world has moved on from 32 bit and Linux isn't the best at 32 bit compatibility.
10) Message boards : Number crunching : Telling lies in cc_config.xml (Message 56178)
Posted 8 May 2017 by MarkJ
Post:
Though you can use it to limit the number of cores/project

e.g. <project_max_concurrent>1</project_max_concurrent>
</app_config>

Useful if you are running a project that requires virtualbox. On this machine if I don't make sure one core isn't running vitual box it makes it unusable at times. Also useful for those who want to micromanage different projects without having to play about with share ratios.


Yes you can use an app_config file to limit the number of apps per project although its nothing to do with cc_config and the ncpus element the OP was talking about.
11) Message boards : Number crunching : Telling lies in cc_config.xml (Message 56176)
Posted 8 May 2017 by MarkJ
Post:
From the BOINC client configuration wiki. See https://boinc.berkeley.edu/wiki/Client_configuration

<ncpus>N</ncpus>
Act as if there were N CPUs; e.g. to simulate 2 CPUs on a machine that has only 1. To use the number of available CPUs, set the value to -1 (was 0 which in newer clients really means zero).

There is no reason for you to use it. If you want to reduce the number of cores available to BOINC the recommended way is to set the percentage of available CPU's to a value less than 100%. Setting to to a value higher than the number of cores/threads on your computer will force it to swap things in and out and your machine performance will suffer.
12) Message boards : Number crunching : RECENT UPGRADE (Message 54452)
Posted 8 Jul 2016 by MarkJ
Post:
I was hoping we'd get the updated message board functions with the upgrade but it doesn't seem to be here. Are there any plans to bring the message boards up to date?
13) Message boards : Number crunching : ANOTHER UPLOAD PROBLEM (Message 54422)
Posted 5 Jul 2016 by MarkJ
Post:
I did see that uploader2 was working until yesterday and uploader4 was off. Now it seems to be the other way around. Maybe they're moving files off to make room.

It took almost a week for me to upload everything. I just left it alone with network enabled.
14) Message boards : Number crunching : ANOTHER UPLOAD PROBLEM (Message 54380)
Posted 26 Jun 2016 by MarkJ
Post:
I've got uploads failing for the hapam3p-eu models (all 4 zips) as well. I have around 128 zip files waiting to go.
15) Message boards : Number crunching : PNW upload issues (Message 52653)
Posted 2 Oct 2015 by MarkJ
Post:
And the http_debug output...


climateprediction.net 02-10-2015 09:38 PM [http] HTTP_OP::libcurl_exec(): ca-bundle 'C:\Program Files\BOINC\ca-bundle.crt'
climateprediction.net 02-10-2015 09:38 PM [http] HTTP_OP::libcurl_exec(): ca-bundle set
climateprediction.net 02-10-2015 09:38 PM Started upload of hadam3p_pnw_xfiq_2002_1_010348924_0_10.zip
climateprediction.net 02-10-2015 09:38 PM [http] HTTP_OP::libcurl_exec(): ca-bundle 'C:\Program Files\BOINC\ca-bundle.crt'
climateprediction.net 02-10-2015 09:38 PM [http] HTTP_OP::libcurl_exec(): ca-bundle set
climateprediction.net 02-10-2015 09:38 PM Started upload of hadam3p_pnw_xfiq_2002_1_010348924_0_11.zip
climateprediction.net 02-10-2015 09:38 PM [http] [ID#26] Info: timeout on name lookup is not supported
climateprediction.net 02-10-2015 09:38 PM [http] [ID#26] Info: Hostname was NOT found in DNS cache
climateprediction.net 02-10-2015 09:38 PM [http] [ID#26] Info: Trying 128.193.64.193...
climateprediction.net 02-10-2015 09:38 PM [http] [ID#27] Info: Found bundle for host boinc1.coas.oregonstate.edu: 0x2696e20
climateprediction.net 02-10-2015 09:38 PM [http] [ID#27] Info: timeout on name lookup is not supported
climateprediction.net 02-10-2015 09:38 PM [http] [ID#27] Info: Hostname was found in DNS cache
climateprediction.net 02-10-2015 09:38 PM [http] [ID#27] Info: Trying 128.193.64.193...
climateprediction.net 02-10-2015 09:38 PM [http] [ID#26] Info: connect to 128.193.64.193 port 80 failed: Timed out
climateprediction.net 02-10-2015 09:38 PM [http] [ID#26] Info: Failed to connect to boinc1.coas.oregonstate.edu port 80: Timed out
climateprediction.net 02-10-2015 09:38 PM [http] [ID#26] Info: Closing connection 25
climateprediction.net 02-10-2015 09:38 PM [http] [ID#27] Info: connect to 128.193.64.193 port 80 failed: Timed out
climateprediction.net 02-10-2015 09:38 PM [http] [ID#27] Info: Failed to connect to boinc1.coas.oregonstate.edu port 80: Timed out
climateprediction.net 02-10-2015 09:38 PM [http] [ID#27] Info: Closing connection 26
climateprediction.net 02-10-2015 09:38 PM [http] HTTP error: Couldn't connect to server
climateprediction.net 02-10-2015 09:38 PM [http] HTTP error: Couldn't connect to server

And then it happily connects to Google as the reference site.

It seems its trying to connect to boinc1.coas.oregonstate.edu and timing out.

ping
C:\windows\system32>ping boinc1.coas.oregonstate.edu

Pinging maui.oce.orst.edu [128.193.64.193] with 32 bytes of data:
Reply from 128.193.88.130: Destination host unreachable.
Reply from 128.193.88.130: Destination host unreachable.
Request timed out.
Reply from 128.193.88.130: Destination host unreachable.

Ping statistics for 128.193.64.193:
Packets: Sent = 4, Received = 3, Lost = 1 (25% loss),
16) Message boards : Number crunching : PNW upload issues (Message 52650)
Posted 2 Oct 2015 by MarkJ
Post:
Is anyone else having issues uploading? In the last couple of days I have seen all my PNW zip files stuck trying to upload. Between 6 machines I have over 180 zip files that don't seem to want to start their transfers. BOINC is saying they are Transient HTTP errors.

Internet access seems fine, even to the CPDN web site and scheduler requests seem to work. I've tried resetting the router and flushing DNS cache.

They are all the recent pnw_x??? series work units downloaded in the last week or so. Machines are all Win10 x64 and running BOINC 7.6.9.
17) Message boards : Number crunching : How to default Boinc to Network activity suspended (Message 52317)
Posted 24 Jul 2015 by MarkJ
Post:
Why not get 7.6.2 from LocutusOfBorg's ppa? He builds the latest BOINC versions for Ubuntu. He has versions for Wily, Vivid, Utopic and Trusty.

There are instructions on there how to add his ppa. Once you've done that you can use apt-get to pickup the latest version.
18) Message boards : Number crunching : hadam3p_anz download errors (Message 52167)
Posted 4 Jul 2015 by MarkJ
Post:
A heap more. This time they are the hadam3p_anz_u???_2013_1_??? series. They all appear to be checksum errors. And before I'm told it must be me, my wingmen also have download errors for the same work units.

The last lot were also Jan 2013 work units so maybe that's a common theme here.

04-07-2015 11:12 PM [error] MD5 check failed for hadam3p_anz_u0ot_2013_1_009998589.zip
04-07-2015 11:12 PM [error] expected b031e384b89ac9ee2cf46b76085503a3, got ec890bfc2d0edf85e25aea42dacf335c
04-07-2015 11:12 PM [error] Checksum or signature error for hadam3p_anz_u0ot_2013_1_009998589.zip

04-07-2015 11:12 PM [error] MD5 check failed for hadam3p_anz_u0aq_2013_1_009998082.zip
04-07-2015 11:12 PM [error] expected 37bfe482d612f8e4c03c54c0b765acb0, got 910b15c08459b1b50179cdf9c54348cf
04-07-2015 11:12 PM [error] Checksum or signature error for hadam3p_anz_u0aq_2013_1_009998082.zip

04-07-2015 11:06 PM [error] MD5 check failed for hadam3p_anz_u0t2_2013_1_009998742.zip
04-07-2015 11:06 PM [error] expected 32c5e563e2fa8a5b71ca68c9f9bf4428, got 68e14bf99e8553b60ac6b386341a54c0
04-07-2015 11:06 PM [error] Checksum or signature error for hadam3p_anz_u0t2_2013_1_009998742.zip
19) Message boards : Number crunching : hadam3p_anz download errors (Message 52079)
Posted 23 Jun 2015 by MarkJ
Post:
It seems the latest batch submitted in the last few hours are missing some files.

climateprediction.net 23-06-2015 11:01 PM Started download of hadam3p_anz_d003_2013_1_009937381.zip
climateprediction.net 23-06-2015 11:01 PM Giving up on download of hadam3p_anz_d003_2013_1_009937381.zip: permanent HTTP error


If you spot these trying to download AND the hapam3p_anz_xxx.zip is 0k then there is no point downloading the other 46Mb of work unit data. I've been aborting the few that I spotted. If its 10k then it will be fine, its only the 0 bytes ones.

So far I have only seen hadam3p_anz_d???_2013_1_???.zip files that have this problem.
20) Message boards : Number crunching : Linux/Mac/Windows segmentation (Message 51849)
Posted 18 Apr 2015 by MarkJ
Post:
How does that work Mark? I and I suspect a lot of other crunchers have never used virtual boxes. What is the performance hit if any to crunching and other work particularly if running on minimum spec for memory?

You install VirtualBox on your machine. You can do that either via the BOINC+Vbox installer or do each individually. I prefer individually as there have been reports of the BOINC+Vbox installer failing on the Vbox part. Also you can usually get a more up to date Vbox directly from the source (virtualbox.org). The recent versions of BOINC will detect the presence of Vbox once its installed.

The project sends Vbox virtual disk images as work units. The disk image has whatever OS (usually a flavour of Linux) and everything else to run the task.

Overheads-wise they use more memory of course due to the Vbox memory footprint. The CPU usage would depend on the task but CPDN are single-threaded so I'd expect 1 CPU thread per work unit running.


Next 20

©2019 climateprediction.net