How get WCGrid Stats to show in Participating Projects

Nevada Bob
Nevada Bob
Joined: 4 Aug 15
Posts: 9
Credit: 22924880
RAC: 0
Topic 198361

I see that other people have World Community Grid Statistics showing in their participating projects. How do I get my WCGrid stats to show up on my profile page?

noderaser
noderaser
Joined: 9 Feb 05
Posts: 50
Credit: 49785373
RAC: 17357

How get WCGrid Stats to show in Participating Projects

Are you talking about the stats images in signatures (like mine) or on the account page? I just did a search for your name over at BOINCstats, the interesting thing is your account name only shows as being associated with a WCG account--which obviously can't be the case since you have the same name here and your account profile shows you having credit at Rosetta, MilkyWay, and PrimeGrid as well.

One thing that you could try, is to attach one BOINC Manager to all of those projects, let them update at the projects (no work has to be completed, though you'll usually get a task for "project initialization") and wait a few days for all the stats gremlins to catch up. That will rule out/fix the dreaded "split CPID" syndrome.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5848
Credit: 110008615922
RAC: 24340232

RE: How do I get my WCGrid

Quote:
How do I get my WCGrid stats to show up on my profile page?


Make sure you use the same email address at whatever projects you want other project stats to show up on. Use a single machine that has all the projects showing on the projects tab of BOINC Manager. The projects don't need to be active, they just need to show on the projects tab. Projects can even be suspended just as long as they show.

With those two conditions met, select each project in turn and click the 'update' button. It will take a while for all projects to synchronise on a single CPID (cross project identifier). You might be able to make things go faster by doing a second round of updates after a couple of hours if you're still not seeing all projects.

After a day or two, your account page at each project should list all of your projects to which you are attached with that unique email address. The current value of the CPID is listed on your account page. Take a look at this page for each project and see which ones don't line up. They all need to be precisely the same and it should happen eventually.

Cheers,
Gary.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5848
Credit: 110008615922
RAC: 24340232

I was having a look at your

I was having a look at your computers attached to this project, in particular this one which has a Pitcairn series GPU. What concerned me was that BRP6 tasks (Parkes PMPS XT) are taking something like 11 hours to crunch. Your GPU should be able to do those tasks much faster than that - perhaps around 2 hours.

If you were interested, I'd be happy to diagnose the problem and suggest some settings changes to achieve a much faster throughput. If you'd like to do this, just list the projects that currently run on that machine and the mix of task types that you normally have running. Do any of the other projects supply GPU tasks?

Cheers,
Gary.

Nevada Bob
Nevada Bob
Joined: 4 Aug 15
Posts: 9
Credit: 22924880
RAC: 0

1. Gary, I would be grateful

1. Gary, I would be grateful for any help to speed up the process. It appears you are referring to my computer I call P8Z68-V because that is the motherboard model. This is the main computer I use but is idle 98% of the time.

2. Noderaser, I don't understand why my profile page shows this without any WCGRid stats. I have 41 years with WCGrid.

Projects in which Nevada Bob is participating

Einstein@Home 4,211,789 15,470 4 Aug 2015
Rosetta@home 1,018,174 1 11 Jan 2011
MilkyWay@home 165 0 27 Feb 2012
PrimeGrid 37 0 27 Feb 2012[/img]

Nevada Bob
Nevada Bob
Joined: 4 Aug 15
Posts: 9
Credit: 22924880
RAC: 0

I just double checked. I

I just double checked. I 'was' using a different email address in WCGrid and another one for Einstein. A-Ha!!

Don't understand what that means or why it matters. I corrected both to the same email address. Will that fix it? I don't understand how to do the CPID thing you describe. (And I thought I was computer literate...) Guess not, I can build them from scratch...

Nevada Bob
Nevada Bob
Joined: 4 Aug 15
Posts: 9
Credit: 22924880
RAC: 0

I currently do not have any

I currently do not have any GPU tasks running besides Einstein.
Currently I am running WCGrid and only allow GPU for Einstein.
After I catch up on a couple WCGrid projects I will go back to participating in all tasks with WCGRid. I will then add back in Rosetta and split it 50/50 between Rosetta and WCGRid with Einstein still only GPU.

Nevada Bob
Nevada Bob
Joined: 4 Aug 15
Posts: 9
Credit: 22924880
RAC: 0

Oh I think you are referring

Oh I think you are referring to BOINC.
Open up BOINC and the projects tab there and update.
Got it. Done it.
Will see if that fixes it.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5848
Credit: 110008615922
RAC: 24340232

RE: I just double checked.

Quote:

I just double checked. I 'was' using a different email address in WCGrid and another one for Einstein. A-Ha!!

Don't understand what that means or why it matters.


It matters because the email address is used to identify you across different projects. There could be 100 Nevada Bobs around the place but only you should have your unique email address.

Quote:
I corrected both to the same email address. Will that fix it?


Should do. If you've visited the projects tab in BOINC Manager on a machine connected to all your projects and clicked on each project listed to select it and then clicked on the update button, (rinse and repeat until all projects are updated) this should speed up the process of having each project identify you correctly and get an updated version of the CPID if necessary. Since WCG had the 'different' email address and all the other projects are listed here, I think you probably just need to 'update' the WCG project from your machine currently running Einstein. It wont hurt to 'update' all projects.

Quote:
I don't understand how to do the CPID thing you describe.


The CPID synchronising is done automatically by BOINC. To speed it up, on a machine attached to all projects, sequentially update each project in BOINC Manager.

Cheers,
Gary.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5848
Credit: 110008615922
RAC: 24340232

RE: 1. Gary, I would be

Quote:
1. Gary, I would be grateful for any help to speed up the process. It appears you are referring to my computer I call P8Z68-V because that is the motherboard model. This is the main computer I use but is idle 98% of the time.


I'm referring to the machine I linked to - i7-2600K with a HD7850/7870 (Pitcairn) series GPU. It's HT enabled and BOINC sees all 8 cores. Only you can see the computer name and certain other private details. Others see just a subset of what you can see :-).

You are running two types of GPU tasks, Arecibo GPU (BRP4G) and Parkes PMPS XT (BRP6). The BRP4G tasks aren't always available and for best efficiency on your GPU, I suggest you run BRP6 exclusively. If you go to your account page on the website and click on the Einstein project preferences, you will see the various science runs that are available. There are blocks of preferences for different 'locations' or 'venues' (default, home, school, work). If you're not using different venues you will only see the default values. You need to click the edit preferences link for the venue you use (default?) to open an edit screen and then remove the tick for the Arecibo GPU run and make sure the Parkes PMPS XT run remains ticked. While you are there, change the GPU utilization factor of BRP Apps setting from 1 to 0.5. If it wasn't showing 1 please indicate what the previous value was. The default is 1 and I'm assuming you hadn't changed it previously. Once you make those changes, please save them by clicking the 'update preferences' box.

I reckon your computer may be running GPU tasks slowly because all 8 cores are crunching CPU tasks for other projects. AMD GPUs can perform rather badly under those circumstances. The GPU utilization factor change is designed to run two GPU tasks simultaneously and to automatically restrict the number of CPU cores running CPU tasks to 7. I think this may make a rather dramatic improvement to the GPU crunch time. The change happens after your machine downloads new GPU work.

One way to speed this up is to make a small increase to your work cache settings through the computing preferences link on your account page. Whatever value you have for Store at least x.xx days of work, increase it by a small amount (0.2 days, say), to trigger an immediate work fetch. After that, you should see 2 GPU tasks running and a maximum of 7 running CPU tasks (ie. 1 core is available for GPU support duties).

Please ask if you're unsure about any of the above. If we don't get a dramatic speedup in GPU crunching, we'll have to go looking for other factors. For instance, have you placed restrictions on how BOINC is allowed to use your machine's resources?

Cheers,
Gary.

Nevada Bob
Nevada Bob
Joined: 4 Aug 15
Posts: 9
Credit: 22924880
RAC: 0

Gary, Thanks so much, I don't

Gary,
Thanks so much,
I don't have any restrictions on BOINC usage. I've made the changes you suggested. This computer still has 3.5 hours to go on an Archibo project before it gets another project. Changing the cache to 0.2 did not get an immediate work fetch. Maybe because there is still 3.5 hours to go on this project? I have done an update. Should I create a home profile and use that for my other computer so it works on both Archibo and Parkes?

On another note, my page still doesn't show WCGrid stats.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.