TN-Grid challenge

PDW

PDW

Associate
Joined
21 Feb 2014
Posts
1,867
Now Formula Boinc has finished for the year you might be lacking direction as you wander aimlessly towards your Christmas pud :)

Wander no more, a challenge has been issued on TN-Grid

The challenge is at:

http://boincstats.com/en/stats/challenge/team/chat/831

Start time: 2016-12-10 16:00 UTC
End time: 2016-12-16 16:00 UTC

It is likely the project will end up in The Vault unless it can't handle the load.
They have recently upgraded the server CPU, RAM and storage so hopefully that won't be a problem.

The project is now open, the invitation code can be found on the main page http://gene.disi.unitn.it/test/

Can you sign us up for the challenge please hanluc ?
 
Associate
Joined
20 Apr 2014
Posts
1,865
Location
Herts/uk
Has it really finished or has a year turned into 11months while i was asleep?
Dont matter if it runs nov>nov does it?
Dont the points we make now count for anything?

EDIT: Anyways this comp sucks cos its not letting me join the project so ill just have to do my own thing hahah
 
Last edited:

PDW

PDW

Associate
OP
Joined
21 Feb 2014
Posts
1,867
Looks like the crunching part of FB only runs for the first 11 months and then it takes a month to get the medals sorted out :rolleyes:
Unless there is a medal ceremony somewhere exotic that hanluc has been invited to attend and accept our gold medal on behalf of the team :p

You can do some work now for next year, just find a project that takes a long time to validate, some of the Primegrid work can take 3-6 months to validate !


TN-Grid - You need to register on their website first to create an account, then you can attach a client to it.
See the first line immediately below 'Join TN-Grid' on the left hand side of their home page :)
 
Associate
Joined
27 Jun 2014
Posts
28
Ah yer :)

app config for this is

<app_config>
<app>
<name>gene</name>
<max_concurrent>8</max_concurrent>
</app>
</app_config>
Not sure of the minimum Boinc version, but with at least v7.4.22 forward you can use the <project_max_concurrent> tag.
<app_config>
<project_max_concurrent>8</project_max_concurrent>
</app_config>

You can still use the <app> tag structure to control the number of concurrent WUs for a specific app though.
 

PDW

PDW

Associate
OP
Joined
21 Feb 2014
Posts
1,867
Looks like there might have been a few holding back for the last day or two !

Team is currently in 2nd place, a long way behind the hosts :)
hanluc sitting in 3rd place for now, can you keep it ?
 

PDW

PDW

Associate
OP
Joined
21 Feb 2014
Posts
1,867
You should be in 4th overall for a while, no-one else is near :)

I still have over 1,200 units pending so am thinking someone else is still going to drop a load soon.

Andy should be looking in his rearview mirror as Senture is coming up fast :D
 
Associate
Joined
1 May 2007
Posts
2,354
Location
London
Joined in a little late, had some other things going on so I missed the start. Thrown a few Xeons at it, but it all depends on how long units take to be validated.
 

PDW

PDW

Associate
OP
Joined
21 Feb 2014
Posts
1,867
I have received 2 at a time recently but only having 8 running and taking about 43 minutes to run I never run out. Is that your problem on the big box ?

If I've ever increased the cache settings in the past it would happily download dozens at a time for linux/windows.

I see our gold medal has finally turned up in FB :)
 
Soldato
Joined
22 Oct 2010
Posts
2,961
Location
Ratae Corieltauvorum
I have received 2 at a time recently but only having 8 running and taking about 43 minutes to run I never run out. Is that your problem on the big box ?

If I've ever increased the cache settings in the past it would happily download dozens at a time for linux/windows.

I see our gold medal has finally turned up in FB :)

Yes, the big box works best under Linux as I have to turn HT off in Windows because it doesn't handle core assignment well over 64 threads. I think I'll have to do that as my linux cache is hovering around the 30 WU mark after hours of crunching.
 

PDW

PDW

Associate
OP
Joined
21 Feb 2014
Posts
1,867
So it actually runs out of work to do ?

I think the most I have seen downloaded at any one time is 256 for a 64 thread machine, that might be the limit the project allows. Even that would take some time to chew through given that your big box will be taking a lot longer to process them.

If you aren't able to keep 256 in progress then something else is wrong :(
 

PDW

PDW

Associate
OP
Joined
21 Feb 2014
Posts
1,867
You are back up to 3rd :)

I'm still stuck at 7th but Senture has overtaken Andy :D

Not sure why Andy has set <max_concurrent>8</max_concurrent> when he has 16 cores !
 
Soldato
Joined
22 Oct 2010
Posts
2,961
Location
Ratae Corieltauvorum
So it actually runs out of work to do ?

I think the most I have seen downloaded at any one time is 256 for a 64 thread machine, that might be the limit the project allows. Even that would take some time to chew through given that your big box will be taking a lot longer to process them.

If you aren't able to keep 256 in progress then something else is wrong :(

The work server only gives me 1 work unit when I ask for work (sometimes nothing) and I can only ask every two minutes, so by the time I've got around 30 in my cache I start returning work, leaving 40 threads idle.

I have my cache set for 7 days, so if it is me I don't know what I'm doing wrong.

I've switched back to windows now.
 

PDW

PDW

Associate
OP
Joined
21 Feb 2014
Posts
1,867
It was actually a 32 thread machine that got 256 tasks.

You can't even split the threads over two or more clients as they aren't seen as unique. Only other option would be if you can create a unique virtual machine on the linux box to co-exist.

Sounds like Windows it is then !
 

PDW

PDW

Associate
OP
Joined
21 Feb 2014
Posts
1,867
Do you think the previous version was bad or just the current one is newer and handles big boxes better ?

We are still in 2nd place but it looks like we are being caught by EVGA, this time tomorrow we could be relegated to 3rd :(
 
Soldato
Joined
22 Oct 2010
Posts
2,961
Location
Ratae Corieltauvorum
Do you think the previous version was bad or just the current one is newer and handles big boxes better ?

It did the same on another machine, which had the same 14.4 Ubuntu version, I believe it's a combination the that, the boinc repo and the project itself.


We are still in 2nd place but it looks like we are being caught by EVGA, this time tomorrow we could be relegated to 3rd :(

4 vs 17 ... not fair :(
 
Back
Top Bottom