FormulaBOINC

[ao_ika_red]

I guess TN-Grid tasks are leaned towards to my integer core rather than the shared FPU module. If I run Seti cpu tasks with 2 tasks per module configuration, it would certainly bog down my cpu.

 

[Orange Kid]

A little bit of pre sprint stats.
Should be interesting with the PrimeGrid challenge happening at the same time. I don't see either of the Overclockers over there.

Formula Boinc....................Top Five
1

Overclock.net ......................795
2

OcUK - Overclockers UK ....724
3

TeAm AnandTech ...............700
4

Sicituradastra. .....................430
5

Crunching@EVGA ..............423

TN Grid............................Marathon
1 Overclock.net ......................25 ............53,992,181
2 XtremeSystems ...................18 ...........30,906,364
3 Ars Technica ........................15 ...........16,013,239
4 The Scottish Boinc Team .....12 ...........11,219,485
5 TeAm AnandTech ................10 .............9,401,094
6 OcUK - Overclockers UK ......8 ..............7,317,326
7 Crunching@EVGA ................6 ..............2,030,954
8 SETI@Netherlands ...............4 ...............1,635,249
9 AMD Users ............................2 ...............1,195,606
10 France .................................1 ...................904,497

 

[Markfw]

They are not accepting new clients, so I guess I am out.

 

[StefanR5R]

Account creations is supposed to work via web interface, in which an invitation code needs to be entered. This is a measure against automated account creation by spammers.
From the home page:

• invitation code: science@tn
• account creation page: http://gene.disi.unitn.it/test/create_account_form.php
• when this was done (and team affiliation set up etc.), then "add project" in boincmgr
• enter http://gene.disi.unitn.it/test/ as project URL, as TN-Grid isn't listed by default

It's months ago that I did this myself, so I don't know if it still works as it is supposed to work.

--------
The reading of TN-Grid's "science meter" just before the sprint starts:

V. vinifera 'OneGenE' (2nd dataset)

# genes ............. 28013
DB ..................... 28013
Executed ........... 25962 (92.68%)
Last 10 days ...... 300.80/day​

E. coli 'OneGenE' (α=0.01)

# genes ............. 3343
DB ..................... 758
Executed ........... 0 (0.00%)
Last 10 days ...... 0.00/day​

http://gene.disi.unitn.it/test/gene_science.php

 

[ao_ika_red]

I got no problem when registering to TN-Grid yesterday.

 

[Markfw]

Account creations is supposed to work via web interface, in which an invitation code needs to be entered. This is a measure against automated account creation by spammers.
From the home page:

• invitation code: science@tn
• account creation page: http://gene.disi.unitn.it/test/create_account_form.php
• when this was done (and team affiliation set up etc.), then "add project" in boincmgr
• enter http://gene.disi.unitn.it/test/ as project URL, as TN-Grid isn't listed by default

It's months ago that I did this myself, so I don't know if it still works as it is supposed to work.

--------
The reading of TN-Grid's "science meter" just before the sprint starts:

V. vinifera 'OneGenE' (2nd dataset)

# genes ............. 28013
DB ..................... 28013
Executed ........... 25962 (92.68%)
Last 10 days ...... 300.80/day​

E. coli 'OneGenE' (α=0.01)

# genes ............. 3343
DB ..................... 758
Executed ........... 0 (0.00%)
Last 10 days ...... 0.00/day​

http://gene.disi.unitn.it/test/gene_science.php

Thanks, I am in with 29 cores !

 

[StefanR5R]

Mains power went out for a second earlier today. This happens very rarely here, maybe once in a year. So of course this must happen during a DC competition... Fortunately I was present and could restart all clients right away. Unfortunately, internet service was down for about 3 hours.

The clients which had network enabled when I started them without internet access were unresponsive to remote control. Also, I saw many TN-Grid tasks finished with computation error, which I suspect was caused by finishing the tasks while the client had network enabled but internet was out. So I shut down these clients once more, edited client_state.xml to say <user_network_request>3</user_network_request> = network suspended, and started the clients again, thus regained control.

Of course the internet outage also caused some of the clients to empty their work queue. But by a strange coincidence, I had some PrimeGrid llrTRP tasks stored on these hosts so that I could keep their CPUs busy.

 

[Markfw]

Sp it takes about 36 minutes per task, and I have done 317 in 10 hours, but 210 are pending validation/

 

[StefanR5R]

Of my completed tasks, 1/3rds are validation pending, 2/3rds validated. Many of these tasks were downloaded yesterday already though. So, validation lag isn't too bad; and it shouldn't be, since the project does not allow very deep task queues.

 

[Orange Kid]

As we near the end of day one, we find ourselves in first place.
One has to wonder if we should ease up and put some more into the PrimeGrid Challenge or are we being toyed with? Is OcUK playing possum? Time will tell. I say keep the heat coming here

Formula Boinc.................Top Five
1

Overclock.net ......................810
2

OcUK - Overclockers UK ....742
3

TeAm AnandTech ..............725
4

Crunching@EVGA .............436
5

Sicituradastra. ....................430

TN-Grid....................Marathon
1 Overclock.net ......................25 .........54,131,374
2 XtremeSystems ...................18 .........31,016,511
3 Ars Technica ........................15 .........16,073,189
4 The Scottish Boinc Team .....12 ..........11,219,676
5 TeAm AnandTech .................10 ...........9,709,747
6 OcUK - Overclockers UK ........8 ...........7,571,156
7 Crunching@EVGA ..................6 ...........2,086,305
8 SETI@Netherlands .................4 ...........1,635,249
9 AMD Users ..............................2 ...........1,195,859
10 France ...................................1 ...............904,743


TN-Grid...............Sprint...................Day one
1 TeAm AnandTech ................25 ..............518,029
2 OcUK - Overclockers UK .....18 ..............397,471
3 Overclock.net .......................15 ...............337,538
4 Crunching@EVGA ................12 ..............116,851
5 XtremeSystems .....................10 .............112,830
6 Ars Technica ...........................8 ................67,841
7 meisterkuehler.de Team .........6 ................24,917
8 Portugal@Home .....................4 ..................5,389
9 France .....................................2 ..................2,554
10 CANAL@Boinc .....................1 ..................1,040

 

[Markfw]

Is there an individual breakdown ?

 

[ao_ika_red]

Given recent OcUK behaviour, I'll keep mine full blast on TN-Grid. I think they will surprise us on Day2.

 

[StefanR5R]

Is there an individual breakdown ?

Formula Boinc provides only team stats, no user stats. You can of course check at stats.free-dc.org. Their TN-Grid stats are apparently updated once per day, like most of their other project stats.

• TeAm Anandtech
  
• OcUK
• OCN
  
• XtremeSystems
  
• Crunching@EVGA

BTW, XtremeSystems aren't actually sprinting; this is their normal daily pace throughout the last month at least. And Crunching@EVGA are distracted by battling BOINC@Poland for 5th at the current PrimeGrid challenge.

 

[TennesseeTony]

OCN is the team we need to keep an eye on. They are treating this Sprint as a long-jump competition, and went from mid 100's to 400+ thousand in 12 hours or so.

 

[ao_ika_red]

OCN is the team we need to keep an eye on. They are treating this Sprint as a long-jump competition, and went from mid 100's to 400+ thousand in 12 hours or so.

I wish it's just a case when somebody unload whatever they have, otherwise we'll get some problem.
On personal note: Today's weather is not helping my machines. The room temp. hits 30s this afternoon and because of that, my psu is going crazy. I wish Kiska has better day than mine.

 

[Kiska]

I wish Kiska has better day than mine.

Which I do, thank goodness of the cloudy and almost rainy day here

 

[Kiska]

Wait what?
http://gene.disi.unitn.it/test/host_app_versions.php?hostid=17673
The SSE2 app is faster than both the AVX and FMA?!
I think I'll wait for a few more results to come in before making a judgement

 

[StefanR5R]

They perform the same, give or take small random variations.

My host with most points, E5-2696v4: sse2 3.57/ avx 3.62/ fma 3.79 GFLOPS
http://gene.disi.unitn.it/test/host_app_versions.php?hostid=10513

My host with second most points, E5-2696v4 too: sse2 3.63/ avx 3.62/ fma 3.55 GFLOPS
http://gene.disi.unitn.it/test/host_app_versions.php?hostid=10511

 

[Markfw]

My 1700x is stuck downloading a unit, its not doing anything in TN-Grid

 

[Ken g6]

My 1700x is stuck downloading a unit, its not doing anything in TN-Grid

Huh. You should be able to abort the download. Or you could detach from and reattach to the project.

 

[Markfw]

ok, I aborted the download and reset the project. Now its not doing anything

 

[Kiska]

They perform the same, give or take small random variations.

See BOINC scheduler does not care about variations. Once an app has had 11 successful validations. It will select the fastest app(which currently would be the SSE2 app for my 6200U), until the project adds a new app.

 

[Kiska]

ok, I aborted the download and reset the project. Now its not doing anything

Can you manually force it to update?

Click update

 

[Markfw]

I tried update, reset, everything

At least I turned WCG back on, so its no longer idle

 

[Markfw]

Well, 5 hours later, it has decided to work, so I suspended WCG for now on that box.