FormulaBOINC

[StefanR5R]

@crashtech, in short, the configured RAM limit works perfectly for me so far. But I only need it on Linux, not on Windows.

Long story,
on my Linux boxes:

• I looked at RAM usage only during the first ≈2.5 hours, and at one time ≈ten hours later.
• I started with 14e, 15e, and 16e tasks, i.e. a mixture of ≤0.5 GB and ≤1.0 GB tasks. But right now I have 16e tasks only (≤1.0 GB per task). Both situations worked out right.
  
• So far, the percentage of RAM which I configured controls very well how many tasks are allowed to run at any time. I see it by the amount of used vs. free RAM, and by the occasional blocking of tasks in "waiting for memory" on those machines which are tight on RAM.
  
• On dual-socket machines with 64 GB RAM, I chose 88 % RAM "during use" as well if "not in use". I.e. ≈7 GB RAM remain for the system and the rest.
  
• I currently don't have anything else running on these machines. Granted, there will be one or another cron job running at night. I'm crossing fingers that they don't cut too much into the remaining RAM.
• Swap partitions are very small on these machines; I should do something about that eventually. In their "real life", these machines are not supposed to swap. Swap space is currently unused, as it should be.
  
• Earlier this year during Yoyo races with ECM tasks, it happened sometimes that the Linux kernel's out-of-memory killer had to take out a few ECM tasks, sadly after they already ran for quite some time. Those tasks were then shown in boincmgr and in the results tables at the Yoyo web site as error during computing. The host's system log would of course show that the OOM killer was engaged.
  This sort of thing did not happen yet with NFS.

On the only Windows box which I currently use for NFS, I don't approach the configured RAM limit. I recently upgraded RAM of this machine for occasions like this, despite the current high RAM prices. It's a single-socket machine with 64 GB RAM installed, 70 % allowed for BOINC. Windows 7 and light desktop usage.

 

[crashtech]

I may have more time to experiment later, but it was only after limiting CPU usage that I was able to regain remote control of the machine. Usually, all Teamviewer needs to run reliably is one thread free, if that. It could be that some other unknown factor is affecting the PC, though.

 

[TennesseeTony]

First hour results are in, but mean little, ....but 1st across all 3 leagues by a factor of 4.

 

[crashtech]

Where does Formula BOINC get its stats from?

 

[Orange Kid]

OK, so sorry for the late stats. Was BBQ'ing and (ok, drinking) and eating.
We were 13th before the sprint started two hours ago and have already picked up one spot. We have an opportunity to gather some more points along the way.
Us old retired people sometimes don't remember the concept of time

NFS@home..............Marathon................Start
1 Team Norway .......................25 ............11,244,378
2 OcUK - Overclockers UK ......18 .............8,370,398
3 LinusTechTips_Team ............15 ............7,683,918
4 AMD Users ............................12 .............5,597,316
5 Overclock.net .........................10 .............5,508,818
6 Sicituradastra. ..........................8 .............5,192,726
7 BOINC.SK ................................6 .............3,282,340
8 meisterkuehler.de Team ..........4 ..............2,541,402
9 US NAVY .................................2 ..............1,769,836
10 team.se ..................................1 ..............1,664,090
11 CANAL@Boinc - ......................................1,179,246
12 TeAm AnandTech - ..................................1,031,362

 

[crashtech]

...Was BBQ'ing and (ok, drinking) and eating...

 

[StefanR5R]

Where does Formula BOINC get its stats from?

I suppose they poll the project sites using web RPC.
https://boinc.berkeley.edu/trac/wiki/WebRpc

Spoiler: example

output from https://escatter11.fullerton.edu/nfs/team_lookup.php?team_id=125

at 00:35 UTC:

<team>

<id>125</id>
<create_time>1252176608</create_time>
<userid>137</userid>
<name>TeAm AnandTech</name>
<url>forums.anandtech.com/forumdisplay.php?f=15</url>
<type>1</type>
<country>International</country>
<total_credit>43162994.829902</total_credit>
<expavg_credit>122486.2131842525</expavg_credit>
<expavg_time>1501201813.567427</expavg_time>​

</team>​

at 00:56 UTC:

<team>

<id>125</id>
<create_time>1252176608</create_time>
<userid>137</userid>
<name>TeAm AnandTech</name>
<url>forums.anandtech.com/forumdisplay.php?f=15</url>
<type>1</type>
<country>International</country>
<total_credit>43218006.829902</total_credit>
<expavg_credit>127675.5672880378</expavg_credit>
<expavg_time>1501203342.331722</expavg_time>​

</team>​

 

[Orange Kid]

So the end of day one of the sprint leads to some surprising changes.
While we started out in first the UK Overclockers have overclocked us and taken over first.
Meanwhile in the marathon we have risen from 13th to 8th earning us 4 point and probably an update away from 7th and two more points.
The UK Overclockers started out in 2nd,but are looking at the first place spot with some glow in their cpu's.
While we have gained 22 points overall the Overclockers have picked up 25. Looks like someone over there did the math and decided that they can't let us win all the sprints and still maintain 1st place overall.

NFS@home..............Sprint................Day One
1 OcUK - Overclockers UK .............25 ........2,728,436
2 TeAm AnandTech .......................18 ........2,482,178
3 Crunching@EVGA .......................15 ........1,181,504
4 Overclock.net ...............................12 ...........622,878
5 meisterkuehler.de Team ..............10 ...........219,270
6 LinusTechTips_Team ....................8 ..............78,000
7 Sicituradastra. ................................6 ..............25,834
8 US NAVY .......................................4 ..............11,698
9 team.se ..........................................2 ..............11,306
10 AMD Users ..................................1 ..............10,322

NFS@home..............Marathon................Day One
1 Team Norway ..............................25 ..........11,244,378
2 OcUK - Overclockers UK .............18 ..........10,597,184
3 LinusTechTips_Team ..................15 .............7,764,908
4 Overclock.net ...............................12 ............6,096,112
5 AMD Users ...................................10 ............5,607,300
6 Sicituradastra. ................................8 .............5,218,400
7 BOINC.SK ......................................6 .............3,285,116
8 TeAm AnandTech .........................4 ..............3,253,240
9 meisterkuehler.de Team ................2 ..............2,736,930
10 US NAVY .....................................1 ..............1,782,546

 

[TennesseeTony]

Thanks Kid!

I spoke for the TeAm at OcUK. https://forums.overclockers.co.uk/threads/formula-boinc-2017.18763633/page-22

User Xez is currently in hospice care, but STILL CRUNCHING!!! At age 28, the doctors have given up hope on his cancer. Stop by there and give your personal words of encouragement if you desire.

 

[indydude345]

Is anyone else having a problem accessing FreeDC?

 

[TennesseeTony]

yes

 

[crashtech]

Looks like OcUK might be on their way to a well-deserved win. It's actually a little more fun this way, when we take first by a mile it's a bit of an anticlimax.

 

[StefanR5R]

OCN found some weird scheduler decisions with 14e and 15e units, ending in tasks "completed, too late to validate".

Sigh, whats up with NFS sending out two backup units after someone aborts, so 3 in total, and then canceling the 3rd unit mid progress when the 2nd completes:
https://escatter11.fullerton.edu/nfs/workunit.php?wuid=81883138

Or even two at once with quorum = 1 and aborting the 2nd unit:
https://escatter11.fullerton.edu/nfs/workunit.php?wuid=81889752

No more 14e/15e. That's just nutty.

I saw a strange combination of abandonment/ cancellation/ late completion with a 16e unit:
https://escatter11.fullerton.edu/nfs/workunit.php?wuid=80745181

• 1st task sent, not completed by client before 7 days deadline
• 2nd task sent, not completed by client before 7 days deadline
• 3rd task sent, and 1.5 hours later "abandoned", and a 4th task sent
• 2nd task completed and validated 3 hours after deadline (i.e. NFS has a grace period), 4th task "cancelled by server"

 

[indydude345]

As of current time, we have surpassed OcUK.
http://formula-boinc.org/sprint.py?sprint=11&lang=&year=2017

 

[StefanR5R]

Encouraged by the performance of my Phenom II X4 @ 2.5 GHz (1h27m with 16e tasks), I now had my Core 2 T7600 @ 2.33 GHz enter the race too. 1h52m with 16e tasks, and 2 simultaneous tasks of the 32 bit application can be supported with just 2 GB RAM and 32 bit Linux. I haven't noticed any "waiting for memory" situation, but I haven't watched it frequently.

 

[crashtech]

We are neck and neck down the straightaway; I've had every available machine involved since Thursday, so more more help here.

 

[ocukguest]

going to be a close one this time.

 

[crashtech]

going to be a close one this time.

Those are the best kind, imo.

 

[ocukguest]

yes I have to agree,. it adds much more competition and makes everybody on edge lol

 

[Orange Kid]

Day two, I know late again, what can I say, it's summer. Gotta go out and play sometime.

We are still in second in the sprint, but earlier I saw us in first, so it has become a seesaw battle

Meanwhile in the marathon we have move up to 7th and stand to gain another spot or two before this one is over. The Overclockers UK have take the top spot in the Marathon so congrats to them on that move.

NFS@home..............Sprint................Day One
1 OcUK - Overclockers UK .............25 ........2,728,436
2 TeAm AnandTech .......................18 ........2,482,178
3 Crunching@EVGA .......................15 ........1,181,504
4 Overclock.net ...............................12 ...........622,878
5 meisterkuehler.de Team ..............10 ...........219,270
6 LinusTechTips_Team ....................8 ..............78,000
7 Sicituradastra. ................................6 ..............25,834
8 US NAVY .......................................4 ..............11,698
9 team.se ..........................................2 ..............11,306
10 AMD Users ..................................1 ..............10,322

NFS@home..............Sprint................Day Two
1 OcUK - Overclockers UK ..............25 ...........4,749,330
2 TeAm AnandTech ........................18 ...........4,302,112
3 Crunching@EVGA ........................15 ...........2,135,560
4 Overclock.net .................................12 .............990,186
5 meisterkuehler.de Team ................10 .............534,638
6 LinusTechTips_Team .....................8 ..............163,150
7 Sicituradastra. ................................6 .................51,338
8 team.se ..........................................4 .................26,606
9 AMD Users ....................................2 .................26,014
10 US NAVY .....................................1 .................21,676

NFS@home..............Marathon................Day Two
1 OcUK - Overclockers UK ..............25 ............12,320,006
2 Team Norway ...............................18 .............11,244,378
3 LinusTechTips_Team ...................15 ...............7,844,468
4 Overclock.net ...............................12 ...............6,448,864
5 AMD Users ...................................10 ...............5,620,752
6 Sicituradastra. ................................8 ...............5,235,594
7 TeAm AnandTech .........................6 ...............4,988,678
8 BOINC.SK .....................................4 .................3,285,246
9 meisterkuehler.de Team ................2 ................3,011,904
10 Crunching@EVGA ......................1 .................2,082,072

 

[crashtech]

Here are some results that my machines have been turning out. It seems that other factors are having an influence. Perhaps adding RAM speed and amount to the chart would help illuminate some seeming discrepancies.



NFS@Home, 16e Lattice Sieve V5 v1.11, windows_x86_64
Average CPU Time of last 10 Valid WUs, to nearest second:

CPU _______ Family _ Threads/Core _ Speed _________ RAM ___________ Time

i5-6500 ___ Skylake ______ 1 _____ 3.2GHz _ 08GB DDR4 2133 CAS 15 _ 2402
G3420 _____ Haswell ______ 1 _____ 3.2GHZ _ 16GB DDR3 1600 CAS 9 __ 2569
G4560 _____ Kaby Lake ____ 2 _____ 3.5GHz _ 08GB DDR4 2133 CAS 15 _ 3077
i7-3930K __ Sandy Bridge _ 2 _____ 4.0GHz _ 16GB DDR3 1333 CAS 11 _ 3209
i3-4370 ___ Haswell ______ 2 _____ 3.8GHz _ 32GB DDR3 1600 CAS 8 __ 3299
1700X _____ Ryzen 7 ______ 2 _____ 3.4GHz _ 16GB DDR4 3200 CAS 14 _ 3485
E3-1230v5 _ Skylake ______ 2 _____ 3.6GHz _ 08GB DDR4 2133 CAS 15 _ 3559
i3-2120 ___ Sandy Bridge _ 2 _____ 3.3GHz _ 08GB DDR3 1333 CAS 11 _ 4346
W3670 _____ Westmere _____ 2 _____ 3.2GHz _ 12GB DDR3 1333 CAS 11 _ 4755
X5670 _____ Westmere _____ 2 _____ 2.9GHz _ 12GB DDR3 1333 CAS 9 __ 4765
E5-2660v2 _ Ivy Bridge ___ 2 _____ 2.6GHz _ 64GB DDR3 1066 CAS 7 __ 6565
________________________________________________________________________

 

[petrusbroder]

Thanks for all the numbers!

 

[StefanR5R]

Here are some results that my machines have been turning out. It seems that other factors are having an influence. Perhaps adding RAM speed and amount to the chart would help illuminate some seeming discrepancies.

Interesting how the 8C/16T Ryzen 7 holds its own against the 4C/8T Skylake Xeon E3, while both have dual channel RAM controllers. Did you provide the Ryzen system with lower latency RAM than the Xeon E3?

 

[Orange Kid]

Congrats to OcUK for taking First in the Sprint as well as moving into First in the Marathon.
Congrats to the TeAm on a well fought battle and a fine Second place finish.
Crunching@EVGA ended in third and also came from nowhere to the 10th spot in the marathon, well done.
We added 28 point to our total in this sprint as we had none in the marathon before the race.

NFS@home..............Sprint................Day One
1 OcUK - Overclockers UK .............25 ........2,728,436
2 TeAm AnandTech .......................18 ........2,482,178
3 Crunching@EVGA .......................15 ........1,181,504
4 Overclock.net ...............................12 ...........622,878
5 meisterkuehler.de Team ..............10 ...........219,270
6 LinusTechTips_Team ....................8 ..............78,000
7 Sicituradastra. ................................6 ..............25,834
8 US NAVY .......................................4 ..............11,698
9 team.se ..........................................2 ..............11,306
10 AMD Users ..................................1 ..............10,322

NFS@home..............Sprint................Day Two
1 OcUK - Overclockers UK ..............25 ...........4,749,330
2 TeAm AnandTech ........................18 ...........4,302,112
3 Crunching@EVGA ........................15 ...........2,135,560
4 Overclock.net .................................12 .............990,186
5 meisterkuehler.de Team ................10 .............534,638
6 LinusTechTips_Team .....................8 ..............163,150
7 Sicituradastra. ................................6 .................51,338
8 team.se ..........................................4 .................26,606
9 AMD Users ....................................2 .................26,014
10 US NAVY .....................................1 .................21,676

NFS@home..............Sprint................Day Final
1 OcUK - Overclockers UK ..............25 ............6,624,922
2 TeAm AnandTech ........................18 .............5,819,744
3 Crunching@EVGA ........................15 ............2,870,378
4 Overclock.net ................................12 ............1,229,488
5 meisterkuehler.de Team ...............10 ...............660,182
6 LinusTechTips_Team .....................8 ...............235,950
7 Sicituradastra. .................................6 .................69,406
8 AMD Users ......................................4 .................37,352
9 team.se ...........................................2 ..................37,122
10 US NAVY ......................................1 ..................27,892


NFS@home..............Marathon
1 OcUK - Overclockers UK ..............25 ................14,818,370
2 Team Norway ...............................18 .................11,244,378
3 LinusTechTips_Team ....................15 ..................7,923,118
4 Overclock.net ................................12 ...................6,724,666
5 TeAm AnandTech .........................10 ...................6,633,414
6 AMD Users .....................................8 ....................5,633,574
7 Sicituradastra. ................................6 .....................5,261,636
8 BOINC.SK ......................................4 .....................3,285,246
9 meisterkuehler.de Team ................2 ......................3,196,104
10 Crunching@EVGA .......................1 ......................2,898,412

and I was a little more timely in posting

 

[TennesseeTony]

Dag-nab-it. 2nd means "first loser". Well done all.