4 days and counting...

**Lowe** · 29-01-2008, 08:04 AM

Bit of a personal one for me here I guess, but it's time to get the e-peen out.

1 millllllllllion dollars... err, points even.

Predictions are showing it as 4 days worth of crunching, so all being well I'll be a Hexus Folding millionaire.

Party anyone?

tiggerai · 29-01-2008, 12:42 PM

will there be wine?

**directhex** · 29-01-2008, 01:56 PM

so, i have 4 days to beat you?

420,000 points in 4 days means 60.3 SMP WUs a day

it's gonna be CLOSE

0iD · 29-01-2008, 02:14 PM

Anyone got odds on this one

**Lowe** · 29-01-2008, 03:13 PM

I know you've got a super cluster hiding away there Jo, but can you get everything up and running in time? If you can, bring it pal - competition is a good thing

0iD · 29-01-2008, 03:15 PM

The gauntlet has been cast down Hex

**directhex** · 29-01-2008, 03:20 PM

3 machines running as of now.

well, that's a lie. 3 smp instances on a 16-core opteron box. job has a 4-day limit set.

**directhex** · 29-01-2008, 04:17 PM

5 instances running

tiggerai · 29-01-2008, 04:18 PM

oooh the race is on!

Muahahaha

Anyone wanna place some bets?

**directhex** · 29-01-2008, 04:25 PM

Code:

jms@konrad:~$ rsh comp02 ps -ef | grep FahCore_a1.exe
jms       8892  8835  0 16:22 ?        00:00:00 ./mpiexec -np 4 -host 127.0.0.1 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8835 -version 600
jms       8893  8892 85 16:22 ?        00:02:05 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8835 -version 600
jms       8894  8892 93 16:22 ?        00:02:18 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8835 -version 600
jms       8895  8892 63 16:22 ?        00:01:34 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8835 -version 600
jms       8896  8892 39 16:22 ?        00:00:58 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8835 -version 600
jms       8898  8825  0 16:22 ?        00:00:00 ./mpiexec -np 4 -host 127.0.0.1 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8825 -version 600
jms       8899  8898 70 16:22 ?        00:01:45 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8825 -version 600
jms       8900  8898 82 16:22 ?        00:02:01 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8825 -version 600
jms       8901  8898 51 16:22 ?        00:01:16 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8825 -version 600
jms       8902  8898 33 16:22 ?        00:00:49 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8825 -version 600
jms       8904  8845  0 16:22 ?        00:00:00 ./mpiexec -np 4 -host 127.0.0.1 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8845 -version 600
jms       8905  8904 75 16:22 ?        00:01:51 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8845 -version 600
jms       8906  8904 81 16:22 ?        00:02:00 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8845 -version 600
jms       8907  8904 58 16:22 ?        00:01:25 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8845 -version 600
jms       8908  8904 35 16:22 ?        00:00:52 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 8845 -version 600
jms@konrad:~$ rsh comp01 ps -ef | grep FahCore_a1.exe
jms      10336 10312  0 16:22 ?        00:00:00 ./mpiexec -np 4 -host 127.0.0.1 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 10312 -version 600
jms      10337 10336 73 16:22 ?        00:01:51 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 10312 -version 600
jms      10338 10336 82 16:22 ?        00:02:05 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 10312 -version 600
jms      10339 10336 55 16:22 ?        00:01:25 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 10312 -version 600
jms      10340 10336 33 16:22 ?        00:00:51 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 10312 -version 600
jms      10342 10322  0 16:22 ?        00:00:00 ./mpiexec -np 4 -host 127.0.0.1 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 10322 -version 600
jms      10343 10342 74 16:22 ?        00:01:53 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 10322 -version 600
jms      10344 10342 81 16:22 ?        00:02:03 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 10322 -version 600
jms      10345 10342 56 16:22 ?        00:01:25 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 10322 -version 600
jms      10346 10342 35 16:22 ?        00:00:54 ./FahCore_a1.exe -dir work/ -suffix 01 -checkpoint 15 -lifeline 10322 -version 600

these are machines from the old, decommissioned SMP service

i STILL don't have working clusters (rargh!)

**directhex** · 29-01-2008, 05:11 PM

Code:

jms@konrad:~$ tail -1 submit.sh.o672*
==> submit.sh.o6724 <==
[16:59:51] Completed 10000 out of 500000 steps  (2 percent)

==> submit.sh.o6725 <==
[16:56:45] Completed 10000 out of 500000 steps  (2 percent)

==> submit.sh.o6726 <==
[16:56:51] Completed 10000 out of 500000 steps  (2 percent)

==> submit.sh.o6727 <==
[17:00:23] Completed 10000 out of 500000 steps  (2 percent)

==> submit.sh.o6728 <==
[16:58:28] Completed 10000 out of 500000 steps  (2 percent)

20 minutes per percentage point - "4-way" opteron 1.8ghz is about as fast as 2-way 2.13ghz core2 for f@h

**directhex** · 29-01-2008, 05:14 PM

Originally Posted by Lowe

I know you've got a super cluster hiding away there Jo, but can you get everything up and running in time? If you can, bring it pal - competition is a good thing

the INSTANT i get a proper image pushed out to the nodes, i'm good to go. 128 quad machines POSSIBLY pulling an all-nighter as of tomorrow, another 128 later (though those'll be too late for this little contest)

**Zak33** · 29-01-2008, 11:10 PM

COME ON.... who'd have ever though of a race for a million?

Funkstar · 29-01-2008, 11:12 PM

I'm just pleased that both my machines were still folding after leaving them on their own for almost two weeks

**directhex** · 29-01-2008, 11:24 PM

Originally Posted by Zak33

COME ON.... who'd have ever though of a race for a million?

race to a million? meh

race to earn 418k in the time it takes Lowe to earn 30k? now that's where it starts sounding silly

and after some stupid config errors, those hexadeca-core machines are working away, with 6 smp jobs running between them. it's not wasted time, it's allowed me to craft a zero-config job schedule submission script which should work with minimal changes on the clusters

**Lowe** · 30-01-2008, 09:10 AM

I'm feeling the pressure - home machines are now 24/7 (3.1ghz C2D on SMP and PS3) and I've even added my Macbook Pro now (something I don't like doing since it burns my hands when typing lol!) to squeeze a few more units out.