Rally Nov. 21st-23rd

[gd_barnes]

To all,

We're planning another LLRnet rally next weekend from Nov. 21st at 7 PM GMT to Nov. 23rd at 7 PM GMT (2 PM EST / 11 AM PST U.S.).

The rally will be run mostly on port 400 (for the heavier hitters) and port 443 (for smaller resources). We may also reactivate port 5000 depending on the resources running the rally. (Note by Max: G4000 is also now open for rally work. I'm guessing that it can handle a pretty heavy load, though it hasn't been tested and thus the primary load should go on IB400 unless we can get a good pre-rally load test going to verify what it can handle.)

If you can't run our usual project servers and wish to run your own, let us know how many cores you'll be running and we'll send you a range just for the rally.

Max, can you post a response that asks for and shows the cores that everyone will be bringing. Right now, I'm going to plan on a full 40 fast cores. (I may run 3-4 slow cores also.)

Bring on the firepower as we attempt to finish the 1st drive by year end!


Gary

[Mini-Geek]

I'll put my two cores on C443.

[mdettweiler]

Okay, here's a table keeping track of who's on what server:

User          # of cores      Server
------------------------------------
gd_barnes     40 fast         IB400
Mini-Geek     2 slower        C443
mdettweiler   2 fast          G4000(1)/C443(1); possibly will add 4 more fast cores later
MyDogBuster   14 fast; 7 slow IB400
em99010pepe   4 fast          C443
glennpat      4 fast          C443

Total on IB400: 61
Total on C443:  11
Total on G4000:  1

As represented in the table, I'll bring two fast cores, one each on G4000 and C443. I also would really like to bring my quad, but it's not quite ready yet and as such I'm not putting it down as definite at this time.

Note that I mentioned G4000 as one of the servers I'm using for the rally. With my completion of the scripts for CSV and results processing stuff last night, I'm comfortable with officially opening G4000 for business. It's running on a Linux server, so it should be able to handle a decent number of clients (though only time will tell whether it's quite as stable as David's rock-solid servers ). Its status web page and results files can be found at http://nplb-gb1.no-ip.org/llrnet/, and as of last night it was all clear for CSV importing too.

So, if anyone wants to put some cores on G4000 for the rally, go right ahead--it should have plenty of work so that it doesn't dry out even with some "business" from others.

Max

[mdettweiler]

Hey, I just thought of an idea about G4000: since right now we're kind of in the dark as to how much load it can safely handle, why don't we run a pre-rally load test on it? Maybe we could all try to put as many cores as we can on G4000 for, say, 12 hours, and at least get a minimum figure as to how much we can put on it before it breaks.

For those who are interested, here's the specs for the server G4000's on:
-AMD Phenom 9750 quad-core CPU
-2 GB RAM
-running Ubuntu 7.04 Desktop 32-bit (yeah, it's old, that's a long story. We'll upgrade it some time or other, especially now that Ubuntu stopped supporting 7.04. )
-lighttpd web server for the status page stuff

The LLRnet server has a jobMaxTime of 5 days, and a prunePeriod of 15 minutes.

Max

[Mini-Geek]

Originally Posted by mdettweiler

Mini-Geek     2 (unknown)     C443

My cores are my two cores of a 2.5 GHz Athlon X2 (as usual). You usually mark them "slower". Last time you listed them as "slow". Call 'em whatever you want. They take about 760 CPU seconds (around 12.5-13 minutes) to finish a number at n=~547K (that's one core).

[mdettweiler]

Originally Posted by Mini-Geek

My cores are my two cores of a 2.5 GHz Athlon X2 (as usual). You usually mark them "slower". Last time you listed them as "slow". Call 'em whatever you want. They take about 760 CPU seconds (around 12.5-13 minutes) to finish a number at n=~547K (that's one core).

Okay, I've marked them as "slower". Based on those timings, that's probably the closest approximation of their speed.

The "fast/slow" approximations are actually somewhat subjective; hence why the same computer has been listed both as "slow" and "slower" in previous rallies. Anyway, the speed of the individual cores are not really that much of a factor in the server load unless you've got a whole load of them (i.e., 40 Core 2 cores are going to produce significantly more server load than 40 AMD X2 cores, though 2 Core 2 cores vs. 2 AMD X2 cores aren't really going to be that much of a difference).

[MyDogBuster]

Just added more cores this AM.

I will have 14 fast cores and 7 slow cores. All are on IB400. Tell me where you want them.

[mdettweiler]

Originally Posted by MyDogBuster

Just added more cores this AM.

I will have 14 fast cores and 7 slow cores. All are on IB400. Tell me where you want them.

Okay, I've marked you down in the table; IB400 should be good.

Gary, we've got 61 cores on IB400 at this point. At which point do you think we should start telling users to begin putting their cores on G4000 or C443 instead? We've been able to support upwards of 100 cores on IB5000, so presumably IB400 can handle a similar load; however, we probably want to make sure that, especially during a rally, we stay somewhat under that limit, especially considering as how we have plenty of servers anyway.

[em99010pepe]

4 fast on C443

(321 seconds per candidate at n=543k)

[mdettweiler]

Originally Posted by em99010pepe

4 fast on C443

(321 seconds per candidate at n=543k)

Hmm, with timings like that, those are not only fast, but on the faster end of fast! How much do you have that thing overclocked again?

[em99010pepe]

Originally Posted by mdettweiler

Hmm, with timings like that, those are not only fast, but on the faster end of fast! How much do you have that thing overclocked again?

Only at 2.96 GHz, can't go much higher. You can ask Flatlander about his quad timings.

Edit: 321-325 secs between all cores with llr 3.7.1c.