KWSN Orbiting Fortress

[Yankton]

Ran one (not an SR5, but the normal riesel. took 7.5 hours. Noted also in preferences that AVX is only for genefer. Apparently windows app can't detect it yet so you have to force it. It is available for the llr projects as a seperate download according to the forums. Is this what you did to get it running?
_________________
Some days are worse than others.

[Pooh Bear 27]

LLR wrapper is much different than GFN. The wrapper has it built in, so you get the same wrapper for AVX, and non-AVX. In the first seconds of the start up it checks to see what you have and uses the best.

Plus you saw you ran it in 7.5 hours. The estimate is 19+ hours. A real difference!
_________________
Watch the movie about me The 4-1-4s: The Original Teenage Hackers (It is only about 12 minutes long.)
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1

[LanDroid]

PPS (Sieve) 1.39 (cudaPPSsieve)

Dang it, got a bunch of the wrong type of WU. Aborted those, selected the correct one, but now of course I'm getting this message again:

10/4/2013 5:27:02 PM | PrimeGrid | Not requesting tasks: project is not highest priority

Having a rough start as usual...

[Gemjunkie]

[Plomos]

Plus it needs be TRP LLR WUs not PPS.......
_________________

[Gemjunkie]

Yes, it even has "World Animal Day Challenge project" right next to it in preferences.
_________________




(older, before split CPID)

[LanDroid]

Yes, didn't realize I had TWO sub-projects checkmarked by accident. Fixed that and now finally shrubbing the good stuff: The Reisel Problem (LLR) 6.15. Update: Looks like I'll be a slow shrubber - 5+ hours in with 15+ hours to go?

AND at last I got me badge sig updated!


_________________
[img]http://stats.free-dc.org/cpidtagb.php?cpid=6533a276b3a1dde393be350eb3cfda70&theme=16&cols=5/.png[/img]
[img]http://stats.free-dc.org/badgesbanner.php?cpid=6533a276b3a1dde393be350eb3cfda70/.png[/img]

[Yankton]

The WUs for the challenge may be a bit larger then what was sent out before? Also I had let PPS sieve units run on the gpu. claims .1 CPU usage, but task manager shows MUCH higher CPU usage. .5 - .6. Turned off nvidia in primegrid. Rigs are also running cooler with this LLR then with other projects. Also rather counter-intuitive. I always though prime searches ran hotter.

Very wierd
_________________
Some days are worse than others.

[Pooh Bear 27]

It looks like they may have found a prime and illuminated another k. The first tasks I did were in the 1100 credit range, the newer ones are near the 1400 credit range. Larger numbers mean longer crunch times. Looks like added 3-4 hours on the regular processors and 1-2 on AVX.
_________________
Watch the movie about me The 4-1-4s: The Original Teenage Hackers (It is only about 12 minutes long.)
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1

[Gemjunkie]

A lot of mine are taking >24hrs. I need some new hardware.
_________________




(older, before split CPID)

[Pooh Bear 27]

The new FFT is making them run quite a bit longer (the FFT changed during the early part of the challenge because the numbers we are crunching called for it). My C2Q is running them at around 27 hours, so I can see that. My AVX enabled machines are running about 10.5 - 11 hours. So AVX has a huge advantage. My non AVX i7s are at between 21 and 24 hours.

At 2 days in, we have 12 knights with credit. 3 are in top 100. Currently I am in top of our team at 50th position out of over 575 participants. As a team we've been flirting between 12th and 14th positions.

This is an LLR, cleanup will take a while. Scores will change as people who sent back bad units get their scores reduced. Last big one we were in all our team members never lost a point (we all seem to know how to keep stable machines), and we gained several spots in the end.

Good showing team!
_________________
Watch the movie about me The 4-1-4s: The Original Teenage Hackers (It is only about 12 minutes long.)
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1

[Yankton]

The length my new machine takes for them is about triple the one I did before the challenge started. Digging through the forums it appears that running loads of LLRs at the same time on hyper threaded machinges causes a HUGE slowdown. An issue with LLRs using large amounts of CPU cache and cache misses (constant cache misses) due to hyper threading. I want to test this but not sure how without killing my challenge input entirely. I'd just reduce it to 4 instances but I don't think there's a way to restrict the 4 from running two instances on one physical core. joy joy. I've noticed it also clocks back to base speed when running 8 simultaneously. Which is why it's running cool right now. Let it clear it's cache of WCG units and it ramped back up and had them no issues.



Edit: As soon as the current gpugrid unit finishes I'm going to shutdown and disable hyper threading entirely. I only left it on as it looks pretty. Should clear this up.
_________________
Some days are worse than others.

[Pooh Bear 27]

Actually set it to 50%. It will do very well. I do my HT this way and get the times just fine.
_________________
Watch the movie about me The 4-1-4s: The Original Teenage Hackers (It is only about 12 minutes long.)
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1

[Yankton]

I'm not used to having hyper-threading. The only option I remember in the bios was to simply enable/disable it. How do you set it to 50%? I'll give it a whirl if I can. I was thinking of disabling it anyway next week to do a full overclock on it. Should get another .2 Ghz out of it at the same temps I'm running now.
_________________
Some days are worse than others.

[Pooh Bear 27]

Go to your Prefernces page and set On multiprocessors, use at most Enforced by version 6.1+ to 50%
_________________
Watch the movie about me The 4-1-4s: The Original Teenage Hackers (It is only about 12 minutes long.)
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1

[Yankton]

I gave that a shot. No joy. Last night I set boinc local preferences to 50% and that worked. It's ignoring project specific though. I zonked out before the gpugrid task finished. It was a doozy. It also apparently crashed near the end with an access violation. Looking at the gpugrid forums people are reporting that the new beta 331.40 drivers corrects the problem. I guess I'll be installing beta drivers later also.
_________________
Some days are worse than others.

[Gemjunkie]

I spotted the long-lost Wabbit!
479 Wabbit98 The Knights Who Say Ni! 9396.67
_________________




(older, before split CPID)

[Plomos]

Have hit my entry level badge for this subproject and still going strong Ni!!!
_________________

[Pooh Bear 27]

Announcement of a prime has been stated. I am hoping a Knight is either the finder or double checker. It would be worth a lot of prime points.
_________________
Watch the movie about me The 4-1-4s: The Original Teenage Hackers (It is only about 12 minutes long.)
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1

[Pooh Bear 27]

To have my fastest cruncher lock up for about 7 hours while I slept makes a Pooh Bear very upset. That is equivalent to losing 4 units being finished in time for the end of the challenge. It also errored one GPU unit it was working on, right near it's end, so almost 8 ours of lost crunch time there, also.

It looks like once said and done I will probably finish somewhere in the 60s or even 70s. I see Grebuloner has now passed me. As a team we dropped to 14, but are sticking close to 13th. Time will tell. We have just over 28 hours to the finish. If you haven't gotten in yet, you probably can get 1-3 per core finished yet (1 with Core2/i7 first gen or AMD, 2-3 with i7 2nd gen and above using AVX).
_________________
Watch the movie about me The 4-1-4s: The Original Teenage Hackers (It is only about 12 minutes long.)
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1