Badge hunting for both hpf2 and gfam. efficiency drive.

[OldChap]

Well done One.shot and team.

Starting to see why the disparity in completion times ....

My core 2 quad seems happy to complete in 3.3 hours for the couple of wu's I tried. This is not dissimilar to completion times for GFAM

Aracnid, on the other hand, looks set to take between 6.6 and 11.2 hours per wu which is waaaay longer than for GFAM. Are there "double" or maybe just variable length wu's being issued???

[Rob_B]

Are those both Linux crunchers? I had a E6600 (2 cores) running Linux for the last week and that crunched the same number of wu's each day as my i3-540 (4 threads) that is on Win7.
Looks like GFAM is doing great on Linux.

Congrats One.shot Just a few more days with pie left for you.

[Mumak]

Hyperthreading maybe ?

[Rob_B]

No, the other way round. The E6600 doesn't have HT. And the dual core i3 does. So the E6600 is doing the same works with half the cores. HT doesn't double the runtime, but it gets close.
And the E6600 is at 2.4 GHz and the i3 at 3.0 GHz.

[yojimbo197]

Are those both Linux crunchers? I had a E6600 (2 cores) running Linux for the last week and that crunched the same number of wu's each day as my i3-540 (4 threads) that is on Win7.
Looks like GFAM is doing great on Linux.

Congrats One.shot Just a few more days with pie left for you.

If I"m remembering right GFAM, DFSL, and SNS all use Vina which works more efficiently on Linux boxes. I see between a 1-2 hour decrease in run time for those projects on my either my double dual Core X5260 workstation @3.3Ghz or my Core 2 Quad Q9550(2.83Ghz) vs. the i5 2320(3.0Ghz) Windows 7 box.

[OldChap]

*ALL* ViNA sciences in 64 bit are way way faster on Linux than Windows-64. so maybe I am seeing the effect of dropping off a vina project whilst running Linux.

I believe that I read that HT on gives maybe 40% increase in production so maybe this rig running ht processes at 70% for each core which combined with the slow clocks.....

All a moot point of course when looking for days not points.

The only effect as I see it is that I have work for the full 10 days...... and probably don't need it

[Mumak]

No, I wanted to point out that the i3-540 has 2 cores as well + HT. So depending on type of workload + during full load on all threads you might see nearly no performance gain from HT (it might produce similar as without HT).

[Rob_B]

No, I wanted to point out that the i3-540 has 2 cores as well + HT. So depending on type of workload + during full load on all threads you might see nearly no performance gain from HT (it might produce similar as without HT).

Oke. I misunderstood. Never seen that on any project, but It could happen ofcourse.

[Rob_B]

The project chart is showing just one day left for HFP2.

[Mumak]

I suppose the WCG techs will post a message when the end is near.
What makes me worry is that the DSFL remaining times goes down rapidly.. Might be problematic to catch badges soon.

[OldChap]

Not that much reduced time surely?

Needing a full 2 years, if you started as soon as you could after these badges you should be able to get sapphire in 37 days when running 20 cores

You would only be in trouble if there were < 30 days remaining

[Mumak]

Yeah, I hope so. But if you follow SekeRob's chart, maybe you noticed that a few days ago DSFL remaining time was ~120 days, now it's 97.
Also this: https://secure.worldcommunitygrid.or...ad?post=423810

[bluestang]

Well I have my DSFL Sapphire badge, so I would be more than happy to toss my 8 and 4 core rigs to help someone out.

First to ask, gets them!

[0ne.shot]

Congrats One.shot Just a few more days with pie left for you.

Thanks. With my own machines I can average around 250 hrs/day. With everyone on my account, we broke 2000! I'm also starting to put on a few pounds with all of that pie. Maybe a diet is in store.

[0ne.shot]

Yeah, I hope so. But if you follow SekeRob's chart, maybe you noticed that a few days ago DSFL remaining time was ~120 days, now it's 97.
Also this: https://secure.worldcommunitygrid.or...ad?post=423810

Hopefully 97 is a solid number and doesn't go down rapidly.

[Mumak]

Morning (UTC) update:

GFAM - 1:353, 650 InP, 32 PV
HPF2 - 0:064, 4335 InP, 737 PV

So with the next update in 7 hours I should reach GFAM sapphire
I'm going to abort most GFAM WUs.

[D_A]

Well I have my DSFL Sapphire badge, so I would be more than happy to toss my 8 and 4 core rigs to help someone out.

First to ask, gets them!

I have four cores here as well that can help someone with a DSFL deficiency. I'm pure 64bit though, so I can't help anyone with HPF2 or any other 32bit projects.

[Mumak]

GFAM Sapphire !
HPF2: 0:100

You can dump all GFAM tasks now. HPF2 work is still flowing in, but probably won't last long.

Big thanks to you all !! You're a great team

[bluestang]

Mumak....congrats on the GFAM Sapphire.

Now on to the HPF2 Sapphire!

[bluestang]

The last line is something I've never seen or maybe just never paid attention to before:

6/10/2013 11:35:22 AM | World Community Grid | Requesting new tasks for CPU
6/10/2013 11:35:25 AM | World Community Grid | Scheduler request completed: got 0 new tasks
6/10/2013 11:35:25 AM | World Community Grid | No tasks sent
6/10/2013 11:35:25 AM | World Community Grid | No tasks are available for the applications you have selected.
6/10/2013 11:35:25 AM | World Community Grid | Tasks won't finish in time: BOINC runs 86.4% of the time; computation is enabled 99.9% of that

[Mumak]

I've seen that message when my caches are too large and buffer more work than they can process according to deadlines.

[bluestang]

Aah, now that makes sense to me.

[Mumak]

My estimation is that there is approximately 12 hours of new HPF2 work remaining from now.

[OldChap]

What do you make of this then????

such a difference in runtimes???

[Mumak]

What tasks? Must be different ones, you might check their Properties for estimated flops if they are different.