Badge hunting for both hpf2 and gfam. efficiency drive.

[Rob_B]

I will keep my 2600k on his account for as long as he needs it.
At the moment I have 432 wu's on that rig. That is 7.5 days on 8 threads giving a total of 60 days in cache.

I need a few more days of GFAM and after that I may have a few more cores to spare.

EDIT: 255 wu's of those have an end date of June 7th. So those would be finished on time. I got a little worried because all wu's started running on High Priority yesterday.

[0ne.shot]

So there are many systems on my device manager, but I have no idea to see what the rigs are and thread count. So come clean, fellas. Can I get an idea of thread count per machine? I have rigs from Yerloon, Yojimbo197, Rob_B, and OldChap on my account currently. 12 threads on my end here.

[Rob_B]

I see 5 rigs on your account. HERE

i7-2700k 8 threads - yours?
A6-3650 4 cores - yours?
i7-2600k 8 threads - mine
32 threads - OC
i5-2320 4 threads - ?

New rigs are show if they have validated results before the last daily update.

Come on guys, tell him what your helping with.

[OldChap]

I am looking on free dc http://stats.free-dc.org/stats.php?p...cg&name=681782

There I see from the rig id's
http://stats.free-dc.org/stats.php?p...hostid=2173046
http://stats.free-dc.org/stats.php?p...hostid=1781589

both of which I think are yours.

Then

http://stats.free-dc.org/stats.php?p...hostid=2184836 =RobB I guess
http://stats.free-dc.org/stats.php?p...hostid=2378672 =Mine
http://stats.free-dc.org/stats.php?p...hostid=2376521 =I don't know but yojimbo referred to an i5

slow

This list seems to say that either yerloon or yojimbo did not make it on to your account unless done in the last few hours

[yojimbo197]

OC I think my i5 rig is the last one you listed, since its an i5-2320 running Windows Home 7 64 bit. I've only been up and running on One Shots account for about a day now, so my stats may not be totally reported. I know that my computers has done work for him, because I've seen a long line of completed WU's.

[OldChap]

Makes sense. Free-DC is usually only updating every 12 hours.

The 5 rigs listed should do it by my calculation unless new work ends too soon (that is if it ends in <2 days)

Here's hoping

Just an aside: I wonder, when I see a rig go into high priority mode, if it has calculated that some wu,s will not complete in 10 days. If that were the case then, in theory at least, I have 10 days work here.

[bluestang]

33 more days today

Human Proteome Folding - Phase 2
1:191:20:21:41

That leaves 173 days - 30 days PVs = 143 days.
35 cores (Snow Crash 11 + fallwind 12 + my 12)

[bluestang]

Dayys left:

GFAM -- 3 days
HPF2 -- 10 days

Source:
http://i137.photobucket.com/albums/q...Pi1Project.png

[jeanguy2]

33 more days today

Human Proteome Folding - Phase 2
1:191:20:21:41

That leaves 173 days - 30 days PVs = 143 days.
35 cores (Snow Crash 11 + fallwind 12 + my 12)

Bluestang, what formula do you use to change "Pending Valuation" WUs into runtime days? Where does the above 30 days PVs come from? How secure is the final valuation?

Thanks...

[bluestang]

I've been using 4 hrs avg per WU and it has been pretty accurate for my PVs.

So, right now I have 12 pages + 5 PVs = 185 WUs x 4 hrs avg = 740 hrs / 24= ~30 days.

[0ne.shot]

Update #2

Okay, currently I'm at: GO Fight Against Malaria 0:293:05:11:50


05/31/2013 0:044:22:21:28

06/01/2013 0:022:12:58:34

[Rob_B]

Just 2 more days left for GFAM. Time to make sure you have a full cache. That way we should just make it to sapphire.

[OldChap]

day 1: 0:044:22:21:28
day 2: pending, should be higher than yesterday, How many pages of PV's?

This current batch are taking me >5 hours sometimes 6 I have 1360 wu's in cache = 8.9 days @ 5 hours per. = 285 days toward the 437 day target

Just changed my cache settings to 9 days to try to ensure that even if I lose a few out of time (possible because the rig has been running in priority mode since I ran benchmarks) I now have more (96= about 20 target days) that have a deadline a day later than those others.

It is going to be tough to get the timing right for the change to HPF2. I will try to build out another 3770K just in case.

[bluestang]

Only 25 days today

Human Proteome Folding - Phase 2
1:217:04:04:12

That leaves 148 days - 35 days PVs = 113 days

35 cores running (Snow Crash 11 + fallwind 12 + my 12)

[0ne.shot]

Update #3

Okay, currently I'm at: GO Fight Against Malaria 0:321:14:21:560


05/31/2013 0:044:22:21:28

06/01/2013 0:050:11:18:21

I have 6 pages of PVs

[Mumak]

GFAM: 1:097
doing ~30 days/day, 4 pages of PV

[Rob_B]

GFAM: 1:097
doing ~30 days/day, 4 pages of PV

That looks good, Mumak. With a maximum cache on all rigs you should make sapphire in 8-9 days.

[Mumak]

Yep, though still depends on other factors like possible system outage, or not being able to process the 10-day cache on time.

[OldChap]

When it comes to processing the cache "on time" the theory is that if you have enough work in cache then all is good... However, ....If your wu's in cache are all at the same deadline and that deadline is in say 7 days time then some will get sent back unfinished.

I wonder then if it might be a good idea to abort a number of the earlier deadline wu's now and then you should get some later deadline wu's as you complete the work and as the system fills your cache.

This is theory only at this time which I think I have now confirmed as I dumped approx 250 wu's earlier today and I now have new work with a deadline of 12th June.

I wanted to do this now before the work dries up.

[Rob_B]

I wonder then if it might be a good idea to abort a number of the earlier deadline wu's now and then you should get some later deadline wu's as you complete the work and as the system fills your cache.

This is theory only at this time which I think I have now confirmed as I dumped approx 250 wu's earlier today and I now have new work with a deadline of 12th June.

I had done that earlier this week. But not with that many wu's. And it worked for me too.

But if it looks like your wu's will finish on time, you better not do that now. There should be very little work left. We are down to less than 2 days for this project.

If your wu's have a late end date but you don't have enough in cache you can try down clocking the cpu. That way the wu's you have will last longer. It is now all about runtime, not points or number of wu's

[OldChap]

Looking ahead to the HPF2 part of this challenge, the issue as I see it is that the last new wu's will be issued a day or two before we get the GFAM jobs finished....

....but I think I have a plan

Based on what we know at this time about days left: On Friday next I will put another drive in this machine and run a new instance of Boinc. The aim being to dl as many wu's as possible before the work dries up. Then, switch drives and as soon as GFAM target is reached I will run using that new drive with turbo off to stretch them out a bit if necessary although I suspect that the issue then will be the deadlines.

If we can get 7 days of work to run on 56 cores we will just about badge that one too

[OldChap]

Quote from Sekerob on WCG site, Re GFAM:

Overnight quad and octo received a boatload of tasks with the "PfLAMMER" marker in them, 115xxx batches, and then it stopped again when a _1 came in with the inflated TTC, which inflated the rest... no idea why this is seemingly happening only on repairs and quorum 2 assigned ZR tasks.

If this is the last series and remains serial, we've got about 1-1.5 days left coming from the feeder.

[jeanguy2]

Looking ahead to the HPF2 part of this challenge, the issue as I see it is that the last new wu's will be issued a day or two before we get the GFAM jobs finished....

....but I think I have a plan

Based on what we know at this time about days left: On Friday next I will put another drive in this machine and run a new instance of Boinc. The aim being to dl as many wu's as possible before the work dries up. Then, switch drives and as soon as GFAM target is reached I will run using that new drive with turbo off to stretch them out a bit if necessary although I suspect that the issue then will be the deadlines.

If we can get 7 days of work to run on 56 cores we will just about badge that one too

OC, above sounds like a good plan. 1.S can also start crunching HPF2 earlier than next Friday by using the help from the devices of Xtremers who will have spare crunchers available in the next 2-3 days. I recommend that he post his HPF2 core requirements so that folks now what he needs. New helpers coming to him can then switch over to him with a low cache (0.1) day and let him know the name of their switched devices. These devices will then download a minimal amount of GFAM WUs or nothing since his present default profile is set to GFAM. 1.S can then go to his WCG profile and change the profile for these newly added crunchers to the one that is set for HPF2 crunching. Once HPF2 WUs show up, helpers can then raise their cache to whatever 1.S needs.

It look now like 1.S and Yojimbo both need help with HPF2. Yojimbo I will be coming over to your account with 20 cores minimum NLT 5th june 2013. Please insure that the HPF2 profile you use is set as Default.

[Snow Crash]

musings of a badge hunting boinc point hore ...

Cache as much as you can, as soon as you can. If you do that now then as soon as you return work BOINC will request more, no real reason to dump some now and try to refill before the deadline.

As soon as new work stops being distributed lower your clocks immediately as you would rather pass the deadline with WUs left over than run out of WUs altogether.

After the new work stops flowing turn down your cache down only enough so that you are still requesting work each time you report completed WUs. As you get closer to your deadlines you should keep dropping your cache to make sure you are not running in High Priority as I'm pretty sure that excludes you from getting any resends.

Yes, turnaround is also part Resend equation so on the day you will run out of WUs, turn your cache to 2 days and open your profile to another WCG project. Once you have your 2 days full again, reset your profile back to only be the project you want.

I have a couple of emeralds instead of sapphires due to thinking I had as much time as the estimate charts said I did - learned my lesson and here is the approach I use now. Get all your badges now, as soon as you can on all the projects you want so that when a new project gets released you can attack that badge goal and attain it as quickly as possible. This way you're never tormenting your self when a project closing announcement is made and you ask yourself "why didn't I crunch it sooner"?

[yojimbo197]

[QUOTE=jeanguy2;5191153
It look now like 1.S and Yojimbo both need help with HPF2. Yojimbo I will be coming over to your account with 20 cores minimum NLT 5th june 2013. Please insure that the HPF2 profile you use is set as Default.[/QUOTE]

Yep, already done.