Yojimbo, can you post your present HPF2 accumulated runtime as well as the number of cores you now have available for HPF2?Quote:
Originally Posted by yojimbo197
Thanks
Printable View
Yojimbo, can you post your present HPF2 accumulated runtime as well as the number of cores you now have available for HPF2?Quote:
Originally Posted by yojimbo197
Thanks
Is everybody comfortable that they will get GFAM targets?
One.shot and yojimbo are both needing around 365 days on HPF2 I think. One.shot will need about 7 days of cache loaded before new work stops if we all stay with him.
Who is proposing to jump to yojimbo's aid?
I know there should be 9 days of work left but lets say 7 to be safe so he too will need around 56 cores total, all loaded for bear, by the 9th.
Anyone else looking for HPF2?
After I finish my HPF2 badge I'll have 10 cores to help out. Plus another possible 4 cores after Anubis hits his GFAM badge.
Based on comments in previous posts .... maybe it looks like this???
Making the assumption that new work dries up on the 9th:
Jeanguy= 20 cores to yojimbo on the 5th. This equates to around 4 days running + 8 days cache = 240 days
Bluestang= 10 cores to yojimbo on 6th = 3 days running + 8 days cache = 110 days
Maybe if the cache is set high on changeover those will be enough.... another 4 core would seal it though so if he runs that himself..... :D
If anybody can do a bit for One.shot to get his total required down a bit before we (current team) finish the GFAM and go to HPF2, things would be a bit easier
OC, I will switch a minimum of 20 cores to Yojimbo NLT 5 June 2013Quote:
Who is proposing to jump to yojimbo's aid?
I can give another 8 threads by Wednesday too. Maybe a few more.
The 2x 8 threads I have running for Mumak and 1.S will stay there till they reach Sapphire or we run out of work.
Update #4
Okay, currently I'm at: GO Fight Against Malaria 0:349:10:41:40
05/31/2013 0:044:22:21:28
06/01/2013 0:050:11:18:21
06/02/2013 0:028:07:10:03
I have 6 pages + 4 of PVs and I should reach Emerald by tonight's update. The runtime has went up a few hours from yesterday, so far.
For GFAM, I'm currently at 1:111
My 20 cores have 10 days of work buffered. I'll check the situation in detail next week. Maybe Rob and Yeroon could switch their hosts to others needing help few days sooner.
To reach my own target for GFAM I'm running some older hardware. And that shows. :eek:
Average KWh per day.
http://www.robbakker.com/images/verbruik062013.png
Rob, would you consider going to help One.shot as soon as you are free to do so?
I ask because I think we need to run around 7 days to finish up GFAM which is the 9th and that is possibly cutting it close.
I think that if he changes his profile to both GFAM and HPF2 as soon as GFAM dries up, and we all keep our cache settings Big we should be good to go because as we approach the time when HPF2 dries up also we should all have nearly full caches to work on and that should be enough
My request is a sort of insurance,
Human Proteome Folding - Phase 2 1,716,384 2,096 1:008:16:36:27Quote:
Originally Posted by jeanguy2
I have 6 cores dedicated to HPF2 right now. 8 are donated to Anubis and One shot.
Excellent news Yojimbo. Your 6 cores + my 20 cores + Bluestang 10 will give you 36 cores crunching which meets your 34 cores requirement for getting your HPF2 sapphire as stated by OC above. Once 1.S minimum cores requirement to get HPF2 sapphire are met, it would be nice to add another 8 helping cores to your crunching stable to seal the deal for you if no other Xtremer needs help..Quote:
Originally Posted by yojimbo197
Hang in there Yo... the cavalry is a coming...:D
Mumak: Looking at your numbers I thought to play it a little safer I would add a "QUAD"
It is downloading work as I type this. initial dl = 190 tasks @ estimated 3h 53m each so maybe 30 days
We got 35.5 days today!
Human Proteome Folding - Phase 2
1:252:16:16:58
That leaves 112.5 days - 45 days min PVs = 67.5 days
35 cores running (Snow Crash 11 + fallwind 12 + my 12) with 938 WU's In Progress.
Looks like everyone can pull off my account on Tuesday (June 4th) night. I'll know more definitely tomorrow night.
I have added an other 8 threads to One.shot's account.Quote:
Originally Posted by OldChap
Im cool now, you can remove all helpers from my account! Once the WU queue on my own crunchers will clear will have the badge.
GO Fight Against Malaria 2,569,581 3,256 1:292:19:26:08
Much appriciated, guys!
Update #5
Okay, currently I'm at: GO Fight Against Malaria 1:010:22:50:40
05/31/2013 0:044:22:21:28
06/01/2013 0:050:11:18:21
06/02/2013 0:054:19:19:03
I have 104 PV WUs right now, so (104 tasks * 4 hours)/24=17.33 days. 355-17=338 more days with 50 days/day of compute time = just under 7 more days needed on GFAM for Sapphire. Make sure your caches are set to at least 7 days and we should be able to pull it off.
HPF2: 1:005:11:40:30
EDIT!!!: I have 2903 GFAM tasks in my queue. I only need (338*6 tasks per day)=2028 WUs to reach Sapphire. I'm going to set all devices to start accepting HPF2 WUs and I'll let everyone know when to abort all "in progress" GFAM WUs. I hope my math is correct. It's 1AM after a long day!
Thanks, OC :) I see your machine now on my acct.
Current GFAM status: 1:123
I've got a lot of WUs in my cache now, but many of them ending 6/6. That was my fault, since I raised the caches in one turn, I should have done that in steps day by day.. Will probably try to abort them to get new ones with a later deadline. A bit risky so near the end...
I think in 2-3 days I should know whether I need help, but so for I think I should make it :)
EDIT: Aborting works. In turn I got whole new WUs ending 6/13 :)
EDIT2: Got new tasks, but not as many as I have aborted - aborted ~500 WUs, got back ~350. The amount of received tasks raises slowly after each update, but I don't know if I'll get as many as I had, so be careful with such operation near the end...
EDIT3: I have now ~950 WUs cached on my own machines (which return ~100 WUs/day). That should be enough for 200 days of runtime. Adding 5 pages of PVs, I think it should be soon safe for me :)
Alright, nice! :up:Quote:
Originally Posted by anubis
I just aborted 175 WU's from my Q6600 rig and I'll let the 4 WU's currently running finish (about 1.5-2hrs left) an then I'll switch my 4 cores to someone else.
Everyone OK with GFAM?
Mumak, are you sure your good?
I'm currently @ 1:139. I guess I should be OK, but maybe give it 1-2 more days of help, just to be sure. Then I think all my helpers can switch to others needing help. Or, maybe if help is urgently needed on another place, let's say 1 or 2 of them (Rob, Yeroon, OC) could switch to others now and leave 1-2 of them for 2 more days...Quote:
Originally Posted by bluestang
Ok, so once Mumak is in the clear, who else is still cutting it close so that I can switch over.
1.S needs help for HPF2...:up:Quote:
Originally Posted by Yeroon
I threw my 4 core rig onto Mumak's account for now. I can always switch to someone else, but didn't want it sitting idle.
One.shot. May I suggest you switch on HPF2 only now so that we start to build caches for that one.
My thinking goes like this:
When calculating for when to abort GFAM maybe ignore your PV's (or half of them) and use them as a safety net. Also be aware that recently I have been running wu's that took almost 5 hours to complete, today the work is taking less time it seems so maybe all wu's are not created equal???
As I see it the only thing critical to you and your team is to have enough HPF2 in cache when you get your GFAM as I believe it is possible that HPF2 may finish sending new work before you get your GFAM badge. Whilst I hope it does not and we will have enough work I have made contingency plans for this I think.... but only to the tune of maybe 80-100 days.
The other thing to remember is that once you get your GFAM badge you only need a 7 day cache on all the machines running
You don't need to decide on this just yet ... I think we need to do this latest on the 9th, even though the last of the new HPF2 work is still projected to come out on the 12th, then do it based on runtime for each rig individually. On the 9th there should only be a need for 1 day cache :D The timing is our safety net for an early finish.
Rambling a bit but I hope you get the idea
OC, i think 1.S has already turned on HPF2.. see post 117 above:Quote:
One.shot. May I suggest you switch on HPF2 only now so that we start to build caches for that one
Quote:
EDIT!!!: I have 2903 GFAM tasks in my queue. I only need (338*6 tasks per day)=2028 WUs to reach Sapphire. I'm going to set all devices to start accepting HPF2 WUs and I'll let everyone know when to abort all "in progress" GFAM WUs. I hope my math is correct. It's 1AM after a long day!