Printable View
Uh huh. Just wanted him to do it now rather than later after he said he is "going to". EDIT: I should have checked. I already have my first batch of 80 wu's. Estimated runtime is about half an hour longer than GFAM
So everybody, am I being too pessimistic on this or just playing safe?
What do YOU all think will be the last day for new work from the HPF2 project?
I'm already getting some HPF2 wu's on one of the rigs I have running for 1.s.
I think the dashboard from sekerob is one day old, as it is still showing 9 days left. So 8 days and with GFAM finished and not many other projects left, it will be closer to 7 days.
According to latest update posted about 1 hour ago, there's only 3-4 hours of GFAM steady work remaining.
Mumak, got in just in time for you then.
135 WU's in cache and "no new tasks available."
Last day for HPF2 = 10th June 2013Quote:
What do YOU all think will be the last day for new work from the HPF2 project?
OC, you're not pessimistic... You're just darn careful & it's a good way to insure that the final goal is reached...:yepp:
So caches of my own machines are holding now ~ 1150 GFAM WUs.
How is that looking for you Mumak?
Quad is sitting on about 34 days total, Bluestang should have some too, what about your other rigs and those of any other helpers?
Sitting on ~20 days here for Mumak.Quote:
Originally Posted by OldChap
Got 172 wu in cache for Mumak, * 4.5h avg completion time, gives me approx 10 days buffer per core (3), so I have right around 30 days here.
EDIT: Is it only based off runtime, or also WU completion speed?
Also, I have buffer for 10 days but latest deadline is the 11th, so I only have 8 days x 3.
Currently I need 225 days.
My 20 cores are loaded with ~1150 WUs, returning ~100 WUs/day. 70 WUs are PV.
Helper machines have ~ 1000 WUs buffered in total.
So my rigs should have work for ~11 days, however deadline is 10 days, so I assume some of them won't be finished on time. Nevertheless, from now on I can make 200 days of runtime (assuming the WUs are well spread to keep all rigs busy for 10 days, but this is hard to tell so early).
So it seems there's still a little help needed.. I'd say ~1-2 days max to be safe.
436 wu's for Mumak here. That is about 70-75 days.
OK, so let's do the math :cool:
I don't expect I'll catch many resends, so I cannot make more than ~200 days (20 cores * 10 days deadline). Add ~10 days of PV now, that gives ~15 days of additional runtime needed.
So 1 helper should be enough...
Mumak, you should have forty something cores on your account now. I think with tomorrow morning's WCG update you will have a firm idea of where you are at. I'm guessing you'll clear in 5 more days.
Update #6
Okay, currently I'm at: GO Fight Against Malaria 1:038:19:52:44
05/31/2013 0:044:22:21:28
06/01/2013 0:050:11:18:21
06/02/2013 0:054:19:19:03
06/03/2013 0:027:21:02:04
I have 103 PV WUs right now, so (103 tasks * ~4 hours)/24=~17.16 days. 327-17=310 more days with 50 days/day of compute time = just under 6 more days needed on GFAM for Sapphire. Make sure your caches are set to at least 6 days. 310 * 6 WU per day = 1860 WUs (4hr/WU) or 310 * ~6.8 WU per day = 2125 WUs (3.5hr/WU) needed to reach goal. Of course I'm making the assumption of 4 hr/WU to 3.5 hr/WU.
I currently have 2689 GFAM WUs in queue among all devices. I've noticed some of you are completing them in under 4 hours and most of mine are a bit over 4 hours. We need between 1860 to 2125 WUs and we have a surplus between 829 and 564 WUs.
In a worse case scenario with HPF2, we can downclock our rigs for maximum runtime with limited WUs. I'm at HPF2 1:005:11:40:30 and I can cut my clocks in half to double runtime if needed. There are 150 HPF2 WUs in queue among all rigs.
How many days are left for HPF2 WUs? I think it's safe if I abort 200 tasks and start getting more HFP2 WUs.
EDIT! Upon closer examination, Oysterhead rig is pumping out WUs in under 3 hours. The rest are 4 hours and more. The 1860 number may still be a bit conservative on the needed number of WUs. Either way, we have more than enough and I'll continue to monitor the situation.
We did 39 days today!
Human Proteome Folding - Phase 2
1:292:07:48:27
That leaves 73 days - 35 days PVs = 38 days left
35 cores running (Snow Crash 11 + fallwind 12 + my 12) with 792 WU's In Progress.
Well Anubis said he didn't need anymore help, so I pulled my Q9550 rig off and put it on HPF2. So now I am at 10 cores.
Tonights HPF2 run time total: 1:018:02:56:49
I'll be holding with Anubis until I run this machine dry or he rolls over the mark.
I have 36 cores available for those needing assistance with HPF2. PM me with your weak key and how much run time you need and we'll work something out.
My GFAM status is 1:153.
GFAM ran out faster than I expected. 1:301 and not getting more WU's :(
Some guys here might help you if you post your weak key.Quote:
Originally Posted by Red Maw
Nobody can help now that they cannot get new work. Re-sends are only given to proven machines if I remember right.
The best to hope for is to reduce the clocks on the machine and hope that the wu's get done within the deadline.
Red Maw: how many cores are you running and how long does an average wu take? How many wu's do you have in cache?
Not even if a helper having a large cache of GFAM WUs would detach from my account and attach to his ?
Unless I'm mistaken, work units sent to one account have to be returned to that same account or it breaks the audit trail. I think changing the key with units in cache will force you to reset your client and reload new work.
My latest GFAM status: 1:169