Hungry hosts could be a problem to Einstein?
19 Nov 2012 23:04:12 UTC
Topic 196635
(moderation:
I switch 8 hungry host (all with NVidia GPUs) from SETI to Einstein yesterday, still working with 0 (zero) resource share for now.
To avoid any mistake, if i put them to crunch Einstein at full speed that could cause any problem to the project?
Please forgive if that is an "stupid" question but i don´t want to generate any problem with my action.
Hungry hosts could be a problem to Einstein?
)
Welcome to the project Juan,
Nice fleet of cuda crunchers
You will cause no problems and in fact do thousands of tasks and get millions of credits.
I stopped running Seti years ago but here the main thing I do is check the temp. on my video cards once in a while since they run 24/7
You can test yours and see if any cards run better/faster with X1,X2,X3,or more tasks at the same time.
Your RAC will take a week or more to get up to what you are actually getting but your Credits will take off real soon.
Have fun crunching here Juan
edit: I just checked and see you have hundreds of Seti's in progress so trying to do both at the same time may make it hard to finish on time so I would have all of your set to "no new tasks" so you can take care of that.
RE: I switch 8 hungry host
)
Welcome to E@H Juan! :-)
No problem, let it rip ... we like that :-)
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Thanks for the warm
)
Thanks for the warm welcome.
Most of the SETI work still on my hosts are CPU WU, all the GPU works was already done, just can´t report due the SETI well know server/bandwith problem. Eventualy they will clear.
I will start slowly to build a E@H cache on my fleet, one at a time to avoid any trouble, will take days but i have all the time of the world. And BTW stay at the front of a host monitor with a cold beer is a good thing to do.
I start to make some tests regard the 1x, 2x, etc WU at the same time per GPU, and see the performance. Because i have some very diferent types of GPUs (from 560 to 690) i need to verify the actual capacity of each one on separate.
About temperature, thanks for the warming but all my cards have regular manteinance (cooler/termal paste,/cleaning, etc.) they have a full set of collers/fans and i run EVGA precision program to keep them cool as possible in my hot tropical country. And BTW as i could see the E@H apps generate a lot of less heat than SETI optimized Beta App´s i already normal use. They are alredy running 24/7 for months without any trouble.
Thanks again, hope i could do a little contribution to the project in order reaches it´s goals.
And please always forgive me very bad english, as you could note that's not my native language, but be sure i always try to do my best.
RE: Thanks for the warm
)
Occasionally E@H is a victim of it's own success, which is when we add more server resources. :-)
Yes. That would work. :-)
Not per se ( that comes to my mind at least ), but if not fire up a thread yourself as I'm sure you'll get a discussion.
Your English is fine, and way better than my Portuguese or Spanish. :-)
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Juan wrote:I start to make
)
Good to make tests--even share your results. I do think in the past more than one of us has reported very, very poor performance when a SETI WU and an Einstein WU were running on the same GPU at the same time.
One other thing that may surprise you is that the Einstein GPU work is more demanding of CPU (not clear how much this really means CPU I/O bandwidth vs. actual computing) than the SETI case, I think. So much so that it is more than usually helpful to cut down the number of allowed CPU jobs here when one has capable GPU resources, as you surely have.
I wrote a number of posts in a thread I started regarding bringing up a GTX 460 hosts here some months back. You probably don't use that particular model of GPU, and the Einstein application which uses GPUs was updated since my tests, with changed characteristics. Still you might find something of interest there.
With my rather modest fleet of just two GTX 460 hosts, I am over half the monthly bandwidth limit from my cable Internet Provider--Comcast. If you are sensitive to aggregate data transfer from a price, throttling, or limiting cap perspective, this might be something you'll want to watch carefully.
Welcome to Einstein.
RE: I do think in the past
)
With the old Einstein apps my experience running SETI and Einstein sharing the same GPU was that both apps were running faster than when both halves of the GPU were used by Einstein or by SETI... but due to the FIFO nature of the BOINC cache it was rather difficult to achieve that thing of having one of each kind running all the time.
The issue running Einstein and SETI togheter is that as SETI has always issues giving work and Einstein is always ready to give more work then... well, you wont be crunching too much for SETI as your cache will be filled quickly by Einstein... On the long term BOINC will reach a threshold about debts and will try to give SETI more priority... but you know, without babysitting SETI wont go anywere...
RE: but you know, without
)
As everyone must know i was one of the biggest SETI RAC contributor (at least one of the top 10 the last time i see) not leaving SETI or changing it for Einstein, simply i get tired to do babysitting and get no answer (a real and honest one, i hate political answers like the last one given) from the staff.
Because that, i think is time to sail on others seas and why not try to help people that make so warm welcome?
I realy belive, running SETI and Einstein at the same time, with so many diferent hosts and configurations will be a challenge for me (don´t forget my hosts makes a lot of other type of work on my company, most of them in a 24/7 cycle), and of course will need a lot of help to do that, but i´m sure i count with a lot of unknows friends ready to help me, as allways happens in the SETI forums and expect nothing less here.
This could be not the right forum but somethig must be say, the SETI project could be a dissaster in planning and management of the resources, but the volunteers and the comunity are something amazing. I realy belive thats is why the project is still running until today with all its bad days, bugs and lack of money.
Thanks for the info about CPU usage, i all ready note that and leave some cores avaiable to do the job, specialy in the 2x590/2x690 hosts. On the other hand, i belive don´t will have a problem with bandwidth or storage space to run Einstein, diferent from the SETI lab, we never use more than 50% of the avaiable resources at any time, actualy we use 3 diferent ISP (2 cables and 1 ADSL2 each with no less than 10MBps guaranted bandwitdh and unlimited traffic quota) at the same time (with redundancy and load balancing of course) and have a lot of terabytes of storage avaiable at any time.
And the kitties shall be
)
And the kitties shall be sending more rigs here to join Juan soon.
2 right now, 7 more when they run out of Seti.
Hungry kitties!!!
Meow meow meow!
RE: And the kitties shall
)
Follow my advice: Allways follow the kittes path... I´m a kitties fan!
Could some one give a clue on this?
- There are some kind of optimized app (Win XP or 7 - 32/64)f or Einstein like the one avaiable for SETI?
- In any forum, is avaiable a thread just for chance some waste of time reagar some totaly out of focus thing? Like humor, speak about beverages, trips, etc. something realy just to pass the time and make some friends or enemys who knows?.
- Any place we could see some test results that could help to optimizes the number of WU task per GPU? Just don´t want to loose time trying to reinvent the weel.
Thanks in advance.
RE: RE: And the kitties
)
LOL...I'm rather fond of them too!