A variant on this theme, mainly related to older computers is the upgrades to fast GPUs on these systems. Refurbished quads is good, and fast GPUs is another way to keep older computers with a competitive output at a non-wallet killing price.
If I might add ...
I have always bought computers behind the curve. The hottest new machine (not meaning gaming machines) become fairly priced about 9-10 months after introduction. That is when I used to buy. I never saw the point of paying a premium for a box that would leave me hungry again in 9 months.
Since discovering refurbed machines I get them maybe six months behind the curve. I get the performance a few months earlier.
In the long term sense it still takes about three years before there are enough improvements worth buying.
As to GPUs. I did a quick try on my newest machine and had a problem so I took out the old nVidia card. The motherboard has CUDA capable nVidia chips. The test of the computer that is run on startup does not recognize them. Now that it is up and running I will have to try the card again and see if I can get it recognized.
What I have not come across is a rule of thumb to estimate the performance improvement to see how much effort it is worth to get it working. Any suggestions? Seems to me if the CUDA stuff is up to its reputation even the cheapest card is worth the investment.
An easy thing is to look at the stream processors of the card, the more there are the faster the card can crunch. HOWEVER memory speed, amount of memory etc WILL all come into play too. Right now ATI cards are the fastest but Nvidia is working hard to overcome that and of course ATI is trying to maintain its lead. Also some projects can only use one kind of card or the other, some can use both and some can only certains models of a brand of card. MW can only use double precision cards, that leaves alot of cards out of the mix. Collatz can use both brands and alot of different models too, DNETC is similar to Collatz. It seems as though Einstein can only use Nvidia cards. The rest I am so unsure of I will not speculate here. Of course buying the newest latest and greatest and the most expensive card will give you the most bang, but for the best buck I always try to buy gpu's like I do cpu's, last years model as they come down in price and are replaced by newer models.
SETI@home is using both Nvidia and ATI GPUs, the latter on the Astropulse project. You can learn a lot of things on graphics cards. both good and bad, on the SETI forums. Since SETI is the leader in terms of users number of all BOINC projects, what happens in SETI will happen also elsewhere.
Tullio
SETI@home is using both Nvidia and ATI GPUs, the latter on the Astropulse project. You can learn a lot of things on graphics cards. both good and bad, on the SETI forums. Since SETI is the leader in terms of users number of all BOINC projects, what happens in SETI will happen also elsewhere.
Tullio
Hi Tullio,
I have my both pc's attached to SETI. Yes, they have also ATI-based wu's. I got 5 of them in the last 6 months. In fact they do not have ATI-support.
I cannot agree to your words 'what happens in SETI will happen also elsewhere'.
My main project is Milkyway. It is not run by a private organisation but from the RPI (Rensselaer Polytechnic Institute, Troy, New York, USA).
Travis Desell, Developer, Forum moderator (and in close contact to volunteers!!!!) , graduated this year in computer science. Here you can find two publications: http://wcl.cs.rpi.edu/papers/escience2009.pdf http://wcl.cs.rpi.edu/papers/ppam2009.pdf
SETI@home is not run by a private institution but by the Space Science Laboratory of the University of California at Berkeley. The SETI Institute is a private institution and has nothing to do with SETI@home. It has recently launched SetiQuest (see www.setiquest.org) that has roughly the same aims of SETI@home but different ideas on how to reach them. SETI@home gets its data from Arecibo, as Einstein@home, SetiQuest from the Allen Telescope Array which was financed by Paul Allen and whose operating expenses are actually covered by the US Air Force. Only one third of the ATA operations is devoted to SETI, another third to radioastronomy managed by the Radioastronomy Laboratory of Berkeley U., another third to the US Air Force. Why SETI@home cannot get any data from the ATA is a question I cannot answer.
Tullio
SETI@home is not run by a private institution but by the Space Science Laboratory of the University of California at Berkeley. The SETI Institute is a private institution and has nothing to do with SETI@home.
Sorry, my mistake. Seti is not my primary project and I'm not well-informed.
But the point is: there is always a solution for an upcoming problem.
A couple of days ago orbit@home came to life again. It took two days to 'kill' the servers; the echo to the reengagement into the cruching-scene was enormous. Pretty the same situation as currently at seti.
What they did: We have introduced important changes to the validator, mainly to improve its performance. It now runs in about 1 second, compared to the 35 seconds needed by the previous version, so that should take care of the pending WUs and keep the server load down.
About half a year ago at MW: a guy postet: Just bought 2 additional HD5970. Sorry Travis, I have to punch your servers down!
What Travis did: he distributed longer workunits and everything was fine again.
A simple question: Is it really necessary, that a seti cuda wu finishes within 3 to 15 min ? Make them 10 times longer and you have ten times less traffic and validation work.
Is it really necessary to have more than 200.000 wu ready to distribute? MW produces these wu's on demand, they don't have to manage that amount of stock wu's. Randomly checking the serverstatus there you will see less than 2000 wu's ready to send most of the time.
Not long ago Einstein had a similar problem due to too many new crunchers. They invested in a new server and outsourced some work to another PC.
All this has formed my mind: there is always a way, if there is somone who wants to solve a problem.
Seti has 249,513 active hosts.
Milkyway has 34,464 active hosts.
See the difference there already? Now just imagine that all 249,513 hosts try to contact Seti at the same time, telling the project they want to upload, download and report work. That's a pretty sturdy serverbase one needs then, to be able to cope with those amounts of computers hitting your servers 24 hours a day. Seti manages this, on a shoestring budget, and has managed to do so for the last 11 years.
Einstein, by the way, has 190,365 active hosts. Just for comparisons.
More differences between the projects, some of which you neglect to mention. At Seti the new credit system decides how many tasks you're allowed to download by the amount of work your computer has done reliably. Mean is 100, per CPU per 24 hours.
At Einstein it's 32 tasks per CPU per 24 hours.
At Milkyway it's 5,000 or 10,000 tasks per CPU per day. Now what's the use in numbers like that when - as you say - they only have 2,000 or less available at any time? Especially on a 3 day deadline. Madness (personal opinion).
In SETI@home CUDA equipped PCs crunch WUs so fast that they overload the servers. Result: server shutdown 3 days a week and I cannot upload my CPU results. I hope this does not happen in Einstein.
Tullio
With your and many other posts on CUDA I questioned adding the CUDA issue.
I come from a different background. I spent my time in hell getting a BS in Physics but my career was in R&D management where "bang for the buck" was always a factor as bucks were finite. Funny thing, they still are today and in real life.
I looked at buying 9 months behind the curve as more bang for the buck because of the front end premium price charged to early adopters.
I am very comfortable estimating the optimum price of of CPU/$ which is $400 when purchased as a stand-alone. Below that price there is a confusion of features that marketing people love to move the product but it is not possible to price from any advertisement. If you pay less than $400 you will find some compromise in the components that make it profitable at that price. But if you are buying 1.5 years behind the curve you can find components that once were premium but are being sold cheap to make room for new inventory -- like end of year sales for new cars.
But now I need to estimate nVidia card prices. The one I am thinking about using is about three years old. (I was going to get a cheap off-lease computer to drive my TV and make up for the low performance with the card.) Obviously it is out of date and was only $80 or so at the time.
Several model numbers have been given without prices. A couple of performance improvements have been given and met with skepticism. If the best improvement were true I could drag out the emergency spare from the closet and have a screamer.
But now I need real numbers on the price of the nVidia card v 4 core CPUs.
People can do what they want with their money. Here in Italy they buy SUVs just to go the supermarket instead of crossing the Sahara.So do people at SETI@home buying Fermi cards. I wonder if they ever read the "Theory of the leisure class" by Thorstein Veblen. It was one of the first books they made me read at Thomas Jefferson School in St.Louis back in the Fifties. Conspicuous Consumption was the term used by Veblen.
Tullio
RE: RE: A variant on this
)
An easy thing is to look at the stream processors of the card, the more there are the faster the card can crunch. HOWEVER memory speed, amount of memory etc WILL all come into play too. Right now ATI cards are the fastest but Nvidia is working hard to overcome that and of course ATI is trying to maintain its lead. Also some projects can only use one kind of card or the other, some can use both and some can only certains models of a brand of card. MW can only use double precision cards, that leaves alot of cards out of the mix. Collatz can use both brands and alot of different models too, DNETC is similar to Collatz. It seems as though Einstein can only use Nvidia cards. The rest I am so unsure of I will not speculate here. Of course buying the newest latest and greatest and the most expensive card will give you the most bang, but for the best buck I always try to buy gpu's like I do cpu's, last years model as they come down in price and are replaced by newer models.
SETI@home is using both
)
SETI@home is using both Nvidia and ATI GPUs, the latter on the Astropulse project. You can learn a lot of things on graphics cards. both good and bad, on the SETI forums. Since SETI is the leader in terms of users number of all BOINC projects, what happens in SETI will happen also elsewhere.
Tullio
RE: SETI@home is using both
)
Hi Tullio,
I have my both pc's attached to SETI. Yes, they have also ATI-based wu's. I got 5 of them in the last 6 months. In fact they do not have ATI-support.
I cannot agree to your words 'what happens in SETI will happen also elsewhere'.
My main project is Milkyway. It is not run by a private organisation but from the RPI (Rensselaer Polytechnic Institute, Troy, New York, USA).
Travis Desell, Developer, Forum moderator (and in close contact to volunteers!!!!) , graduated this year in computer science. Here you can find two publications:
http://wcl.cs.rpi.edu/papers/escience2009.pdf
http://wcl.cs.rpi.edu/papers/ppam2009.pdf
Look into this thread:
http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=1256#33386
So many developers, forum moderators, project scientists in close contact to the volunteers. They started 2006, now MW is the #2 project in BOINC.
Regards,
Alexander
SETI@home is not run by a
)
SETI@home is not run by a private institution but by the Space Science Laboratory of the University of California at Berkeley. The SETI Institute is a private institution and has nothing to do with SETI@home. It has recently launched SetiQuest (see www.setiquest.org) that has roughly the same aims of SETI@home but different ideas on how to reach them. SETI@home gets its data from Arecibo, as Einstein@home, SetiQuest from the Allen Telescope Array which was financed by Paul Allen and whose operating expenses are actually covered by the US Air Force. Only one third of the ATA operations is devoted to SETI, another third to radioastronomy managed by the Radioastronomy Laboratory of Berkeley U., another third to the US Air Force. Why SETI@home cannot get any data from the ATA is a question I cannot answer.
Tullio
RE: SETI@home is not run by
)
Sorry, my mistake. Seti is not my primary project and I'm not well-informed.
But the point is: there is always a solution for an upcoming problem.
A couple of days ago orbit@home came to life again. It took two days to 'kill' the servers; the echo to the reengagement into the cruching-scene was enormous. Pretty the same situation as currently at seti.
What they did:
We have introduced important changes to the validator, mainly to improve its performance. It now runs in about 1 second, compared to the 35 seconds needed by the previous version, so that should take care of the pending WUs and keep the server load down.
About half a year ago at MW: a guy postet: Just bought 2 additional HD5970. Sorry Travis, I have to punch your servers down!
What Travis did: he distributed longer workunits and everything was fine again.
A simple question: Is it really necessary, that a seti cuda wu finishes within 3 to 15 min ? Make them 10 times longer and you have ten times less traffic and validation work.
Is it really necessary to have more than 200.000 wu ready to distribute? MW produces these wu's on demand, they don't have to manage that amount of stock wu's. Randomly checking the serverstatus there you will see less than 2000 wu's ready to send most of the time.
Not long ago Einstein had a similar problem due to too many new crunchers. They invested in a new server and outsourced some work to another PC.
All this has formed my mind: there is always a way, if there is somone who wants to solve a problem.
Kind regards,
Alexander
RE: Not long ago Einstein
)
Err ... do tell?
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
Seti has 249,513 active
)
Seti has 249,513 active hosts.
Milkyway has 34,464 active hosts.
See the difference there already? Now just imagine that all 249,513 hosts try to contact Seti at the same time, telling the project they want to upload, download and report work. That's a pretty sturdy serverbase one needs then, to be able to cope with those amounts of computers hitting your servers 24 hours a day. Seti manages this, on a shoestring budget, and has managed to do so for the last 11 years.
Einstein, by the way, has 190,365 active hosts. Just for comparisons.
More differences between the projects, some of which you neglect to mention. At Seti the new credit system decides how many tasks you're allowed to download by the amount of work your computer has done reliably. Mean is 100, per CPU per 24 hours.
At Einstein it's 32 tasks per CPU per 24 hours.
At Milkyway it's 5,000 or 10,000 tasks per CPU per day. Now what's the use in numbers like that when - as you say - they only have 2,000 or less available at any time? Especially on a 3 day deadline. Madness (personal opinion).
The SETI servers have been
)
The SETI servers have been shut down again due to an air conditioning failure. Thermodynamics still rules.
Tullio
RE: In SETI@home CUDA
)
With your and many other posts on CUDA I questioned adding the CUDA issue.
I come from a different background. I spent my time in hell getting a BS in Physics but my career was in R&D management where "bang for the buck" was always a factor as bucks were finite. Funny thing, they still are today and in real life.
I looked at buying 9 months behind the curve as more bang for the buck because of the front end premium price charged to early adopters.
I am very comfortable estimating the optimum price of of CPU/$ which is $400 when purchased as a stand-alone. Below that price there is a confusion of features that marketing people love to move the product but it is not possible to price from any advertisement. If you pay less than $400 you will find some compromise in the components that make it profitable at that price. But if you are buying 1.5 years behind the curve you can find components that once were premium but are being sold cheap to make room for new inventory -- like end of year sales for new cars.
But now I need to estimate nVidia card prices. The one I am thinking about using is about three years old. (I was going to get a cheap off-lease computer to drive my TV and make up for the low performance with the card.) Obviously it is out of date and was only $80 or so at the time.
Several model numbers have been given without prices. A couple of performance improvements have been given and met with skepticism. If the best improvement were true I could drag out the emergency spare from the closet and have a screamer.
But now I need real numbers on the price of the nVidia card v 4 core CPUs.
I knew I should have kept my mouth shut.
People can do what they want
)
People can do what they want with their money. Here in Italy they buy SUVs just to go the supermarket instead of crossing the Sahara.So do people at SETI@home buying Fermi cards. I wonder if they ever read the "Theory of the leisure class" by Thorstein Veblen. It was one of the first books they made me read at Thomas Jefferson School in St.Louis back in the Fifties. Conspicuous Consumption was the term used by Veblen.
Tullio