The obvious thing is to have credit equal to so many flops, so you can see how much work your computer has done.
Except raw flops is just a measurement... It doesn't take into effect efficiencies different architectures might have, or even more so the fact that some projects favour ATI/AMD over nVidia (or vice versa), even if the hardware is comparable in other areas (be it benchmarks, gaming performance, or rendering in general). Some, like say MilkyWay with it's requirement for dual precision, also perform better on OLDER hardware due to there being more of a focus on such in the hardware (with Nvidia, Kepler will outperform Maxwell if dual precision is required).
This is why it has been, and will remain for quite some time, a never ending debate like another member linked to. Such is yet another reason why so many will say "choose your projects based upon the science you wish to support" instead of silly things like "I want the project that gives me the highest points in a day". Making a million credits in some projects can be done in a day if you have the right hardware (Collatz and PrimeGrid as an example) while others can take a while (Einstein@Home being one, especially if your focus is on cpu tasks), and some making 20k points in a day is something that not even everyone in the top 50 for that project are doing (Rosetta was used for this example), and even at that kind of production you're talking 50 days to reach a million points.
EDIT: Whoops, my bad, Rosetta@Home the top 55 are making 20k RAC..... Out of how many thousands of active volunteers? (end edit)
In the end, the biggest goal for any distributed computing project is progress towards their goal, be it finding the largest prime number in the world, searching for a cure for (insert disease here), wanting to know more about our universe in general, trying to solve mathematical theories and/or problems, or even just searching for extra terrestrial life through radio waves.
And then there is a project like Bitcoin Utopia where someone can earn 12 million credits in about 65 seconds as long as you have the right hardware. Zero flops involved just asic boxes crunching to help keep some of the stats sites and Boinc Projects running. When asked even Dr David Anderson, of Seti, commented that even that amount of credits was TOO LOW compared to the amount of work being done!! It was hoped that asic's would be used at other projects too but so far that has not happened, probably due to the lack of support from the asic hardware folks and the wide diversity of asic crunchers being made. Trying to program Boinc to use them is NOT easy and resources are very tight at most projects, but Projects like Seti's Astropulse Team have said it would be a huge benefit if they could use them. I have no idea about any other projects, but others have speculated there are several who could benefit from them.
Can you just imagine the brewhaha being bantered about if those types of machines were brought on line for the general Boinc projects? People with cpu's AND gpu's wouldn't stand a chance comparing credits then!!
For me credits are only worth comparing within a given Project, comparing project a to project b seems okay until you dig down and realize that each project has it's own programmers and developers and either optimizes or doesn't optimize it's apps all with almost zero outside help whatsoever!! That means that even IF credits were standardized across the Projects people crunching project a would STILL get more credits than people crunching project b and then how many people would project a be able to retain?!! Good projects would go away and those with the resources to have better programmers and developers would stay. Keeping the project credits non aligned means we crunch for our own reasons, not for credits. BTW 100 Billion credits and 10 bucks can get you a large cup of coffee at Starbucks, but without the 10 bucks you aren't getting ANY coffee!!
Seti and the Einstein Project are 2 different Universes to me and I wouldn't think about comparing them in any way.
But I stopped running any Seti tasks back in 2005 when this project started.
I think the probability of me seeing any alien creatures flying around or sending me any signal would be as good when I look up in the sky at night here.
So all I do is work here and for CERN........but that is just me.
It was hoped that asic's would be used at other projects too but so far that has not happened, probably due to the lack of support from the asic hardware folks and the wide diversity of asic crunchers being made.
That's the whole point of "application specific integrated circuits" - they are hard-wired to perform one task and nothing but that task well. And costs a lot to design and manufacture them, so people generally only build them for large to very large markets. And that's why we're using general processing units for most task.
If someone built an "FFT accelerator" ASIC, that might be able to help several projects including Einstein and SETI. However, GPUs are already pretty suitable for that task, hence both projects have a GPU app.
Before anyone creates an ASIC for something as specific as one application of a BOINC project, they should rather implement GPU acceleration with HSA and FPGA acceleration.
Peter Hucker wrote:
Most ASICs I used for bitcoin burnt themselves out (and yes they were well cooled). They're just not as robustly made as graphics cards
Designing and fabricating chips properly is not rocket science.. I'd say it's way more complicated than this. There are good reasons companies usually take a long time to "get it right". But the Bitcoin gold rush didn't leave them time. they "testing" was probably running the ASICs internally for the first few months, when they were the most profitable, before selling them.
Designing and fabricating chips properly is not rocket science.. I'd say it's way more complicated than this. There are good reasons companies usually take a long time to "get it right". But the Bitcoin gold rush didn't leave them time. they "testing" was probably running the ASICs internally for the first few months, when they were the most profitable, before selling them.
It wasn't the chips that blew, more the power circuitry on the board.
If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.
Designing and fabricating chips properly is not rocket science.. I'd say it's way more complicated than this. There are good reasons companies usually take a long time to "get it right". But the Bitcoin gold rush didn't leave them time. they "testing" was probably running the ASICs internally for the first few months, when they were the most profitable, before selling them.
It wasn't the chips that blew, more the power circuitry on the board.
Let me guess you were running the U3's? They STILL quit after a while, sometimes a few years, but they still run themselves to the point of not working anymore. The Monarchs by Butterfly Labs and the numbered series by Avalon seem to last much longer, though they are not nearly as cheap as the U3's are or were.
The whole point of this though was about credits and to bring it back to that...how could the credits from an asic be counted since it doesn't do flops in it's processing? You would have to come up with something else that would then be just as contentious as flops are now in finding an agreed upon standard. To me credits are best when viewed within a single project, trying to compare them between project a and b usually ends up like comparing apples and oranges and you end up with a fruit salad.
Bitcoin Utopia doesn't use cpu's or gpu's anymore, the difficulty with Bitcoins has made them very inefficient, so trying to compare BU with Einstein or Seti that does not use asic miners but does use cpu's and gpu's is unfair at best and a nightmare at the worst. BUT since BU IS a Boinc Project you can't just say 'we will just ignore that Project because it doesn't fit our thinking', so let's just ignore ALL Projects that don't do things 'the Seti way' and voila we are where we are now!!! Each project does it's own thing with credits and only the Stats Sites are the ones showing the combined stats. When I click on my account here at Einstein it does NOT show the total Boinc combined stats for me, it lists most of the Boinc Projects I have crunched for and each one with their own individual credits.
You completely misunderstand the discussion Michael. Credits are a nice way of telling how much work you've done for each project. If I needed to "get a life" as you so rudely put it, I'd be wanting more credits not less. Try engaging your brain in future before putting your fingers to the keyboard.
If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.
WhiteWulfe wrote:Peter Hucker
And then there is a project like Bitcoin Utopia where someone can earn 12 million credits in about 65 seconds as long as you have the right hardware. Zero flops involved just asic boxes crunching to help keep some of the stats sites and Boinc Projects running. When asked even Dr David Anderson, of Seti, commented that even that amount of credits was TOO LOW compared to the amount of work being done!! It was hoped that asic's would be used at other projects too but so far that has not happened, probably due to the lack of support from the asic hardware folks and the wide diversity of asic crunchers being made. Trying to program Boinc to use them is NOT easy and resources are very tight at most projects, but Projects like Seti's Astropulse Team have said it would be a huge benefit if they could use them. I have no idea about any other projects, but others have speculated there are several who could benefit from them.
Can you just imagine the brewhaha being bantered about if those types of machines were brought on line for the general Boinc projects? People with cpu's AND gpu's wouldn't stand a chance comparing credits then!!
For me credits are only worth comparing within a given Project, comparing project a to project b seems okay until you dig down and realize that each project has it's own programmers and developers and either optimizes or doesn't optimize it's apps all with almost zero outside help whatsoever!! That means that even IF credits were standardized across the Projects people crunching project a would STILL get more credits than people crunching project b and then how many people would project a be able to retain?!! Good projects would go away and those with the resources to have better programmers and developers would stay. Keeping the project credits non aligned means we crunch for our own reasons, not for credits. BTW 100 Billion credits and 10 bucks can get you a large cup of coffee at Starbucks, but without the 10 bucks you aren't getting ANY coffee!!
Seti and the Einstein Project
Seti and the Einstein Project are 2 different Universes to me and I wouldn't think about comparing them in any way.
But I stopped running any Seti tasks back in 2005 when this project started.
I think the probability of me seeing any alien creatures flying around or sending me any signal would be as good when I look up in the sky at night here.
So all I do is work here and for CERN........but that is just me.
Most ASICs I used for bitcoin
Most ASICs I used for bitcoin burnt themselves out (and yes they were well cooled). They're just not as robustly made as graphics cards.
If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.
Mikey wrote:It was hoped that
That's the whole point of "application specific integrated circuits" - they are hard-wired to perform one task and nothing but that task well. And costs a lot to design and manufacture them, so people generally only build them for large to very large markets. And that's why we're using general processing units for most task.
If someone built an "FFT accelerator" ASIC, that might be able to help several projects including Einstein and SETI. However, GPUs are already pretty suitable for that task, hence both projects have a GPU app.
Before anyone creates an ASIC for something as specific as one application of a BOINC project, they should rather implement GPU acceleration with HSA and FPGA acceleration.
Designing and fabricating chips properly is not rocket science.. I'd say it's way more complicated than this. There are good reasons companies usually take a long time to "get it right". But the Bitcoin gold rush didn't leave them time. they "testing" was probably running the ASICs internally for the first few months, when they were the most profitable, before selling them.
MrS
Scanning for our furry friends since Jan 2002
ExtraTerrestrial Apes
It wasn't the chips that blew, more the power circuitry on the board.
If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.
Peter Hucker
Let me guess you were running the U3's? They STILL quit after a while, sometimes a few years, but they still run themselves to the point of not working anymore. The Monarchs by Butterfly Labs and the numbered series by Avalon seem to last much longer, though they are not nearly as cheap as the U3's are or were.
The whole point of this though was about credits and to bring it back to that...how could the credits from an asic be counted since it doesn't do flops in it's processing? You would have to come up with something else that would then be just as contentious as flops are now in finding an agreed upon standard. To me credits are best when viewed within a single project, trying to compare them between project a and b usually ends up like comparing apples and oranges and you end up with a fruit salad.
Bitcoin Utopia doesn't use cpu's or gpu's anymore, the difficulty with Bitcoins has made them very inefficient, so trying to compare BU with Einstein or Seti that does not use asic miners but does use cpu's and gpu's is unfair at best and a nightmare at the worst. BUT since BU IS a Boinc Project you can't just say 'we will just ignore that Project because it doesn't fit our thinking', so let's just ignore ALL Projects that don't do things 'the Seti way' and voila we are where we are now!!! Each project does it's own thing with credits and only the Stats Sites are the ones showing the combined stats. When I click on my account here at Einstein it does NOT show the total Boinc combined stats for me, it lists most of the Boinc Projects I have crunched for and each one with their own individual credits.
It's about science, not about
It's about science, not about credits. Get a life.
Om mani padme hum.
You completely misunderstand
You completely misunderstand the discussion Michael. Credits are a nice way of telling how much work you've done for each project. If I needed to "get a life" as you so rudely put it, I'd be wanting more credits not less. Try engaging your brain in future before putting your fingers to the keyboard.
If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.