... the top 50 computers, all of them running exclusively FGRP on GPU, none of them crunching GW or others, so by the looks of it those projects kinda loose out...
There are a couple of things you're not taking into account.
Firstly, whilst the top 50 would indeed have (probably several) high end GPUs per host, they may also be using their CPU cores to crunch CPU tasks. Unless you have checked each one, you can't just assume "exclusively FGRP on GPU".
Secondly, the top 50 is a tiny fraction of the total number of hosts contributing to Einstein. Their combined total production would be way less than that of the many thousands of hosts not in the top 50.
Thirdly, most people are more interested in making an efficient and cost effective contribution to science. Many will choose to support a particular search like FGRPB1G because they see the maximum contribution for every dollar they spend. I'm certainly in that category. None of my hosts are anywhere near being in the top 50 and being there is absolutely not a factor of importance to me.
Fourthly, if you look at the server status page for the number of tasks returned last week for O2AS compared to FGRPB1G, those numbers currently show ~135K compared to ~320K. Sure, there's an imbalance but when you consider that a GW task might take 10-20 hours each whereas a FGRPB1G task might take something similar in minutes, it's a pretty good indication of just how many CPU cores there are out there crunching on these long running CPU tasks.
Fifthly, any imbalance is likely to change RSN :-). With a GW GPU app in testing at the moment, and with the ironing out of any teething issues, there is bound to be a switch of resources from FGRP to GW once people can see the likely efficiency improvement. Already, mid-range GPUs are showing the possibility that GW tasks may be processed 5-10 times faster than on CPU cores. Since the detection of continuous GW is a pretty big deal, a GW GPU app should considerably increase participation in the O2AS search so that the current imbalance mentioned above will probably disappear over time.
Attila wrote:
Like if you want to be in the top 50, you have to run FGRP on GPU, and halt the rest of the projects ...
Realistically, how many people could have aspirations of being in the top 50 and why are you so fixated on the very small number it could be? I'm sure the project welcomes any and all contributions and there's no problem from the project's point of view if a whole bunch of people add extra resources of any description in an attempt to improve their ranking.
Checked all 50 of them. Some of them had a few, but not more then 5. And those who had more just aborted them CPu tasks. Its not really worth it since you run them 10k sec cpu and earn 700credits while a gpu task take 290 sec (and i run two per gpus, just because it can do it easily (tried 4 task too/gpu, but then the processing time/task grew considerably, so went back to two)), plus a little bit of CPu to feed the gpu and earn 4k credits. 1/task more then enough and it wont even use much.. around 1 percent per gpu task in my case.
By GPU they earn avarage 50 mil credits, and the avarage credit earned by gpus sum 280mil.
Im not saying, that everybody run thise just to see the highest numbers next to their names, but there are a few, who do like to see it, and since i have a too powerful computer compared to how much of it im using usually (for 4k gaming plus hefty video editing and programming was in my mind when i built it, but i very rarely do these next to my main job..), so my computer sits idle most of the time, even when im sitting in front of it, watching something. So im in that category, whod like to see that high number next to it.. XD
Attila wrote:... the top 50
)
There are a couple of things you're not taking into account.
Firstly, whilst the top 50 would indeed have (probably several) high end GPUs per host, they may also be using their CPU cores to crunch CPU tasks. Unless you have checked each one, you can't just assume "exclusively FGRP on GPU".
Secondly, the top 50 is a tiny fraction of the total number of hosts contributing to Einstein. Their combined total production would be way less than that of the many thousands of hosts not in the top 50.
Thirdly, most people are more interested in making an efficient and cost effective contribution to science. Many will choose to support a particular search like FGRPB1G because they see the maximum contribution for every dollar they spend. I'm certainly in that category. None of my hosts are anywhere near being in the top 50 and being there is absolutely not a factor of importance to me.
Fourthly, if you look at the server status page for the number of tasks returned last week for O2AS compared to FGRPB1G, those numbers currently show ~135K compared to ~320K. Sure, there's an imbalance but when you consider that a GW task might take 10-20 hours each whereas a FGRPB1G task might take something similar in minutes, it's a pretty good indication of just how many CPU cores there are out there crunching on these long running CPU tasks.
Fifthly, any imbalance is likely to change RSN :-). With a GW GPU app in testing at the moment, and with the ironing out of any teething issues, there is bound to be a switch of resources from FGRP to GW once people can see the likely efficiency improvement. Already, mid-range GPUs are showing the possibility that GW tasks may be processed 5-10 times faster than on CPU cores. Since the detection of continuous GW is a pretty big deal, a GW GPU app should considerably increase participation in the O2AS search so that the current imbalance mentioned above will probably disappear over time.
Realistically, how many people could have aspirations of being in the top 50 and why are you so fixated on the very small number it could be? I'm sure the project welcomes any and all contributions and there's no problem from the project's point of view if a whole bunch of people add extra resources of any description in an attempt to improve their ranking.
Cheers,
Gary.
Checked all 50 of them. Some
)
Checked all 50 of them. Some of them had a few, but not more then 5. And those who had more just aborted them CPu tasks. Its not really worth it since you run them 10k sec cpu and earn 700credits while a gpu task take 290 sec (and i run two per gpus, just because it can do it easily (tried 4 task too/gpu, but then the processing time/task grew considerably, so went back to two)), plus a little bit of CPu to feed the gpu and earn 4k credits. 1/task more then enough and it wont even use much.. around 1 percent per gpu task in my case.
By GPU they earn avarage 50 mil credits, and the avarage credit earned by gpus sum 280mil.
Im not saying, that everybody run thise just to see the highest numbers next to their names, but there are a few, who do like to see it, and since i have a too powerful computer compared to how much of it im using usually (for 4k gaming plus hefty video editing and programming was in my mind when i built it, but i very rarely do these next to my main job..), so my computer sits idle most of the time, even when im sitting in front of it, watching something. So im in that category, whod like to see that high number next to it.. XD