@IAN&STEVE C. and KEITH MYERS. Thank you for your help! I got it working now. Unfortunately, I would have to dethrone you both from the top of the leaderboard.
There are only two questions left pending in my mind - would my considerable contribution to this project lead to a great scientific discovery, and how many Gridcoins can be generated with that much processing power. I almost feel guilty capitalizing on the fact that I'm using multi million dollar worth of hardware for free and not paying the cost of electricity.
I calculated the power consumption of the GPUs alone 1000 Nvidia Tesla T4 x 70 W each = 70000 W. Running the servers 24/7 for 30 days would consume a staggering 50.4 MWh. With a national average price of electricity in the US at $0.168 per kilowatt-hour (taken from here https://fred.stlouisfed.org/series/APU000072610) The cost of electricity for 30 days would be $8467. And that is just the GPUs, not accounting for losses in the power supplies or any other component on the motherboard. Now an even better question would be how much profit (and it's 100% profit) from Gridcoins converted to US dollars can be made in 30 days.
This is your future leaderboard leader signing off.
you wont make any "profit" from Gridcoin. I mean, it's all profit when you're not paying electricity, but not profit if you're including any of those costs. the price of GRC is already falling.
what you earn from GRC is dependent on your magnitude, and how often you get a payout depends on how much you're staking in your wallet. you'll get a higher magnitude contributing to multiple projects actually.
why the 30-days timeline? is that how long you have access to these systems?
My understanding is that all VMs on such a server of yours share the same filesystem. Thus when you start boinc (probably the standard BOINC client from the distribution), they all run in the same "data directory", usually /usr/lib/boinc IIRC.
There are a couple of command-line options you can give to the BOINC client that are mean to allow you to run multiple clients on the same server. These include
"--dir /path/to/unique/dir": you should create a unique dir for each client, and start it to use that as the data dir. Or 'cd' to that directory first, then start the client - should have the same effect.
"--no_gui_rpc:each client will try to open a port on which it will listen for commands from a GUI (or boinc_cmd). Use that option to avoid this, and in particular avoid conflicts between multiple clients
[Edit] I do these things regularly, it's has been done before ;-)
[edit2] Sorry, wrote that before I read your message #208905
> would my considerable contribution to this project lead to a great scientific discovery
All of our searches are basically lotteries - there are occasional discoveries being made, and the likelyhood for your specifically grows with your contribution. However it's still possible that someone else is the luckier one that runs the specific task which bears a new discovery.
I'm not doing this for profit. And I know there is none if you pay just the electricity alone.
The servers that I'm dedicating to this project, belong to a major, major company. They are left idling most of the time, because Nvidia Tesla T4 is an obsolete piece of data center hardware. So, instead of consuming electricity doing nothing, I thought it would be of scientific benefit to put that computational power to good use.
According to this https://www.gridcoinstats.eu/cpid/66b4359476520659b8db798e6d933131 you generate 1227 gridcoins daily, across multiple Boinc projects, so in 30 days that's 36810. At current price of 0.0097 $ per gridcoin that equals $357 per month. I assume you pay the cost of electricity yourself. I also assume you are not dedicating the rather impressive hardware I see on you profile to all these projects you are contributing to, for the purpose of making a profit. So, it seems to me that you are involved in Gridcoin, just because you can get some cash, however insignificant it may be. Sure, you can make more money flipping burgers at McDonald's for a peasant wage. But, then again you have to do something in order to get paid. My point is - setting up 1000 GPUs to run some calculations is not working in the true sense of the word. So I get some cash doing nothing at all. It doesn't matter me none if Gridcoin crashes into the ground. But for as long as I can, I'm going to recieve my free money with an open arms. And given the huge computational power of these 1000 GPUs I just may get something approaching a minimal wage in Gridcoins.
To answerer your question about the 30 day timeline, no I'm not limited to 30 day access to these servers. I would have access until they are decommissioned.
I would make sure that this "major major company" is aware of and explicitly allows you to use their systems for the purposes of making money through GRC. you could get yourself in serious hot water otherwise. I would get an agreement in writing. just my .02.
and you can't just take my earnings, with my hardware, and scale that up to your hardware. the GRC earned is a flat rate across all users. so when you add one huge user to a single project, you dont just ADD your power and earn accordingly, you reduce the share from everyone else. so my (and everyone elses) magnitude contribution from einstein will go down a lot, and our reward will proportionally reduce. each project is capped at 7187.5 magnitude. if you are contributing 60% of the total project RAC, your magnitude will be ~4300. which is high, but not really that high as a total, mine is currently higher than that. so you would actually earn less than I am right now. but in reality, you're just increasing the size of the pool and diluting the contribution from other users. the project currently has a total RAC of about 850 million 166 million. looking at your runtimes for some of your T4s, and scaling linearly to 1000 of them, you're looking at about 660 million RAC. 660/(660+850) = 43.7% of total RAC contribution. 7187.5*(0.437) = 3141 magnitude. Sorry, that was a little incorrect. the magnitude only considers users who are on the gridcoin network, and the vast majority of people are not contributing to gridcoin. I pulled my 850M number from BOINCstats which accounts for most users who are exporting stats. realistically you're looking at about a 5700 magnitude; [660/(166+660)*7187.5]. which is still only a mere 17% more than my current. My Einstein magnitude would drop to about 231, but my total would still be about 4000 due to my contribution at other projects.
read about how the magnitude and rewards work via the gridcoin website. if your goal is to maximize GRC earned, you will have better success distributing your computational power across multiple projects.
USD valuation of GRC is meaningless, since there is no GRC/USD market or exchange. you'll have to exchange it from GRC to BTC or ETH, and then to USD if you so desire. so you'll eat some fees in those trades. then the biggest problem is liquidity, the daily volume of GRC trade is quite low. don't expect to trade a large volume of it at any one time unless you want to tank the price. either trade it slowly every day, or watch the market and trade when favorable.
It was never my intension to take from your Gridcoin profit. I don't care about money. This is not why I'm here. If I wanted to make money, I could just mine Bitcoin on 1000 GPUs at no cost to me and I would make profit as inefficient as that would be. So, I'm more than happy to split my computing power across other Boinc projects. That way you get to keep your magnitude at what it is now. Just say the word.
i don't care about it either. but as I showed, you coming into Einstein won't really meaningfully affect my total earnings, it would only reduce my total magnitude by about 20%. i treat my GRC earned as basically monopoly money. it largely sits in my wallet untouched, except for recently when I converted a bunch to BTC. I was just suggesting that you'd earn a lot more if you diversified on different projects. every person can do whatever they like. if you want to put all that power here, more power to you! do whatever you want. it doesn't bother me lol.
even 1000 T4s mining BTC would be mosquito's fart compared to the hashing power on the BTC network. all 1000 of them would be like 1/10th the power of a single modern ASIC box lol, you'd earn even less mining BTC than you would earning GRC via BOINC. would be better to mine a GPU coin. BTC hasnt been GPU mineable for over a decade.
One thing you also need to think about is can any single project provide you with enough tasks to keep 1000 T4 gpu's running 24/7 for the next 30 days, that might be a huge ask so spreading them out, while also earning more GridCoins, might be easier to manage as well. As an example SRBase, which is also a GC supporting project, only has 46k gpu tasks while Einstein has about 64k total tasks ready to send out, MilkyWay also supports GC's and has about 10k gpu tasks available, GPUGrid also support GC's but has zero tasks in their cache, they generate tasks as needed so your gpu's would be out of tasks and not earning you any GC's at all at least part of the time with them. In short I think this is going to be a little more time consuming than a set and forget thing with those 1000 VM's you setup!!
Here is the full list of Boinc Projects that give out GridCoins for crunching:
@IAN&STEVE C. and KEITH
)
@IAN&STEVE C. and KEITH MYERS. Thank you for your help! I got it working now. Unfortunately, I would have to dethrone you both from the top of the leaderboard.
There are only two questions left pending in my mind - would my considerable contribution to this project lead to a great scientific discovery, and how many Gridcoins can be generated with that much processing power. I almost feel guilty capitalizing on the fact that I'm using multi million dollar worth of hardware for free and not paying the cost of electricity.
I calculated the power consumption of the GPUs alone 1000 Nvidia Tesla T4 x 70 W each = 70000 W. Running the servers 24/7 for 30 days would consume a staggering 50.4 MWh. With a national average price of electricity in the US at $0.168 per kilowatt-hour (taken from here https://fred.stlouisfed.org/series/APU000072610) The cost of electricity for 30 days would be $8467. And that is just the GPUs, not accounting for losses in the power supplies or any other component on the motherboard. Now an even better question would be how much profit (and it's 100% profit) from Gridcoins converted to US dollars can be made in 30 days.
This is your future leaderboard leader signing off.
you wont make any "profit"
)
you wont make any "profit" from Gridcoin. I mean, it's all profit when you're not paying electricity, but not profit if you're including any of those costs. the price of GRC is already falling.
what you earn from GRC is dependent on your magnitude, and how often you get a payout depends on how much you're staking in your wallet. you'll get a higher magnitude contributing to multiple projects actually.
why the 30-days timeline? is that how long you have access to these systems?
_________________________________________________________________________
My understanding is that all
)
My understanding is that all VMs on such a server of yours share the same filesystem. Thus when you start boinc (probably the standard BOINC client from the distribution), they all run in the same "data directory", usually /usr/lib/boinc IIRC.
There are a couple of command-line options you can give to the BOINC client that are mean to allow you to run multiple clients on the same server. These include
"--dir /path/to/unique/dir": you should create a unique dir for each client, and start it to use that as the data dir. Or 'cd' to that directory first, then start the client - should have the same effect.
"--no_gui_rpc:each client will try to open a port on which it will listen for commands from a GUI (or boinc_cmd). Use that option to avoid this, and in particular avoid conflicts between multiple clients
[Edit] I do these things regularly, it's has been done before ;-)
[edit2] Sorry, wrote that before I read your message #208905
hth,
BM
> would my considerable
)
> would my considerable contribution to this project lead to a great scientific discovery
All of our searches are basically lotteries - there are occasional discoveries being made, and the likelyhood for your specifically grows with your contribution. However it's still possible that someone else is the luckier one that runs the specific task which bears a new discovery.
BM
@IAN&STEVE C.I'm not
)
@IAN&STEVE C.
I'm not doing this for profit. And I know there is none if you pay just the electricity alone.
The servers that I'm dedicating to this project, belong to a major, major company. They are left idling most of the time, because Nvidia Tesla T4 is an obsolete piece of data center hardware. So, instead of consuming electricity doing nothing, I thought it would be of scientific benefit to put that computational power to good use.
According to this https://www.gridcoinstats.eu/cpid/66b4359476520659b8db798e6d933131 you generate 1227 gridcoins daily, across multiple Boinc projects, so in 30 days that's 36810. At current price of 0.0097 $ per gridcoin that equals $357 per month. I assume you pay the cost of electricity yourself. I also assume you are not dedicating the rather impressive hardware I see on you profile to all these projects you are contributing to, for the purpose of making a profit. So, it seems to me that you are involved in Gridcoin, just because you can get some cash, however insignificant it may be. Sure, you can make more money flipping burgers at McDonald's for a peasant wage. But, then again you have to do something in order to get paid. My point is - setting up 1000 GPUs to run some calculations is not working in the true sense of the word. So I get some cash doing nothing at all. It doesn't matter me none if Gridcoin crashes into the ground. But for as long as I can, I'm going to recieve my free money with an open arms. And given the huge computational power of these 1000 GPUs I just may get something approaching a minimal wage in Gridcoins.
To answerer your question about the 30 day timeline, no I'm not limited to 30 day access to these servers. I would have access until they are decommissioned.
I would make sure that this
)
I would make sure that this "major major company" is aware of and explicitly allows you to use their systems for the purposes of making money through GRC. you could get yourself in serious hot water otherwise. I would get an agreement in writing. just my .02.
and you can't just take my earnings, with my hardware, and scale that up to your hardware. the GRC earned is a flat rate across all users. so when you add one huge user to a single project, you dont just ADD your power and earn accordingly, you reduce the share from everyone else. so my (and everyone elses) magnitude contribution from einstein will go down a lot, and our reward will proportionally reduce. each project is capped at 7187.5 magnitude. if you are contributing 60% of the total project RAC, your magnitude will be ~4300. which is high, but not really that high as a total, mine is currently higher than that. so you would actually earn less than I am right now. but in reality, you're just increasing the size of the pool and diluting the contribution from other users. the project currently has a total RAC of about
850 million166 million. looking at your runtimes for some of your T4s, and scaling linearly to 1000 of them, you're looking at about 660 million RAC.660/(660+850) = 43.7% of total RAC contribution. 7187.5*(0.437) = 3141 magnitude.Sorry, that was a little incorrect. the magnitude only considers users who are on the gridcoin network, and the vast majority of people are not contributing to gridcoin. I pulled my 850M number from BOINCstats which accounts for most users who are exporting stats. realistically you're looking at about a 5700 magnitude; [660/(166+660)*7187.5]. which is still only a mere 17% more than my current. My Einstein magnitude would drop to about 231, but my total would still be about 4000 due to my contribution at other projects.read about how the magnitude and rewards work via the gridcoin website. if your goal is to maximize GRC earned, you will have better success distributing your computational power across multiple projects.
USD valuation of GRC is meaningless, since there is no GRC/USD market or exchange. you'll have to exchange it from GRC to BTC or ETH, and then to USD if you so desire. so you'll eat some fees in those trades. then the biggest problem is liquidity, the daily volume of GRC trade is quite low. don't expect to trade a large volume of it at any one time unless you want to tank the price. either trade it slowly every day, or watch the market and trade when favorable.
_________________________________________________________________________
@IAN&STEVE C. It was never
)
@IAN&STEVE C.
It was never my intension to take from your Gridcoin profit. I don't care about money. This is not why I'm here. If I wanted to make money, I could just mine Bitcoin on 1000 GPUs at no cost to me and I would make profit as inefficient as that would be. So, I'm more than happy to split my computing power across other Boinc projects. That way you get to keep your magnitude at what it is now. Just say the word.
i don't care about it either.
)
i don't care about it either. but as I showed, you coming into Einstein won't really meaningfully affect my total earnings, it would only reduce my total magnitude by about 20%. i treat my GRC earned as basically monopoly money. it largely sits in my wallet untouched, except for recently when I converted a bunch to BTC. I was just suggesting that you'd earn a lot more if you diversified on different projects. every person can do whatever they like. if you want to put all that power here, more power to you! do whatever you want. it doesn't bother me lol.
even 1000 T4s mining BTC would be mosquito's fart compared to the hashing power on the BTC network. all 1000 of them would be like 1/10th the power of a single modern ASIC box lol, you'd earn even less mining BTC than you would earning GRC via BOINC. would be better to mine a GPU coin. BTC hasnt been GPU mineable for over a decade.
_________________________________________________________________________
I was wondering, do you own
)
I was wondering, do you own the computers with server CPUs on your profile here?
One thing you also need to
)
One thing you also need to think about is can any single project provide you with enough tasks to keep 1000 T4 gpu's running 24/7 for the next 30 days, that might be a huge ask so spreading them out, while also earning more GridCoins, might be easier to manage as well. As an example SRBase, which is also a GC supporting project, only has 46k gpu tasks while Einstein has about 64k total tasks ready to send out, MilkyWay also supports GC's and has about 10k gpu tasks available, GPUGrid also support GC's but has zero tasks in their cache, they generate tasks as needed so your gpu's would be out of tasks and not earning you any GC's at all at least part of the time with them. In short I think this is going to be a little more time consuming than a set and forget thing with those 1000 VM's you setup!!
Here is the full list of Boinc Projects that give out GridCoins for crunching:
https://docs.google.com/spreadsheets/d/1JWf8twPVIpOWWn1tg5358Xqh3oGa_mat1KCluYG1I8I/htmlview#gid=0