Gridcoin (GRC) - first coin utilizing BOINC - Official Thread

[TribalBob]

Are we sure the scrypt-sleep function is working properly...? My BOINC Daily Credits seems to actually be decreasing (I've gone from 1373 down to 1360 credits since block 62000)... If my gpu is spending time sleeping it should be working on BOINC projects and INCREASING my daily credits, shouldn't it...?

[1713284010]

1713284010

[1713284010]

1713284010

[1713284010]

1713284010

[1713284010]

1713284010

[andrewp]

Are we sure the scrypt-sleep function is working properly...? My BOINC Daily Credits seems to actually be decreasing (I've gone from 1373 down to 1360 credits since block 62000)... If my gpu is spending time sleeping it should be working on BOINC projects and INCREASING my daily credits, shouldn't it...?

It's working fine on my rigs. How many GPU tasks have you done since then + what projects?
Sometimes there can be a delay of x hours or even a couple days before the BOINC project grants claimed credit for returned tasks. So I would wait a couple days to see.
Also if you go to the projects homepage you can check for your returned tasks that are waiting to receive credit.

Hope this helps,
Andrew.

[traderman]

Which Boinc projects are you running?

Are we sure the scrypt-sleep function is working properly...? My BOINC Daily Credits seems to actually be decreasing (I've gone from 1373 down to 1360 credits since block 62000)... If my gpu is spending time sleeping it should be working on BOINC projects and INCREASING my daily credits, shouldn't it...?

[s4mmy]

I have two rigs mining and both are dropping from sync. Anyone has sync problems or is it just me?

My boinc stats also decreased quite much at first but now my points are climbing up very oddly

edit: explorer is out of sync too

[andrewp]

My boinc stats also decreased quite much at first but now my points are climbing up very oddly

BOINC average credits or total utilization?
What CPU do you have?

Andrew.

[TribalBob]

Are we sure the scrypt-sleep function is working properly...? My BOINC Daily Credits seems to actually be decreasing (I've gone from 1373 down to 1360 credits since block 62000)... If my gpu is spending time sleeping it should be working on BOINC projects and INCREASING my daily credits, shouldn't it...?

It's working fine on my rigs. How many GPU tasks have you done since then + what projects?
Sometimes there can be a delay of x hours or even a couple days before the BOINC project grants claimed credit for returned tasks. So I would wait a couple days to see.
Also if you go to the projects homepage you can check for your returned tasks that are waiting to receive credit.

Hope this helps,
Andrew.

Hi, thanks for the info! How can I check my completed GPU tasks...? I've been using BOINC for a year or two now and never learned how to do that, lol...

Which Boinc projects are you running?

I am signed up to pretty much every project except for "DistrRTgen", "GPUGrid.net" and the two that require purchasing sensors in order to participate. I've thought about dropping some projects but I have been running these ones for about a couple of years now...

Can someone tell me exactly which projects have GPU tasks or tell me how I can figure out which ones have them myself...?

[s4mmy]

My boinc stats also decreased quite much at first but now my points are climbing up very oddly

BOINC average credits or total utilization?
What CPU do you have?

Andrew.

avg credits. 4-core athlon x4 640. Using 2 cores to boinc if all 4 cores (or 3) are boincing then my hashrate decreases.

[andrewp]

Are we sure the scrypt-sleep function is working properly...? My BOINC Daily Credits seems to actually be decreasing (I've gone from 1373 down to 1360 credits since block 62000)... If my gpu is spending time sleeping it should be working on BOINC projects and INCREASING my daily credits, shouldn't it...?

It's working fine on my rigs. How many GPU tasks have you done since then + what projects?
Sometimes there can be a delay of x hours or even a couple days before the BOINC project grants claimed credit for returned tasks. So I would wait a couple days to see.
Also if you go to the projects homepage you can check for your returned tasks that are waiting to receive credit.

Hope this helps,
Andrew.

Hi, thanks for the info! How can I check my completed GPU tasks...? I've been using BOINC for a year or two now and never learned how to do that, lol...

If you go to your account for a project and look for tasks, there is an info at the right - if it is/was GPU task, there should be either CUDA or ATI at the end.

Which Boinc projects are you running?

I am signed up to pretty much every project except for "DistrRTgen", "GPUGrid.net" and the two that require purchasing sensors in order to participate. I've thought about dropping some projects but I have been running these ones for about a couple of years now...

Can someone tell me exactly which projects have GPU tasks or tell me how I can figure out which ones have them myself...?

Here is a list of BOINC projects that shows which work with GPUs.
https://boinc.berkeley.edu/projects.php

Andrew.

[andrewp]

My boinc stats also decreased quite much at first but now my points are climbing up very oddly

BOINC average credits or total utilization?
What CPU do you have?

Andrew.

avg credits. 4-core athlon x4 640. Using 2 cores to boinc if all 4 cores (or 3) are boincing then my hashrate decreases.

I'm not sure if that happened due to scrypt sleep feature. Sometimes, some returned BOINC tasks that you received credits for get rejected later by the project system and the credits are substracted.
That happens to me quite regularly..BOINC just does that.

Andrew.

[traderman]

Scrypt sleep started @ block 62000 if I am not mistaken. So it is already activated.

[TribalBob]

Are we sure the scrypt-sleep function is working properly...? My BOINC Daily Credits seems to actually be decreasing (I've gone from 1373 down to 1360 credits since block 62000)... If my gpu is spending time sleeping it should be working on BOINC projects and INCREASING my daily credits, shouldn't it...?

It's working fine on my rigs. How many GPU tasks have you done since then + what projects?
Sometimes there can be a delay of x hours or even a couple days before the BOINC project grants claimed credit for returned tasks. So I would wait a couple days to see.
Also if you go to the projects homepage you can check for your returned tasks that are waiting to receive credit.

Hope this helps,
Andrew.

Hi, thanks for the info! How can I check my completed GPU tasks...? I've been using BOINC for a year or two now and never learned how to do that, lol...

If you go to your account for a project and look for tasks, there is an info at the right - if it is/was GPU task, there should be either CUDA or ATI at the end.

Which Boinc projects are you running?

I am signed up to pretty much every project except for "DistrRTgen", "GPUGrid.net" and the two that require purchasing sensors in order to participate. I've thought about dropping some projects but I have been running these ones for about a couple of years now...

Can someone tell me exactly which projects have GPU tasks or tell me how I can figure out which ones have them myself...?

Here is a list of BOINC projects that shows which work with GPUs.
https://boinc.berkeley.edu/projects.php

Andrew.

Thanks much!

[andrewp]

Thanks much!

I'm glad I could help.
Andrew.

[traderman]

I just checked one of my rigs and there was a sync problem. I am not sure if this is related to the scrypt sleep or not.

Once I restarted the wallet client it was back to normal.

I have two rigs mining and both are dropping from sync. Anyone has sync problems or is it just me?

My boinc stats also decreased quite much at first but now my points are climbing up very oddly

edit: explorer is out of sync too

[Gridcoin]

Linux Clients:

For everyone elses benefit and to clarify for Zulzedd and heretolearn:  I am going to explain the Linux dilemma in historical terms, current status, and the abstracted wine architecture so you may understand all sides without drawing false conclusions.

1) Regarding slow development and launching without Linux: We launched October 17th 2013.  At that time, I already had two Indian programmers working for me full time (out of my pocket), and they were not cheap, and their task was to finish the equivalent of what I did in Windows in Linux and Mac.  By the end of the year, what I received was not enterprise quality or compatible enough with the boinchash security portion to go live.  I hired a 3rd Linux dev, to fix the problems.  This attempt was much better, some of the code was compatible, but he took a cryptographic approach inside c++ that would expose the class to fraud.  I was not about to compromise security, so I had some disagreements and he asked me to spend $5000 to fix the problem.  I had some issues spending more for a pie in the sky promise and he dissapeared.  To add to the drama, I had some specific requirements to maintain security as far as the chosen classes, interfaces between windows and Linux that had to be met.  When I consolidated the code review, the code was rewritten Microsoft code in c++ that had certain flaws that hackers can use to attack us.
    On the windows side, we have our cpuminer integrated into our gvm.  We also have the ability to verify credits from the boinc API.  If we were to port the Russian version, we would no longer have a GVM and I did not want to compromise security.  So in essence, we burned through money and time and ended up at Christmas time, when I took over the Linux side.  In the next 60 days, I ported the necessary underlying regsvr32, security descriptors, GVM, and boinchash compatibility to Linux.  Wine is required to run my version, but security is maintained.  In my humble opinion 60 days was a relatively short and aggressive timeline and put us 90% of the way back on schedule to release a Linux and mac version.  I also hired two architects from Odesk who also costed me substantial money and that code is also useless as it is not compatible.  The requirements are relatively complicated.  Heretolearn assumes the client must be native but does not understand the risk involved.
2) Regarding Wine:  I want to point out, we are NOT emulating windows by running gridcoin-qt in wine.  The client is being re-written with a compatibility layer between Microsoft c#.net and Linux in a way that the functions can be executed on any platform; Linux;mac & wine.  Before we go into Why that is the case, the point is we have 100 megs of source code to do much higher level operations dealing with SQL, the sql ledger, the api verification, and the boinc hash, the cpu miner beacons, etc.  If we were to maintain 3 codebases:  We would need 3 Rob H's full time to debug and maintain the code over the long term.  But even simpler than that, I have chosen 1 codebase so that we can move on to polishing gridcoin for the end user.  If we can maintain one code base, it will be a magnitude better for Gridcoin in the long term, as the plans are not to create a simple coin that can send money and mine, the plan is to create a coin that can do confirms, help humanity, have internal voting for our future foundation, sql ledger, all kinds of features other coins don't have.  And you are not living in the current state of reality if you think you can write all of that 3 times, do 3 releases, support 3 versions "easily".  It is NOT the case.  Maintaining software takes hundreds of hours per year.  Think of the 80%/20% rule for new development and maintenance.

Next:  Installing wine on your Linux or Mac system is a single command: apt-get install wine
  It's a dependency for gridcoin-qt; it has nothing to do with "wanting" to install wine.  We could add it to the installer!
 Wine is not an emulator.  So, the code running on Linux is running natively. 

So, please get it out of your head that this is not the way to go, is not native code, and that we are slacking, because all of these things are false.

I hope I have justified the reason that wine will work:  When features are added to Gridcoin in the future we will run a deploy script and all 3 versions of the client will have an updated installer with a compatible version that runs on all platforms.  The version inside "about" will be the same, and therefore we will support one codebase.

Regarding downsides with running Wine related to performance:  There ARE NO DOWNSIDES.  Read the wine FAQ if you don't believe me.  The code is running natively.


Cheers,
Rob Halford

[Gridcoin]

I wrote about same things about 2-3 weeks ago(https://bitcointalk.org/index.php?topic=324118.msg4610130#msg4610130). I switched to pool mining yesterday and i see same - no payouts from pool, but i got 1 150 GRC payment, as if i solo mining. Previously i mined in pool for 3-4 days and was gettin payouts at 30-70 GRC after week (!) from when i started to mine in pool. I didn't calculadet profits, but seems pool mining working correctly, but with great latency because of Wallet Hour.

Thanks, zulzedd. I had forgotten about your earlier post - back then I had not yet tried pool mining so it went right over my head at the time.

Sounds like we have similar effects when pool mining. Did you notice whether the address your pool mining payouts went to was normal (presumably your default/leaderboard address)? Mine - only the one so far - showed as being sent to a receiving address I use for a different purpose, one that the mining pool presumably should not have been aware of. I'm now guessing it's a glitch in the wallet display - perhaps it shows "last referenced address" or some such thing.

I'll keep this machine in pool mode a while longer and see what develops.

All,

Regarding PoolMining:

I want to reiterate this:  Its a huge risk for us to open Gridcoin to a standard pool: Its hard to enforce a network packet containing Boinc authenticity information and "believe" the pool operator is going to be trusted with that information; it opens us up to fraud.  If that were NOT the case, I would add that to our spec and work with a pool mining operator.  No one likes a central authority, so we built integrated pool mining inside the client.

In light of that, everyone liked the old integrated pool with the small payments - the only thing they hated was finding the first block.  I viewed that as a miniscule problem; a training issue.

When we went to "Wallet Hour" payments to cut down on bloated wallets; everyone jumped ship and now we only have 23 pool miners. 

The root of the problem is the current pool only pays you in your wallet hour when another pool miner finds a block.  Its easy for a few of the pool miners to lose faith in it when a couple of days go by with no payments.

What I propose as a bridge between now and when we have 100+ pool miners is modifying the client to pay in 4 hour blocks; 2 hours before and 2 hours after the wallet hour.  I think this will satisfy our pool miners for the time being.

What I also want to say about our integrity:  I do not believe anyone has been ripped off- remember, any "advantage" or "disadvantage" is shared among the whole network; so no one is getting something in the pool and ripping other users.  But I will do what I can to satisfy everyone. 

So, as a stopgap I will modify the client to do the 4 hour window for now.

Grid

[Gridcoin]

I just checked one of my rigs and there was a sync problem. I am not sure if this is related to the scrypt sleep or not.

Once I restarted the wallet client it was back to normal.

I have two rigs mining and both are dropping from sync. Anyone has sync problems or is it just me?

My boinc stats also decreased quite much at first but now my points are climbing up very oddly

edit: explorer is out of sync too

Thanks man, keep us posted if you have more problems.  Also, remember that every 20 mins we restart Gridcons net layer; so it is supposed to auto-recover now.

[Gridcoin]

I just checked one of my rigs and there was a sync problem. I am not sure if this is related to the scrypt sleep or not.

Once I restarted the wallet client it was back to normal.

I have two rigs mining and both are dropping from sync. Anyone has sync problems or is it just me?

My boinc stats also decreased quite much at first but now my points are climbing up very oddly

edit: explorer is out of sync too

Thanks man, keep us posted if you have more problems.  Also, remember that every 20 mins we restart Gridcons net layer; so it is supposed to auto-recover now.

Note: This problem will slowly go away as 51% - 100% upgrade to the new version: The problem is Gridcoin is rejecting sleep blocks; I just confirmed it on one machine that went out of sync; It did auto recover.
Specifically, the fork has to go on for more than 3 blocks before the client ditches that chain and syncs fully. 

[traderman]

Hey Rob,

Thanks for letting us know all that. Really puts the complexity of this whole project into perspective.

Linux Clients:

For everyone elses benefit and to clarify for Zulzedd and heretolearn:  I am going to explain the Linux dilemma in historical terms, current status, and the abstracted wine architecture so you may understand all sides without drawing false conclusions.

1) Regarding slow development and launching without Linux: We launched October 17th 2013.  At that time, I already had two Indian programmers working for me full time (out of my pocket), and they were not cheap, and their task was to finish the equivalent of what I did in Windows in Linux and Mac.  By the end of the year, what I received was not enterprise quality or compatible enough with the boinchash security portion to go live.  I hired a 3rd Linux dev, to fix the problems.  This attempt was much better, some of the code was compatible, but he took a cryptographic approach inside c++ that would expose the class to fraud.  I was not about to compromise security, so I had some disagreements and he asked me to spend $5000 to fix the problem.  I had some issues spending more for a pie in the sky promise and he dissapeared.  To add to the drama, I had some specific requirements to maintain security as far as the chosen classes, interfaces between windows and Linux that had to be met.  When I consolidated the code review, the code was rewritten Microsoft code in c++ that had certain flaws that hackers can use to attack us.
    On the windows side, we have our cpuminer integrated into our gvm.  We also have the ability to verify credits from the boinc API.  If we were to port the Russian version, we would no longer have a GVM and I did not want to compromise security.  So in essence, we burned through money and time and ended up at Christmas time, when I took over the Linux side.  In the next 60 days, I ported the necessary underlying regsvr32, security descriptors, GVM, and boinchash compatibility to Linux.  Wine is required to run my version, but security is maintained.  In my humble opinion 60 days was a relatively short and aggressive timeline and put us 90% of the way back on schedule to release a Linux and mac version.  I also hired two architects from Odesk who also costed me substantial money and that code is also useless as it is not compatible.  The requirements are relatively complicated.  Heretolearn assumes the client must be native but does not understand the risk involved.
2) Regarding Wine:  I want to point out, we are NOT emulating windows by running gridcoin-qt in wine.  The client is being re-written with a compatibility layer between Microsoft c#.net and Linux in a way that the functions can be executed on any platform; Linux;mac & wine.  Before we go into Why that is the case, the point is we have 100 megs of source code to do much higher level operations dealing with SQL, the sql ledger, the api verification, and the boinc hash, the cpu miner beacons, etc.  If we were to maintain 3 codebases:  We would need 3 Rob H's full time to debug and maintain the code over the long term.  But even simpler than that, I have chosen 1 codebase so that we can move on to polishing gridcoin for the end user.  If we can maintain one code base, it will be a magnitude better for Gridcoin in the long term, as the plans are not to create a simple coin that can send money and mine, the plan is to create a coin that can do confirms, help humanity, have internal voting for our future foundation, sql ledger, all kinds of features other coins don't have.  And you are not living in the current state of reality if you think you can write all of that 3 times, do 3 releases, support 3 versions "easily".  It is NOT the case.  Maintaining software takes hundreds of hours per year.  Think of the 80%/20% rule for new development and maintenance.

Next:  Installing wine on your Linux or Mac system is a single command: apt-get install wine
  It's a dependency for gridcoin-qt; it has nothing to do with "wanting" to install wine.  We could add it to the installer!
 Wine is not an emulator.  So, the code running on Linux is running natively. 

So, please get it out of your head that this is not the way to go, is not native code, and that we are slacking, because all of these things are false.

I hope I have justified the reason that wine will work:  When features are added to Gridcoin in the future we will run a deploy script and all 3 versions of the client will have an updated installer with a compatible version that runs on all platforms.  The version inside "about" will be the same, and therefore we will support one codebase.

Regarding downsides with running Wine related to performance:  There ARE NO DOWNSIDES.  Read the wine FAQ if you don't believe me.  The code is running natively.


Cheers,
Rob Halford

[zulzedd]

For everyone elses benefit and to clarify...

This is answer of developer. Wide and full. This is great that you shared info about this problem with whole community and with every new user in it. Now i have understanding of problems which you got and i can explain it to potential members of community. Thank you!

[heretolearn]

For everyone elses benefit and to clarify for Zulzedd and heretolearn: ...

Rob : now that you have clarified the situation, it is easier for me and others to understand the reasons. Before this there was no way I or anybody else could know why things are the way they are. I appreciate the work you have put in and more so for the detailed response outlining the problem.

HOWEVER (try to take this positively,cause that is my sole intention)

From a user perspective, people will still begin by asking : Why not a "native" linux client ? (Obviously they would not bother to read the explanation you provided above at first go).

So I "propose a solution" : To overcome this problem of "Why not a 'native' linux client ?" question asked by the user, we can put a small tag line (or have an FAQ page) on the site that would say something like "one codebase for all versions/ smooth and easy to maintain codebase of all cryptocurrencies" or something along the lines (you get the idea) to answer such questions then and there.

So in a way we could use our "perceived weakness" for the project to our "perceived strength". That way we can not only answer future new users but also gain a significant advantage from a marketing perspective (which we need more and more). Kill two bird with the same stone strategy.

Maybe I am wrong or someone has a better suggestion. All open to it. Let me know if it makes sense.

Thanks