Messages

3706

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 13, 2018, 08:04:00 AM »

https://en.wikipedia.org/wiki/BOINC_Credit_System

Please read the part on RAC, is there any information on exactly how it's calculated? Even though the statement needs verification on the wiki, I think we should calculate our own from the credits of each task which would be more accurate. If the statement is true, it looks like maybe you can trick the RAC to being higher possibly by having a higher benchmark score. I will also try to find out how the credits are calculated.

Its just the half life of your recent credit added in the last month.  Actually its very accurate. 
I dont understand why you would want something different - how could anyone tamper with it?  Its not possible.  No researcher has access to the Rosetta validator or write access to the boinc network.

Its already more accurate than anyone would ever need.

3707

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 13, 2018, 08:01:38 AM »

The first machine I tried was an ARM (aarch64) and didn't work.  This is because, I believe, Rosetta does not craft their work units to process correctly in ARM.

Yeah, I can add a couple additional things need done to host an ARM project on boinc.  Its not a limitation of Boinc or the Rosetta Project.

Basically, the Build of Rosetta produces a long compatibility string, something such as "Linux 64 - ARM - Kaeplin" and when the Rosetta developer finds that string, they have to ask Rosetta web admin to add it as a project type (So then Rosetta supports Windows, Mac, Linux, Arm Kaeplin).  After that nothing works yet.  Then they have to have config manager build a Kaeplin build.  Then Network manager configure the outputs to go in the Kaeplin directory.  Then enable Boinc Rosetta to serve those workunits.  Then script a program to Create work units for Kaeplin. 

So its everything from a custom build for that proc to enabling that proc for the project.  They probably only support one Arm build at this time, whatever one is in the google play store that works on the galaxy s7.

You can make posts in the Rosetta forums asking them to support more Arm flavors.   You can get the string from the task scheduler error, where it says Project has no tasks available for "Arm xxxxxx".

3708

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 13, 2018, 07:56:21 AM »

Why do they make RAC decrease by a factor of 2 every week. So even if I'm doing 200 cobblestones a week the next week it's a 100 RAC?

http://boinc.berkeley.edu/wiki/Computation_credit

RAC is a very cool float.  Think of it as the half life of a piece of uranium.  It takes computing power to build it up, then it decays over a one month period with the half life being 2 weeks.  Its very useful to show exactly where a researcher is in the work curve.  If you have been researching for one month solid before you come to Biblepay, we know with certainty how much effort you already put in and can confidently pay you in the next superblock based on your RAC.  It ends up being a perfect representation to rank Rosetta users side by side using disparate equipment.  ( Our biblepay magnitude is also based on the sum of Rosetta RAC and our available budget, so your in-client magnitude is Your next DC superblock share as a percent * 10).

If you want to see work done over a one day period, you can instead look at total credit delta.  Look at todays total credit minus yesterdays total credit.

3709

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 13, 2018, 07:49:34 AM »

That's great! And the exchanges can have a different version of the daemon to run without the additional libraries?

I was able to do it with Open SSL, so they dont need anything.  I made sure to say around our function - If this is Not a Sanctuary - exit immediately.  This way when Bittrex audits our code and says what is this for, they wont get question listing us, as they are highly suspicious of anything that has outside access inside an exchange.  As long as they are not a sanctuary, their node will not do anything funny.  It should pass code review.

But anyway, no they (and we) don't need any additional software libs now.

3710

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 13, 2018, 07:46:03 AM »

Anyone know how to maximize BOINC CPU in command line on Linux?

I just checked on my Linux machine with top command noticed only 7 boinc/rosetta threads running at 100% CPU,
when I was expecting to see 16 threads at 100% CPU (like I saw after first installing BOINC and linking Rosetta)

I found some config files in the /etc/boinc-client folder: cc_config.xml and global_prefs_override.xml

References:
http://boinc.berkeley.edu/wiki/Installing_BOINC_on_Ubuntu
https://boinc.berkeley.edu/wiki/BOINC_Data_directory
http://boinc.berkeley.edu/wiki/Boinccmd_tool
http://boinc.berkeley.edu/wiki/Client_configuration

This guide shows a giant config file:
https://www.digitalocean.com/community/tutorials/how-to-set-up-seti-home-on-ubuntu-14-04-or-debian-7
cpu_usage_limit option stands out

Ill poke around some more tomorrow

Just an fyi btw, another way to do it is to copy the client_state.xml in linux to a backup, then install boinc manager GUI on linx (apt-get install boinc-manager), then run the boinc-manager from the linux desktop (requires graphics) then go into Advanced mode from Tools | Advanced, then make the changes using the settings and preferences, then take a diff of client_state.xml (see what changed) then export those keys to a file, then you will know what to change on every new box in the future.

3711

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 12, 2018, 09:40:53 PM »

EDIT:  Never mind, I decided Im going to jump on working on headless PODC now... It will be more professional...

Alright, finally cracked it.  Starting tomorrow we can have headless Sanctuaries again.

Ill get this networking file manager and CPID business logic rule tested in the morning then we will need a mandatory testnet upgrade so we can all test the enforcement of the botnet busting rules (and test the headless).

3712

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 12, 2018, 09:38:27 PM »

Hello from Slovakia,  my name is Imrich Schindler   (for klondike)
First, thanks to all, which cooperate on this nice idea(Rob, Togo, Alex, Lichtuscher .....) It is great and hard  job.
 I am minnig biblepay since december 2017. I like idea of Biblepay and I'm continualy reading main bitcointalk thread and now this one about PODC testing.

I am testing PODC in testnet on linux. I have installed biblepay 1.0.8.9 from  Lichtsucher ppa(very usefull) and BOINC manager on ubuntu 17.10 .
Testing BOINC on android too (honor 6X)
From my side of view all is working.

Excuse my English, my children are better in english then me .

Welcome aboard!

Glad to have you here, now we have users from High Tatras, Skok, Sucha Bela, and Bratislava!  Its an honor to have you here, we roll out the red carpet.

3713

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 12, 2018, 09:34:48 PM »

Rob, I'm not sure how far along you've gotten with the validation but I think the best way to do this is to have our own Rosetta manager. I know your aware that there are many task managers out there and they all call out the Rosetta client to do the computational work. From there, you can tie in a piece of code that tells Bible pay that the client is actually running and keep track of the tasks run. From our own manager, you can then also send out the performance data for validation use.

That is what I've come up with after thinking about it through out the day. Then, the only piece left that is not currently in our control is the Rosetta and Rosetta mini clients.

Possibly.  I think I can actually scrape all that intercommunication between rosetta and boinc anyway (using the opensource boinc protocol), but I was thinking of that part of it, and whether we trust that log or trust our auditor - Im wondering if we start by assigning an auditor, thats you, we create some web reports (thinks like task turnaround speed per machine, ghz & cores per machine, with some metrics) and make the auditors record the "reference machine"for biblepay.  You can read your row and "certify"it (thats our scale).

Then I base the integrity value on other rolled up users values vs. the auditors.  Then we look at it and try to figure ways to improve it - and try to figure out how "weak"it actually is.

3714

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 12, 2018, 05:19:08 PM »

EDIT:  Never mind, I decided Im going to jump on working on headless PODC now... It will be more professional...

3715

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 12, 2018, 04:52:57 PM »

I understand, but why it is needed to have gui mode, if you can install gui mode and libraries on command line too.
So will it be necessary in prod version too, or it is just temporary?

Its different, because headless doesnt compile against the QT5 libs.


The Sanctuaries have to process the PODC files, so they need the QT libs in memory.

Im working on the integrity side now, Im trying to create a pool report that shows some type of integrity level per researcher.
EDIT:  Im going to work on headless PODC now...

3716

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 12, 2018, 04:48:22 PM »

I was little bit afraid because of boinc GPU support, but then I've found this thread https://boinc.bakerlab.org/forum_thread.php?id=6475 and I'm calm again

I'm glad you found that thread!  Thats awesome.  That brings up two Extremely important points we can add to our arsenal, yes three:

Headcount of high paid developers working on rosetta commons Protein Docking Research:
There are 150 developers in 23 universities and laboratories working on various aspects of the coding (Rosetta Commons). So, the idea has been considered from many many perspectives. And, as you quoted, some serious efforts have been made to utilize GPUs.  (In contrast to a few who worked on pow initially, and it is not nearly as elaborate).

No edge for GPUs yet:
Because of the low memory per core, running Rosetta in GPUs takes a little longer than on the CPU.

Anti-hack limiter:
The scientist said that they reject bad or tampered with results, and increment the cpids error count.  If I remember correctly, a few years ago when I ran my boinc project (a quant project for distributed hedge fund analysis), you would be disqualified for asking for more work if you kept sending us bad results (for a certain amount of hours, called the back-off period).  So that is also Stunning, because if Swongles hacker were to try to falsify protein fold results, they would find themselves unable to download new work.

This project is getting better every day .

3717

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 12, 2018, 12:41:55 PM »

Rob, I think I already helped him solver the problem over discord. Maybe there is some sort of communication roadblock.

Yeah he probably started more than one CPID anyway.

Irmisch was your friends CPID on our team.

I just wanted to show how to pull up RAC by a CPID on the web.

3718

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 12, 2018, 12:38:39 PM »

Hi, I've tried boinc on my ARM based Samsung Galaxy S7... It works fine, after approx. 9 hours it solves 4 tasks (I set up 4 cores) with approx. 60 credits/task.

Thats awesome dude, this is extremely good to know. 

I wonder what the S7 will earn someone in a full month by itself in 90 days... 

3719

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 12, 2018, 12:36:43 PM »

ROB what are you talking about... who is imrich?

im here

https://boinc.bakerlab.org/rosetta/hosts_user.php?userid=1982089


is this true?

10% Charity
10% IT/PR/P2P

35% Sanctuary
35% BOINC Rosetta Resarching Mining
10% POW Heat Mining

So you lied about your CPID.   If the link you posted is correct your cpid is actually:
e420467693e9d8c8f163e283bef5068a

LOL,  Why even answer any questions if you wont tell the truth about the most basic starting point.

LOL

3720

TestNet Discussion Archive / Re: BIBLEPAY -TESTNET TESTING THREAD-Proof-Of-Distributed-Computing(Cancer Research)

« on: February 12, 2018, 09:54:53 AM »

Did you get it working?

One helpful hint I want to share with you all:  After you associate a CPID and while you still have 0 magnitude, If you can get the newbie to relay to you the amount of RAC from his boinc screen it should match the:
CPID_RAC: XXX  message in 'exec getboincinfo'.  This is going to be extermely helpful in debugging people who are < 24 hours old.  Obviously since he didnt paste his exec getboincinfo like we asked him a few times, hes obviously not even reading anything here.

So have him paste that and look at the line with "_RAC:" in it , and that shows progress between 0 hrs and 24 hours.