Project not quite going well yet

[Dingo]

The project seems to be having a few issues with the web page as well as work units.  I have submitted a few posts on errors:

https://boinc.loda-lang.org/loda/forum_forum.php?id=2

[Daniel]

Thanks Dingo.
I joined the project today, but so far all my workunits are ending with computation errors.

[Dingo]

Thanks Dingo.
I joined the project today, but so far all my workunits are ending with computation errors.

Yes mine as well.  Did you install "gif", I thought that was my problem but I installed and still get errors.

[Daniel]

Yes, I have git installed but still get errors.
On average my workunits run for about half a minute before ending with a computation error.
I noticed that some of yours must be working - you have earned some credit.

[Dingo]

I noticed that some of yours must be working - you have earned some credit.

I think they may have made a change because the first lot worked fine, except I was validating my own tasks, then they all stopped working.

[Dataman]

So far all of mine have completed and validated. I did abort 900+ because when joined the project BOINC estimated 12 minutes/wu when in fact they take 4 hours. The first download gave me over 1,000 most of which would not complete in 7 days.

The project scientist does not know how to enable stats export. Hope he figures it out soon.

[Daniel]

The new version of the application that they released last night (version 2205.14) is working better for me.
I have several workunits in progress that have been running for almost 3 hours without error.
I should know later today whether they complete successfully.

[Daniel]

Still no luck.
They ran for almost 4 hours then ended with a computation error.
According to the Stderr output the problem is that they exceeded elapsed time limit.
I will try again.

[Cruncher Pete]

According to the Project Admin the aim is to run 1 work unit with 1 GByte of RAM that will take 4 hours to complete.  I am afraid that is far off on my machines.  My 64 Core, 128 Thread machine with 32Gbyte of RAM  will only run 20 work units.  Anything over that will result in waiting for memory.  What a waste of resources..  That rig cost me over $ 8,000.00  and I could only afford 32GBytes of RAM to run it at that time.  I am not much better off now for I can't afford to buy more RAM.  My 14 core chinese machine only cost my a couple of hundred dollars and can run 14work units.  You figure...

[Dataman]

According to the Project Admin the aim is to run 1 work unit with 1 GByte of RAM that will take 4 hours to complete.  I am afraid that is far off on my machines.  My 64 Core, 128 Thread machine with 32Gbyte of RAM  will only run 20 work units.  Anything over that will result in waiting for memory.  What a waste of resources..  That rig cost me over $ 8,000.00  and I could only afford 32GBytes of RAM to run it at that time.  I am not much better off now for I can't afford to buy more RAM.  My 14 core chinese machine only cost my a couple of hundred dollars and can run 14work units.  You figure...

Me too Pete. My 32 thread machines have 64GB RAM and can only run a few threads on Vbox projects without waiting for RAM. It appears the same with LODA running GIT.

LODA will have to make some changes if it wants to attract curnchers. No checkpoints is a show stopper for me. Mr. Krauss says that it is not a problem for the project but it certainly is for crunchers. 4-8 hours with no checkpoints is not acceptable. Anyway, I set the project to no new tasks and will drain the queue. I will keep checking the forum to see if it becomes more cruncher friendly.

Maybe it is because I am a "grumpy old man" but I can count the number of projects I want to support on one hand. 

[shift]

If you are running LODA, have a look at your recent tasks. I and others have noticed they are all ending in compute errors (after 2 hours mostly).

EDIT: They have fixed it now.