I'll post in here too, as Ralph message boards are 99% spam.

I got 1 task on each of 2 machines. 5GB vdi image downloaded for Python.

1st machine said download failure, not sure why. Computer may have crashed of it's own accord, it has a dodgy old GPU.
https://ralph.bakerlab.org/results.php?hostid=49339

2nd machine got it ok, but caused a computation error immediately.
https://ralph.bakerlab.org/results.php?hostid=48821

Incredible - they have 11 `tasks in progress` on ralph . I thunkd it was ded .
Though even if they did release another million python`s upon us , nnnnnn . . . .
for me this winter`s project , I am all in on cosmology@gnome ,

Incredible - they have 11 `tasks in progress` on ralph . I thunkd it was ded .
Though even if they did release another million python`s upon us , nnnnnn . . . .
for me this winter`s project , I am all in on cosmology@gnome ,

I'm fine with pythons. I overtook loads of people in the stats with those, I'm guessing a lot of folk didn't like them. There were many complaints. It's just a VB program, dunno what the fuss was about. I think it needed a modern processor, with AVX or something.

I got 1 task on each of 2 machines. 5GB vdi image downloaded for Python.

Seems that they don't implement multi-attach disk boinc feature...
It's the old app with the old problems :-(

---

What is multi-attach disk boinc feature?

What is multi-attach disk boinc feature?

It's an old story

---

Not really necessary, don't we all have large SSDs?

Not really necessary, don't we all have large SSDs?

I believe the idea is a Squid thing like LHC crunchers use to download tasks to a single pc then send them out to multiple pc's and then return them from the original pc that downloaded them thereby bypassing one of the security features built into Boinc that is that the pc that downloads a task must be the one that returns it back to the project.

Not really necessary, don't we all have large SSDs?

Great idea, waste space it's good!!

Becoming serious again, it's also a question of perfomance/optimization/etc

---

I believe the idea is a Squid thing like LHC crunchers use to download tasks to a single pc then send them out to multiple pc's....

I believe the idea is to have Rosetta@Home developers that fix code bugs and solve problems.
And, maybe, introduce features that help the volunteers.

---

I believe the idea is a Squid thing like LHC crunchers use to download tasks to a single pc then send them out to multiple pc's and then return them from the original pc that downloaded them thereby bypassing one of the security features built into Boinc that is that the pc that downloads a task must be the one that returns it back to the project.

A`Superhost` was considered some time ago ,
page "Last modified 13 years ago" https://boinc.berkeley.edu/trac/wiki/SuperHost

I believe the idea is a Squid thing like LHC crunchers use to download tasks to a single pc then send them out to multiple pc's and then return them from the original pc that downloaded them thereby bypassing one of the security features built into Boinc that is that the pc that downloads a task must be the one that returns it back to the project.

Squid doesn't do that, it just caches. PC 1 downloads a task, involving a GB disk image. Squid keeps that on the main PC. PC 2 downloads a different task, needing the same GB disk image, but squid already has it, so hands it to PC 2. It's just an internet cache, nothing specific for Boinc.

And anyway it's impossible to make it work, it's a badly ported piece of shit with no GUI from Linux. I had it running for about 2 months (after a lot of hassle), then it suddenly decided to stop working, just kept crashing. No amount of reinstalling, uninstalling, wiping every part of it, would make it go again. I tried it on another PC, wouldn't work there either. So LHC gets everything downloaded 120 times for my 120 cores across 10 computers.

What LHC should be doing is at least caching the image on each PC. I think they're doing that here. On one PC only, the first task downloads a GB of data, then another task downloads the same GB again?!? Only Stephanie (Stefan) (Computerzrmelsmelrle) could think up something that monumentally stupid.

Not really necessary, don't we all have large SSDs?

Great idea, waste space it's good!!

As I said, that kinda space is nothing nowadays. Why waste effort designing something to save what there's plenty of?

Becoming serious again, it's also a question of performance/optimization/etc

It takes a second to read that much data off an NVME. Nothing compared to the processing time of several hours.

I believe the idea is a Squid thing like LHC crunchers use to download tasks to a single pc then send them out to multiple pc's....

I believe the idea is to have Rosetta@Home developers that fix code bugs and solve problems.
And, maybe, introduce features that help the volunteers.

Developers cost money. Nobody funds science. They could ask for donations, many projects do. They could also ask for volunteer programmers, many projects do.

A`Superhost` was considered some time ago ,
page "Last modified 13 years ago" https://boinc.berkeley.edu/trac/wiki/SuperHost

An excellent idea, I've asked why it's not been implemented over at Github where the Boinc programmers hang out:
https://github.com/BOINC/boinc/discussions/5414

As I said, that kinda space is nothing nowadays. Why waste effort designing something to save what there's plenty of?

Good question. It is like one very big nation invaded another one with a lot of space avaiable inside...

---

Developers cost money.

It's not a problem for IPD

Nobody funds science.

Team of Alphafold does not agree

They could also ask for volunteer programmers, many projects do.

The main part of the cose is closed, up to now

---

As I said, that kinda space is nothing nowadays. Why waste effort designing something to save what there's plenty of?

Good question. It is like one very big nation invaded another one with a lot of space avaiable inside...

It's the resources not the space they want, the big nation has loads of room. Resources would be the equivalent of our CPU time, which we allow them to use.

Developers cost money.

It's not a problem for IPD

IPD can mean 79 different things, pick one: https://acronyms.thefreedictionary.com/IPD

Nobody funds science.

Team of Alphafold does not agree

Sometimes you get lucky, I used to work in a university, funding is like a lottery.

They could also ask for volunteer programmers, many projects do.

The main part of the cosecode is closed, up to now

No reason for it to be closed.

It's not a problem for IPD

IPD can mean 79 different things, pick one: https://acronyms.thefreedictionary.com/IPD

C'mon, you are here since 2006 and you don't know the Istitute for Protein Design?

The main part of the cosecode is closed, up to now

No reason for it to be closed.

You have to speak with RosettaCommons team..

---