Folding Forum

Come on ... release a Solaris client.

I am responsible for four servers that don't do very much at night, and a workstation. These machines would be able to crank out a lot of analysis. It's hard to understand why there isn't a Solaris client.

Cheers!
Monsoon

Hello MonsoonNewZealand,
Welcome to the folding forum.

The Solaris lament appears occasionally in the forum; there is other misery to keep you company.

The answer goes like this...
Pande Group has its hands full and would love to roll out a Solaris platform (and HP-UX, et al) if only they could find the cost/benefit time to do so.

In the mean time, any cycle-stealing Linux emulator will work in a pinch.

You are to be congratulated for your interest in this noble project.

With all the volunteers this project has, there would definitely be some (like myself!?) that would be willing to spend the time and effort porting the program to these platforms. Yes, "they" would still need to spend some of their time and resources to double check whatever these volunteers create, but certainly the "cost/benefit ratio" of that effort would be low.
Other reason to reject volunteer programming help:
- the program includes some super secret intellectual property?
- opening the code might reveal some security secrets?
- "open source" projects never work!?

A little disappointed,
Cheers!
Monsoon

MonsoonNewZealand wrote:- opening the code might reveal some security secrets?

This is the reason most often given - the Pandegroup go to some lengths to keep the integrity of the data - if you've ever had a system crash which has resulted in the message "checksums don't match - starting from initial work packet" that is the Pande Group making sure that there are no errors in the data by forcing whatever work has been done to be redone due to the checksum error.

They also use 2048-bit encryption for the work as it is transferred between Stanford and donor machines, to stop forgers submitting duff work in the hope of rising through the ranks to become the top folder. Opening up the client and cores for porting by other people leaves this security vulnerable, and may result in some people abusing the privilege of having the software go opensource. (BOINC anyone?)

I'm sure there are other measures too but these two mean that the Pande Group does not have to repeat work due to data integrity fears, thereby speeding the production of useful results.

great explanation John. thanks

MonsoonNewZealand wrote:Come on ... release a Solaris client.

I am responsible for four servers that don't do very much at night, and a workstation. These machines would be able to crank out a lot of analysis. It's hard to understand why there isn't a Solaris client.

Cheers!
Monsoon

I second that!

I have two systems at home right now that's running Solaris x86/x64 and they don't really do much either.

Heck, all they would really have to do is just have a system running Solaris (well..okay...two systems...one's Solaris/SPARC and the other's Solaris/x86) and then install Sun Studio on both of them and run recompile it.

Then I'd be able to use Solaris even more (other than just for collaboration services).

Recompiling the clients would be simple, but that isn't the important part of the FAH software. Getting the MD cores to work on different platforms isn't always as simple as a plain recompile, especially as pandegroup make specialised modifications to the core software. This is evidenced by the fact that there still isn't an A2 core for Windows.

uncle_fungus wrote:Recompiling the clients would be simple, but that isn't the important part of the FAH software. Getting the MD cores to work on different platforms isn't always as simple as a plain recompile, especially as pandegroup make specialised modifications to the core software. This is evidenced by the fact that there still isn't an A2 core for Windows.

Then I wonder what compiler they use?

I don't think that I've actually seen one that's a true fully cross-platform compiler. And especially if they're "borrowing" (I'm using that term very loosely in the sense that the cores in and of themselves aren't written/developed by the PandeGroup) code (in part or in whole) from others; so it doesn't give them total control/access to do whatever it is they want/need to.

If they were to develop their own cores however, then I would suspect that like the distributed.net client, that it would truly be a cross-platform program that you can run on just about anything.

It may not amount to much in the end, but for those of us that have access/control over some of those systems, it's certainly would help the cause, especially since they lay claim to the whole "might as well use your computer's idle CPU's cycles to help us." vs the whole "might as well use your computer's idle CPU cycles to help us...most of the time."


(Not knock the work that the PandeGroup are doing, cuz never in a million years would I be able to do what they're doing.)