what is the most perfectest OS for running EINSTEIN@HOME?

Hello,

First, not sure why your post got rated to a negative 29, someone here must not like your windows theme :)

Anyways, I would say running the windows binary under Linux using Wine has shown the best results for me. Even though it crashes a few times, it seems to process the Workunits quicker than any other configuration I have tried.

Thanks!
Jordan Wilberding

> my opinion:
>
> WINDOWS.
>
> reason: 98% of the world pupolation (5,7 billion peoples) uses windows. this
> got to say somethin about it´s:
>
> - efficiency (and speed)
> - stability (and comfortablitiy)
> - profesionalty (and mutli tasking ability)
>
> however i am open minded and lookin forward to hear other argumentative
> reasons to consider another OS (but i am heavy to convince).
>
>
> greeetings,
>
>
> ps.
>
> P.S. by the way: my first OS i use was GEM from ATARI 1040 STFM. but i prefer
> windows because it has more colors (red green blue yelow) and the garbage gets
> full when i throuw something in it...like in reality. unless there is a
> SwweeeiEet+t+E hiddden micky mouse in the garbage who likes to eat some cherry
> cake crumbles or cheeese pieces (my favorit foods) which i put in the garbge.
>

> my opinion:
>
> WINDOWS.

hihihihihi, me finding precioussss with calculator. me don't know about opparating systems. me only looking for preciousssss.

You point is well noted. However, the *nix einstein binaries are not always compiled to take advantage of current chip technologies where floating point math is most important. This, in itself, steers the project towards Win-Intel or 68K-OSX based operating systems which take the highest advantage of related chip technologies. You can't say that all of the calc binaries run exactly the same across all available O/S's. So, there *is* a preferable calculation platform. Correct? Mac OS running dual-CPU is obviously going to provide the best math performance due to RISC architecture, while a x386 Linux ELF binary is going be the worst. Am I wrong?

> This question seems to me to miss the point. Choosing an OS for a particular
> DC is like buying paper just to fill the recycling bin. The whole point of a
> Distibuted Computing project is to re-use whatever you've got.
>
> That is my point in a nutshell, but because I like the sight of my own
> typing, I'll say it all again in much more detail.
>
> In a Grid computing project, that is a project which owns its own productive
> machines there would be 100's or 1000's of identical machines all running the
> same code and all with the same operating system. There it would make sense
> to ask which OS is best. Given that Grid projects depend on networking, they
> have sometimes gone for the Unix/Linux variants due to the fact that these
> tend to be more efficient at networking and easier to set up for it; however
> other criteria including the ones you mentioned may push the choice in another
> direction.
>
> However, this is not a Grid project but a distributed computing project: the
> computers are owned by a diverse bunch of donors who all own their computers
> for some other purpose. It might be games, surfing, TV recording, word
> processing, spreadsheets, science, whatever. The DC element is secondary
> for any specific user. The DC client is designed to run as unobtrusively as
> possible. This means it should disrupt the real work of the machine as
> little as possible, and ideally not at all.
>
> This means that the most perfectest OS for running a DC project (in BOINC or
> elsewhere) is whatever OS you already have. Anything else is a mistake. To
> waste time thinking about OS's is a mistake within the context of donating to
> a DC project, important though that choice is elsewhere.
>
> For a project designer the best OS's to support are the ones (plural) that
> seem to them to have the widest base of potential donors. That is not a
> technical point but what might loosely be called a marketing one, and does
> not affect the fact that from a donor point of view the best OS is one they
> already have.
>
> Put it this way: if someone did not already have a Windows licence, would
> you really expect them to pay money just to be allowed to donate time to E@H?
> If their hardware does not run Windows (Mac, say) should they change their
> hardware too?
>
> Or should they pirate the OS and risk jail just to donate to a DC project?
>
> And if someone else does have a Windows licence no doubt they thought before
> they made the purchase. They decided it was worth the investment compared to
> a free OS and I expect they had a good reason for paying all that money, or
> at least a reason that seems good to them. Would anyone expect them to
> sacrifice that investment just for a DC project? Of course not, they will
> use what they've already got.
>
> The only exceptions I can think of are if you dual-boot or have a tool which
> lets you run one OS's code on another OS, then you choose between one and the
> other, probably on the basis of performance. It does not (in my opinion)
> make sense for anyone else to even think about it.
>
> I see your question has already been downrated to -27, and I will be adding
> another -1 to that score. I felt I'd like to let you know why (though I can't
> speak for the other critics of course).
>