Some of us have hardware that others don't / won't have access to. Whether it be a SPARC machine or a 24 core server, maybe even just a Mac G5 server with a lot of open accounts hosted. I dunno.
Point is I offered some people space on one of my sparc boxes and they said sure. So I got the idea of what if we had a couple of machines available to users ON A LIST that we TRUST with access to some hardware? They can get access to some stuff that they have never used for and get space for projects and build on some interesting stuff. I wouldn't mind leaving a sparc machine open to hosting.
Shall we discuss?
Edit: On top of that we don't have to give out our IP's. You can set up 5 free DNS domains here:
How much storage are we talking? As referenced in some of my older posts, I have a need to play around with 36+TB. Being able to play with that at some point and learning to pair it down to maybe fit on my system would be amazing.
The thing it, it's a minimum of 36TB. I'm estimating ~70TB before I can go in and prune and compress. It's the Gene Expression Omnibus. To be honest, the most "practical" solution is for me to sink 40-80 hours of dev time after I finish a large refactoring project. I could use dev time more than anything else but school is keeping what mental energy I have from being used to do research. Once I get the development finished, I could skip much of the superfluous data and take strictly what I need to compute on.
I have a major revision coming down (effectively taking it to v3) that will be properly structured, a little faster, be extendable, etc. There is a thread around here somewhere where I got feedback about it and I'm still incorporating it.
I'm actually doing one better through code. I'm going to made a grid/cluster compatible distributed parallel solver. I'm going to be implementing it through BOINC, so when I'm done I'll be kicking the folding@home thread pretty hard to get people to help out.
Well TBH X86 is great and all, however PPC, POWER, and SPARC are stupid efficient at processing numbers. Where in gene folding you would want 10 4 core X86 chips from intel you could do 1 12 core PPC chip to do the same amount of work or more (don't quote me on numbers, just that the PPC chips for example are stupid efficient). Sparc is used in super computers for a reason :P and Sparc is easily available in a lot of places. Might be older, but I imagine US3i could be useful maybe.
Lol I dunno release your code for your client and let people port it :P you'll get some interesting results I bet.
well depends on os.. basically if your running say linux, you can just compile the c with gcc on sparc and run the same c code, but is risc is designed differently than cisc, cisc is you try to optimize the processor to do everything, where risc is you optimize it to do the general purpose so much you can still do more advanced stuff by it just being that much faster at the basic shit.
so if you want to do things like advanced floating point shit sure the intel whatever is fast, but if you want to do file compression or encryption or something risc has its advantage. arm is another part of risc and performs similarly, you get lots of integer performance/can do some stuff significantly more efficiently but depends what you code/how you try to solve whatever problem
but onto the super old stuff i got kicking around dunno