Beuwolf Mini-supercomputer Cluster

Quote from Runningbear:

You left out something else. The power bill. You might want to allow another $2000 per annum to cover that.

Runningbear
More...

If you can't afford the power; you are in deep shit ...
 
Quote from corvus:

What are you planning on doing with this?
More...

I think he's just trying to build a system that will run "Jasper" without crashing...
 
Quote from metooxx:

You spelled it wrong; BEOWULF ...
More...

Why do I have this sneaking suspicion that you have 3 or 4 of these things already running, and these are just your low end machines? :D
 
Quote from TriPack:



Why do I have this sneaking suspicion that you have 3 or 4 of these things already running, and these are just your low end machines? :D
More...

Where do you think he got the idea ...
 
this is great.
I see 2 great business models for you :
- software development
can be profitable, but be careful : hotlines cost money. also traders are volatile (!), they will switch to another software if you don't keep up with improvements. think monthly lease instead of one time fee anyway.
- hardware business
well problem is inventory. you need just in time delivery. the market for clusters is not as big as you think though.
however, it may be easier for you to build one than to actually trade.
 
Quote from aphexcoil:

I am gathering more information on the computer cluster and I have determined that I can build a mini-supercomputer for less than $10,000.

Here is the configuration of each of the client machines that I am pondering:

Motherboard: ASUS P2PE Intel 845e Chipset ($120)
CPU: Intel 3.08ghz Pentium IV ($250)
Memory: 2700 533mhz 1 Gig Modules ($150)
Network: FiberOptic 1gigahertz ($150)
Hard-Drive: 20 Gigabyte ($40)
Case: $50

Total cost for client machine: approx. $750

10 machines would run around $7,500

I don't need video-cards since the motherboard has in intergrated one. I also don't need monitors since I can log into them from the server.

Linux is free.
More...

Well, I've sat back and watched this project grow for a few days now. All I can really say is, I hope you're not just pulling my leg here. Your budget is a bit low, but manageable. When I started my cluster development, I first outlined the software layout and the starting (planned) hardware configuration. It took me several passes before I laid out how this setup would logically work. Then I spent time building it. Developing the various information pathways and untangling the many conflicting instruction loops.

For now, I would suggest you pass on the fiber plan, it can always be implemented later as your development bugs give way to smooth operations. Get those first three nodes under your belt then step back and take a breather. Re-evaluate your progress and then continue onward. Good luck! :)
 
Quote from tntneo:

this is great.
I see 2 great business models for you :
- software development
can be profitable, but be careful : hotlines cost money. also traders are volatile (!), they will switch to another software if you don't keep up with improvements. think monthly lease instead of one time fee anyway.
- hardware business
well problem is inventory. you need just in time delivery. the market for clusters is not as big as you think though.
however, it may be easier for you to build one than to actually trade.
More...

ohhh ohmost forget.,.. grood lik two you
 
You may want to build your software first and test it on a limited data set before embarking on such a grand journey.

Find other ways to test that a potential edge is there to be found before you begin. This may sound catch-22 but there are ways to identify the potential existence of inefficiencies without identifying the specific inefficiency itself.

This could save you lots of time or narrow your focus such that you wouldn't need such long processing cycles.
 
You must log in or register to reply here.
Back
Top