1) Get 25 to 50 ft USB, and UVGA or DVI extension cables so you can run PC in next room. Otherwise fan noise will drive you crazy.
2) Upgrade UPS. You'll need at least 1 KVA.
3) Upgrade electrice circuits to 20 amps minium
4) Upgrade Central Air Conditioning to something serious.
5) Upgrade electric service to 200 amps minimum to take care of increased electrical load for above.
That's what I'd do if I had the money and resources. What I probably end up doing is getting the VIA VT310-DP dual processor board if it is priced reasonable, can be passively cooled and I can find a mini-itx power supply without a whiney little fan.
Or are you talking about software? I've been ready for years. When is Intel coming out with a 16 way multi-core cpu? I want to see something serious.
jseigh wrote: Or are you talking about software? I've been ready for years. When is Intel coming out with a 16 way multi-core cpu? I want to see something serious.
What kinds of applications do you envision will need (or be able to use effectively) a 16-core processor? I don't think the "killer app" is written that might be able to harness this kind of power, but we might be able to envision something.
Right now, we have a chicken and egg problem in that I expect too few ISVs are ready to use multi-core processors. After the hardware is more readily available, though, I expect an expanded set of threaded applications. But, can we think of what applications or application areas would benefit best/most/fastest?
The problem is that the people who own the apps which would benefit most from multi-core cpu's probably aren't the most thread savvy programmers around. They have a huge investment in techniques to extract the most performance from a single threaded application. I don't think they're going to jump on an area that isn't their strong suit.
Message Edited by hagabb on 04-26-2005 04:29 PM
Another case. An open source file server project was looking at making some internal table lookups lock-free for the same reason, to reduce contention and improve performance. This was a year ago and I haven't heard anything about it.
So I suspect this is fairly common. Some awareness but nobody is going to make any major changes yet. What would it take to get everyone moving on this? Maybe not a "killer app" but at least a high profile example with dramatic enough performance effect that people would start to connect the dots and realize that their own product could be at risk if they didn't start to exploit multi-processing.
Start publicizing performance studies that show dramatic performance improvements using multi-processing in mechanisms commonly used by major applications.