Dual Core CPU's

[AtW]

The only things it slows down are poorly implemented.

LOL?!?! Maintaining backwards compatibility (this includes performance) is job of the chip maker, not programmer, even though in some cases advance notice can be given that certain techniques (ie SSE) will yield better performance.

You are 100% right on Linux though - threading has never been strong point of Unix where as Windows NT is very good at it.

 

[TheMonkey]

LOL?!?! Maintaining backwards compatibility (this includes performance) is job of the chip maker, not programmer, even though in some cases advance notice can be given that certain techniques (ie SSE) will yield better performance.

You are 100% right on Linux though - threading has never been strong point of Unix where as Windows NT is very good at it.

There is backwards compatibility. There is an off switch in the BIOS.

There is also way too much backwards compatibility these days. People should develop software which is FORWARDS compatible rather than relying on eternal support for the technology that they use.

 

[Cowboy Bob]

You are 100% right on Linux though - threading has never been strong point of Unix where as Windows NT is very good at it.

To be fair I've never developed for Windows in anger since Windows 98 so have never experienced the NT threading model. All my apps have ended up being deployed on something else, even if Windows was the desktop environment. However, I have had a lot of experience with the various *nix systems (Solaris, AIX, Linux - as well as more esoteric OSs like the AS400) and I've never had a problem with the POSIX system even though I've built my fair share of multi-threaded apps.

 

[AtW]

There is also way too much backwards compatibility these days.

Yes and for a good reason - Intel won pretty much on the basis of backwards compatibility that did not require old software to be recompiled.

Microsoft is a funny company with very different people working for them - some small groups and individuals are extremely good and their work is truly exceptional, say .NET/C# are very well done, and Windowing system was very good from speed point of view at the time when hardware Windows accelerators did not really exist. Threading in Windows is also good example how well they can do things if they really want.

 

[Mailman]

Dabs will ship orders at up to 7.45pm...

I tried ordering through Dabs once but the customer service was sh1t!

Mailman

 

[AtW]

I tried ordering through Dabs once but the customer service was sh1t!

Worked fine for me, though don't expect to phone them.

 

[Mailman]

People should develop software which is FORWARDS compatible rather than relying on eternal support for the technology that they use.

So what happens in the next round when people have to develop something backwards compatible to work with your forward compatible application?

Regards

Mailman

 

[TheMonkey]

Yes and for a good reason - Intel won pretty much on the basis of backwards compatibility that did not require old software to be recompiled.

Microsoft is a funny company with very different people working for them - some small groups and individuals are extremely good and their work is truly exceptional, say .NET/C# are very well done, and Windowing system was very good from speed point of view at the time when hardware Windows accelerators did not really exist. Threading in Windows is also good example how well they can do things if they really want.

The CPU shouldn't matter. It's just a tool. That's the approach Apple takes and it works. If AMD or Sun did power efficient CPUs which fitted Apple's requirements then they'd just switch tomorrow.

.Net is a good example of what I mean by forwards compatible. No dependency on a particular hardware or software environment. It just works (tm).

 

[TheMonkey]

So what happens in the next round when people have to develop something backwards compatible to work with your forward compatible application?

Regards

Mailman

That's the job of the framework or tool you use, not the programmer.

 

[darmstadt]

I build quite a lot of dual core systems, the Intellistation 6217 A Pro and the eServer e326 using AMD Opterons and we have found that they are a lot faster in regards to our application. I am now looking at the Intel dual core in the x3800 when it comes out to replace the x260 but they don't seem as fast. I do have a T60P here which is also dual core and it actually seems slower than the T43p.