Dual/Quad Core Specific Linux Distro

[xxqq]

Microsoft plan to release future versions of VISTA that need at least a dual core machine. The first part of the OS will run on the first core (continous security scanning, firewall etc). The second part of the OS will run on the second core (ordinary applications: spreadsheet, word processing etc). On Quad core, the third and fourth cores will be used for gaming, multimedia etc.

Is there any Linux or Linux+BSD combo distro with similar specification that need at least dual core machines?
Perhaps this distro will use Xen, OpenVZ, Pacifica, Vanderpool or something similar. Is this idea possible?

[techieMoe]

Any Linux distribution with a 2.6.x or 2.4.x-SMP kernel will work just fine with dual/quad/etc core processors. I don't know of any that specifically *require* them however.

[Roxoff]

No, none require more than one processor or more than one core. In fact the smp-enabled kernels will run on a single-processor system quite happily in most cases.

Just shows how much more versatile and flexible Linux will be than the new Vista - Linux can schedule any of its processes to run on any of its cpu units, thereby maximising its own flexibility.

[Juan Pablo]

How do you do that Roxoff?

[Roxoff]

I dont do it, I didn't write the kernel thread scheduler.

Or did you mean about running an smp kernel on single processor machines? Surely you've done that by accident before - I have (although not for an awful long time...)

[fingal]

Surely you've done that by accident before - I have (although not for an awful long time...)

Lol ... we're not all as clever as you mate. Personally I hardly do anything directly to my kernel, apart from the occasional modprobe. Just a simple desktop user you see.

I did once roll my own kernel on Slackware, thought I'd messed it up and it worked!

[Roxoff]

Lol ... we're not all as clever as you mate. Personally I hardly do anything directly to my kernel, apart from the occasional modprobe. Just a simple desktop user you see.

I did once roll my own kernel on Slackware, thought I'd messed it up and it worked!

Clever? Anyone can tick the boxes in the interface provided by 'make xconfig'... How do you think I managed to install the wrong kernel on occasions?

I eventually gave up building my own kernel. It's nice to be able to do it, and I'd still consider it for a commercial situation where performance is a premium. But for home use, I realised that if you took all the fractional-second performance gains that could be had betweek kernel releases and added them all together, it wouldn't be more than the time taken to configure and build the kernel in the first place. And when you've got 4 machines all running Linux, you've quadrupled your problem.

[apoorv_khurasia]

The last I knew abt dual/multi core processing was that the OS distributes the threads on the two "processors" so as to minimize sum of some function of the individual loads while avoiding "dependency issues". The idea that the OS will now distinguish between the threads as multimedia threads or a gaming thread sounds quite interesting but I was just wondering that what would be the possible benefits of such a task. I mean we are surely not dealing with special purpose processors, so why divide the tasks on the basis of "type"? As I view it this is primarily limiting as to how much processing power I can have for gaming or so....does not seem like a nice idea to me. Hence need expert insight in here

[Juan Pablo]

Roxoff: I mean, how can I send a process to a specific core?

[xxqq]

> Specific core.

Affinity:

http://www.cyberciti.biz/tips/settin...r-process.html

There is a strange non x86 gCore (grid on chip) processors from Boston Circuits that enable non smp ARC Linux kernel (Codito) to run non SMP kernel on multi cores ARC 760D CPU.

Will AMD/Intel create similar "time machine" module for their CPU?

http://arc-linux.org/ARC_linux

"Multicore Virtualization," takes the complexity of multi-core software and moves it into hardware. An on-chip "Time Machine" handles scheduling, resource allocation, and object synchronization dynamically, simplifying the software development process while minimizing latency and overhead.