Thanks All! This thread has just the kind of info I need and will help me make a short list.<br /><br />I plan to use Pure Data to do effects on my guitar (in particular a kind of live granulation patch I'm quite pleased with) which is why low latency would be good. At some stage I'll be recording that using about 4 tracks simultaneously so the lappy will need some power (plus in Windows I need to do some heavy lifting for work). I've got a Zalman laptop cooler which I'd recommend if anyone has overheating problems. Keeps it quieter as well. <br /><br />Cheers!<br /><br /><br /><br /> <p>On Sep 1, 2009, <strong>Kiernan Holland</strong> <rofthorax@gmail.com> wrote:</p> <div class="replyBody"> <blockquote class="email_quote" style="border-left: 2px solid #267fdb; margin: 0pt 0pt 0pt 1.8ex; padding-left: 1ex;"><br />My Dell Dimension 9150 has APIC, but only 23 interrupts.. I suppose with the 750i there will be more since it is quality hardware, but someone I talked to was unsure whether it would have good linux support.  Anyhow.. It seems to be that the best choice is not to get the Core 2 Duo's but the i7, as it supports hyper-threading, evidently according to the wikipedia article on the i7, the Core Duo doesn't support hyper-threading.. Maybe that's just the core duo and not the core 2 duo, nope the core 2 duo doesn't, says it is a descendent of the Pentium 3 architecture. The Pentium Extreme is the one I'm building, is overclockable to 4.2 Ghz.  Also said the Pentium Extreme has 31 stages in it's execution pipeline, but that the branch prediction penalty is greater with this many stages, but that the internal ALU's run twice as fast as the clock cycle to make up for it. All I know, from my discussions with my brother who's done work in things like fault tollerant systems, you need a high degree of branch prediction for linux because of the greater amount of indirection. I also read that part of the problem with the netburst is that it makes other operations go slower, and references to bit-shifting operations were given. If it has hardwired floating and integer operations, I don't know how much better it would be to have bit shifting operations since they are only going to improve multiplications and divisions by multiples of 2. I think a shared cache would be better than a divided cache, but it's a cheap chip, cost me 99 dollars at tiger direct. A Core 2 Duo would probably run me more and i7's that compare to the Pentium Extreme in having hyper threading, go from 200 to 1000 dollars. I'd rather get the best of the last line, than wait for the best of the next. <br /> <br /><br /><br />On Tue, Sep 1, 2009 at 3:45 PM, Kiernan Holland <span dir="ltr"><<a class=" parsedEmail" href="mailto:rofthorax@gmail.com" target="_blank">rofthorax@gmail.com</a>></span> wrote:<br /> <div class="gmail_quote"> <blockquote class="gmail_quote" style="border-left: 1px solid #cccccc; margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> <div class="im"><br /> <div class="gmail_quote"> <blockquote class="gmail_quote" style="border-left: 1px solid #cccccc; margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><br /> I would not compare this interrupts file with a more modern machine.  A<br /> lot has changed in the years since the P4, and for the better.  Netburst<br /> is dead, hurray.  The PIC has been replaced by the APIC.  If you care<br /> look them up in wikipedia, the point is that a modern machine will not<br /> look like the above, and this is not only to be expected, but is a good<br /> thing.<br /> <br /></blockquote> </div> </div> </blockquote> </div> <br /> <br /> <hr size="1" /> <br />-- <br />Ubuntu-Studio-users mailing list<br /><a class="parsedEmail" href="mailto:Ubuntu-Studio-users@lists.ubuntu.com" target="_blank">Ubuntu-Studio-users@lists.ubuntu.com</a><br />Modify settings or unsubscribe at: <a class="parsedLink" href="https://lists.ubuntu.com/mailman/listinfo/ubuntu-studio-users" target="_blank">https://lists.ubuntu.com/mailman/listinfo/ubuntu-studio-users</a><br /></blockquote> </div>