[cctalk] Old UNIX Systems

On 4 June 2013 22:46, Fred Cisin <cisin at xenosoft.com> wrote: The serious question is: why. If current trends continue, discrete GPUs are going away, integrated into the chipset. Motherboard logic is going away too - e.g. the latest "Haswell" Intel Core CPUs now incorporate the voltage regulators on the die as well. More and more is getting built into the CPU as a way to use the extra transistors that Moore's Law keeps delivering, until soon enough, there will be about 3 components to a PC: a CPU containing all the logic and some storage chips - some RAM and some Flash. The next step after that is to integrate those into the CPU die too, partly to shorten the buses and increase speed, partly to save cost and improve reliability. Or memristors will kick in and the difference between volatile and nonvolatile storage will disappear. Then you have one chip, basically, and some power and some cooling. There will be relatively slow external serial buses - some descendant of Intel Lightning, perhaps, quick enough for slow third-level storage, sound and video, but not for adding a different GPU. Once we get there, why would you want or need a big box any more? What would be the point? Frankly, today, only hobbyists & specialist workers change GPU or sound card - for most users, the built-in stuff is fine. Which is another part of the reason for the rise in notebook sales and the drop in desktop sales, because for nonspecialist users, notebooks are good enough now. The trend of increasing integration has continued for as long as just about anyone on this list has been alive, and I'm aware we have some septuagenarians. Why should it slow now? We already have 6-physical-core 12-logical-core chips with powerful onboard GPUs, memory controllers, bus controllers, VRMs and most of the rest of the computer on the die. There are server x86-64 CPUs with 16 physical cores. There are specialist MIPS chips with 64 cores. That trend is going to continue until your general-purpose PC is just as sealed and non-expandable as an iPad. All you'll be able to do is add slower third-level (I'm not counting caches here) storage. The big looming limit is when photolithography stops being able to make smaller features. Seriously, the stuff with "expansion slots" is a historical phase that we are now nearing the end of. You'd have to use it in the dark, that's the snag. You are aware that actual 3D displays of various forms are now on the mass market, aren't you? That you can buy 3D cameras with 3D LCD screens on the back, and pocket games consoles with 3D displays, and 3D monitors and TVs? Not ones that need goggles - although of course those exist too - but open-space 3D displays. Then there are fog screens and helical-scan screens and so on. Lots of 3D display tech out there for 3-5y now. Yes including stuff that you can walk around the back of and see the rear of the object. As tricky as playing a theremin, I imagine. :?) But live instant visual feedback would help. There are lots of solutions for the problems of the size of big squishy ol' human fingers, from zoomable displays, to display regions, to styli. That's not an issue either. And there are mice that you can use on Wacom-type tablets, as well. For anyone who thinks that you can't possibly draw with a big fat finger on even a tiny 3?" multitouch display, have a look at http://www.iphoneart.com/ and see some really impressive stuff. You chaps are all throwing up objections that were solved years ago, in some cases decades ago. I really don't get it. -- Liam Proven ? Profile: http://lproven.livejournal.com/profile Email: lproven at cix.co.uk ? GMail/G+/Twitter/Flickr/Facebook: lproven MSN: lproven at hotmail.com ? Skype/AIM/Yahoo/LinkedIn: liamproven Tel: +44 20-8685-0498 ? Cell: +44 7939-087884