Accumulation.

[dsrtao]

First: historical cost of computing, adjusted for GDP:
http://www.flickr.com/photos/pkedrosky/468721697/
(from Paul Kedrosky at http://feeds.feedburner.com/~r/InfectiousGreed/~3/111094704/progress_of_com.html

Second: the government knows all your prescriptions: http://nielsenhayden.com/makinglight/archives/008872.html

Third: Michael Dell uses Linux ( http://www.dell.com/content/topics/global.aspx/corp/biographies/en/msd_computers?c=us&l=en&s=corp ) and Dell will make Windows XP available again on selected machines, ( http://www.pcmag.com/article2/0,1895,2116076,00.asp ) because nobody likes Vista, especially if they can't afford a $3000 desktop. In related news, a $650 computer can be quite a screamer (dual core AMD 3800+, 1 GB RAM, 250 GB RAID-1 disks) but the price goes up by $370 if you need Microsoft Vista and Office...

Fourth: Macs are easy to use, and a low-end MacBook with a gig of RAM is sturdy, well-designed and performs decently, all at a price point not terribly dissimilar to a brand-name PC notebook. And the OS is much more stable than Windows anything... Just saying.

Comments

[cellio]

I've heard tell of a way to run Windows software on a Mac without rebooting (I'm not talking about the dual-boot thing). Can you clue me in? Since I still have a few Windows-only apps that I care about, this is relevant to me. (I do not want Vista on my next machine.)

The price difference for desktops was rather significant when last I looked (factor of two?), but maybe that will change. I should have some time before I need to replace my current machine. (No, not interested in laptop + dock.)

[dsrtao]

The options are:
http://www.parallels.com/ (works, available now)
http://www.vmware.com/macos (probably works, about to beta)

[metahacker.livejournal.com]

And if you must have a tablet-style, the ModBook really looks rather well done.

[metahacker.livejournal.com]

I take a bit of issue with the usefulness of the graph to the lay-person, because it measures MIPS or MOPS or something like that. I can't remember the last time that was relevant...maybe 1995 or something. Nowadays few people complain their computer is too slow because it's too slow; end users are "good enough" on processor, modulo writing software that is more and more complex to do the same thing. It's almost like you want a graph that maps the worst of {processing power, hard drive speed, network latency} vis-a-vis a 'standard' application like word processing, telephony, or blogging...

[dsrtao]

Gamers, of course. Anyone who's work includes the phrase "high performance" and/or "visualization". Finance people are fond of large datasets.

For many workloads, I/O is a narrower bottleneck than CPU. x86 and its extensions are just not well designed, having evolved in the classic "trowel on another glop" methodology -- hence the years and years in which SPARC processors handled multitasking loads far better than the comparable-MHz x86.

For any gain in affordable computation, there are N things which are now trivial to accomplish, M things which are reasonable, and Q things which now look hard instead of fiendishly difficult. Consider ray-traced rendered graphics...

[metahacker.livejournal.com]

Exception, though a large one. Exception, technical field. Exception, specialized market. All valid markets, and yeah, I belong to those groups, but it's the 90/10 argument all over; those people are willing to pay a significant premium for their hardware. But summarizing all of computing into a single line/scatter involves averaging, and on average, computers have been 'fast enough' for 'everyone' for at least a decade.

(Why are we telling each other things we already know, and that we know we know? Are we trying to prove to each other how smart we are? I'm pretty sure we know. Maybe it's just reflex, or reflexive.)

I/O's a good point. Someone was objecting to how long it took to wake from hibernate until I pointed out that that was about the time it took to read 2GB of RAM off disk. (Of course one could argue that smarter demand-driven reads could improve that time...)

Ray-tracing is an interesting evolution to me. Gaming did an end-run around the ray-tracers and other mathematically-exact rendering folks (howdy) to get stuff that worked. Now that computation is making it tractable, RT in RT is moving back into vogue. Very interesting. Some twenty years ago I claimed I wouldn't be impressed until I could move a reflective/refractive crystal ball around in a lit virtual room in real time and have the image update appropriately; I'm pretty close to impressed with modern computing.

[cellio]

Thanks!