Follow Slashdot blog updates by subscribing to our blog RSS feed


Forgot your password?

Comment Umm, I hope that translation is to blame. (Score 1) 34 34

I really hope that "proud to declare that we are at the cusp of a reclaiming our heritage of being connected to each other and connected to the world." made a lot more sense before some translator mangled it; because otherwise it seems like absurd nonsense. If people were connected long enough and far enough back in time for it to count as 'heritage', the technology behind those connections must have been comparatively primitive. Is he saying that communications have regressed since that time? What golden age of connectedness is he talking about?

Comment Re:Why not both? (Score 1) 231 231

I didn't make it suitably clear; but the 'complexity' is really more of a historical issue. The fact that you can get power transistors, digital logic, and similar solid-state goodness for peanuts, possibly even less than the carbon brushes or other electromechanical alternatives, is a comparatively recent thing in historical terms.

Now that you can, doing so is pretty compelling for any but the highest-power tasks; but it has not always been the case that you can throw semiconductors at a problem for astonishingly tiny amounts of money. Today it is; but a lot of very clever electromechanical, inductive, and similar tricks were developed during the time that it was not.

Comment Re:Low cost chip, high cost support (Score 3, Interesting) 91 91

What I find a bit weird about SPARC's near-total obscurity is that(please correct me if I'm wrong on the details; but to the best of my understanding from what I've read) the ISA is available for use on a royalty-free basis, and there are even a few BSD or GPL verilog implementations out there. That's even less encumbered than MIPS(which has some patents that the owners like to wave around on a couple of useful instructions).

My naive expectation would have been that SPARC on such liberal terms would show up a bit more often embedded in various chips that need some sort of CPU to do housekeeping, as the ISA of security and/or nationalism driven 'indigenous technology' efforts, and potentially even as the cheaper-than-ARM option for application processors.

Clearly that hasn't actually happened, and it's mostly ARM in SoCs and application processors(with PPC holding out in certain automotive and networking niches for some reason; and MIPS in router SoCs and the occasional Chinese vanity project); so ARM's license fees must just not sting that much.

Building SPARC parts that go toe to toe with Xeons would obviously be a much more ambitious project(as well as an act of directly fucking with Intel's juciest margins, which they probably won't take very kindly); but I am surprised by the fact that SPARC is so rare among the zillions of devices that have no need for x86 compatibility and are mostly about delivering performance in the gap between beefy microcontrollers and weak desktops for as little money as possible.

Comment Re:Disappointed (Score 2) 231 231

The competition for good DC-DC conversion is reasonably fierce(given the existence of DC telco and datacenter operations, and the fact that even 'AC' shops are really just doing the conversion in each chassis(and unlike the old AT PSU days, an ever larger chunk of the output power is 12v going directly to a DC-DC converter on the motherboard to feed the CPU and RAM, with fewer and fewer components, aside from HDD motors, being sufficiently high voltage to feed directly from the PSU); so even modest improvements in DC-DC efficiency would make you quite wealthy indeed.

I haven't kept a close eye; but I think that the present standard for DC-DC modules still uses a number of off-chip components(whether because the needed capacitance and such simply can't be done in silicon, or are cheaper as discretes, I don't know); but you can get some very, very, dense little modules.

Comment Re:Why not both? (Score 1) 231 231

Especially if you are dealing with solar-derived DC, which is going to be (reasonably) steady. For some purposes, it doesn't matter that much whether or not the voltage actually crosses zero; but the ripple is essential if you want the magnetics to work correctly. If you have to go to the trouble of switching the DC to get that working, odds are that it's time for a pure DC design.

Comment Re:Why not both? (Score 1) 231 231

There are good reasons to be skeptical of complexity(such as basically all software); but at least for motors small enough that solid state switching isn't heroically expensive and/or practically disposable, 'complex' and 'digital' also mean 'has convenient ways to monitor and control the motor's activity in some detail'. Simple is good; but when the complexity is in solid-state parts you get a serious discount and sometimes some cool features.

Comment Re:Why not both? (Score 1) 231 231

Certainly possible; but probably only efficient or sufficiently cheap for some items.

My (admittedly limited) understanding is that motors are one of the parts where you can get a relatively refined and mature design for almost any remotely normal flavor of electrical input; but the design of the motor is going to reflect your choice, and work either inefficiently or not at all from some other input. Since AC units are basically all motor(some fans, a compressor, maybe a pump), that is a problem.

Some devices just don't care very much(resistive heating works for basically anything except voltages too much lower than the system was designed for); and others(pretty much all modern electronics) always use DC internally, so you can re-use almost the entire design; but need to choose the correct PSU. Some hardware makes swapping one after the fact really obnoxious; but at very least it's an easy thing to choose at assembly time.

Comment Honest question. (Score 1) 91 91

Can someone explain why the program handling interaction with assorted media files would be so closely linked to the rest of the system working? I understand that parsing the ghastly mess of different standard and pseudo-standard formats out there, as poorly or even maliciously interpreted by various 3rd parties, is a difficult and dangerous task; so I'm not surprised by the fact that there is a bug in the media component; but if it is known to do such a dangerous job why isn't it compartmentalized more aggressively? Why does losing the mediaserver process make a mess of the phone, rather than just causing it to mark the file that killed it as tainted, restart the process, and carry on?

Comment Re:Old news is so exciting (Score 1) 80 80

The article named the phone as the Motorola C123. Apparently that model has an atypically well-understood baseband, which is probably why it was picked; but that handset is dumb as a rock except by comparison to the utter antiques from the age of analog cellular or something. I don't even think it has one of the teeny little JREs that phones used to have.

Comment Re:Wrong question. (Score 1) 351 351

I think that it works both ways: the campaign gets face time and spending money from assorted big names in tech because of the hope that it will make programmers cheaper; but it gets buy-in from educators and parents and politicians looking for feel-good photo ops because of the hope that somehow every kid can be a well paid knowledge worker.

Compare to H1-Bs. Those are similarly favored as a way to drive labor costs down; but are more or less politically toxic; so they have none of the popular chatter. The major tech employers are in favor of both; but only one has the buzz in the other direction as well.

Time-sharing is the junk-mail part of the computer business. -- H.R.J. Grosch (attributed)