Catch up on stories from the past week (and beyond) at the Slashdot story archive


Forgot your password?

Comment We are screwed. (Score 4, Insightful) 53

So, 3.6 Gb/s is cool and all; but I did a quick check and Verizon is calling 18GB/month the 'XXL' plan, so this appears to be largely an exercise in accruing overage fees even faster.

It seems like what will matter much more(unless somebody is planning to use the same tech for highly directional point-to-point wireless links, in which case raw speed is pretty useful); is how well these '5G' arrangements handle congestion; and how efficiently the amazing-fancy-theoretical-peak-throughput can be divided across a large number of users. Unless you are made of money, the problem with wireless data isn't so much how slow it is; but how costly it is(in part because of scarcity, which more efficient RF technology might actually alleviate, the 'because we can' part is a separate issue); and how it has a habit of just collapsing in a screaming heap under heavy load.

If the impressive peak bandwidth numbers indicate a larger pool of usable transmission capacity extracted from a given chunk of spectrum, fantastic, that is progress. If they simply represent what you could do if a single client used every doesn't-play-well-with-others trick in the book to get better speeds, that's utterly useless.

Comment Re:Still loaded with shovelware (Score 1) 78

You don't get the "Browbeat your rep" option; but I'm pretty sure that Dell will sell you Optiplex and Latitude systems in quantity 1, if you have a credit card. I think even Precisions and at least the more boring Poweredge stuff should be available as well.

You obviously don't have to go with Dell; but unless they've changed something recently; buying small quantities of business class machines should be no more difficult than buying consumer grade.

Comment Re:either integrated Intel HD Graphics 530 or a po (Score 1) 78

There have been several different flavors of Intel Integrated/Nvidia combinations on the market; with slightly different requirements and options depending on the details of how they are implemented.

My memory is a little fuzzy; but I think that the earliest implementations had actual 'video out' from both the IGP and the GPU, with switching silicon on the motherboard that sent one or the other to the LCD. Those offered the most visible control over which graphics device was in use(the one that wasn't was more or less fully shut down); but I think you had to at least log out, possibly reboot, to switch between them; that era definitely had BIOS options for permanently setting one or the other.

OEMs didn't like the cost of the added switching silicon, and users didn't like the clunkiness of switching between GPUs, so subsequent generations refined the process, with increasingly seamless cooperation(I think that the standard now has only the intel IGP connected to the LCD and any video outs; but the Nvidia GPU can write to its framebuffer if it is taking care of a given graphical task, so it isn't actually possible for the IGP to ever be fully idle, though the Nvidia GPU can be); but a corresponding increase in unhelpfulness if you are trying to force a configuration that non Optimus aware drivers can recognize and work with.

My Linux and BSD systems don't do much in the way of graphics, so I don't know what the current state of support is.

Comment Re:Betting we'll see thermal issues. (Score 1) 78

I have yet to hear any clear explanation for why Intel appears less than cooperative about the idea of Thunderbolt being used for GPU purposes. There have been a few, heavily integrated and close to model-specific, releases; but the "Here is a box with an x16(mechanical) PCIe slot inside, and a thunderbolt port" market is pretty slim, with the exception of some very, very, expensive cardcages from outfits like Magma, clearly aimed at audiences with expansion cards that make gamer toys look disposably cheap.

Most of the tinkering you see skips Thunderbolt entirely and uses the PCIe 1x->16x adapters that became popular when GPU cryptocurrency mining became a craze; and connect those either to the 1x PCIe lane provided by an Expresscard slot; or the one provided by a mini-PCIe slot.

Comment Re:Beware of Dell Support - Worst I've seen (Score 1) 78

What he told you was true. From a certain point of view: Dell's 'consumer' support has traditionally been somewhere between 'as empty and pitiless as the dark spaces between the stars' and 'actively insulting'; but they've always recognized the value of treating enterprise customers properly(and the warranties cost more, to compensate). There have been some ignoble incidents(their handling of Optiplex GX270 capacitor-plague failures was so egregious it resulted in litigation; ironically the IT guys at the law firm defending Dell were fighting to get their own GX270s replaced with ones that worked at the same time the lawyers were making the case that Dell's handling of the matter was just fine...); but in general their Poweredge, Optiplex, Latitude, and Vostro lines all have pretty decent support; and offer excellent support as an option if you are willing to pay for it.

The 'Inspiron' line, for home peons, has traditionally been pretty atrocious. XPS tacks somewhere between the two; it's a bit more annoying if you are trying to operate at scale(unlike the business/enterprise support guys, they tend not to let you do the "I've already run the diagnostics, here are the error codes, now send me a new whatever" thing); but unlike the low-end home user guys, they don't treat you like a filthy cost center who should fuck off and die.

Comment Re:Non-IPS panels (Score 2) 78

In fairness to Windows, non-integer multiple resizing simply isn't possible to do well unless all your graphics are vector(and even then, the designer's care and attention can have a strong influence on whether the result actually looks good to people at different scales; but at least there is a mathematically 'correct' answer).

If you have bitmap elements, integer-multiple resizing is both relatively trivial and possible to do 'correctly'. Non-integer multiple, like lossy compression, can be done in surprisingly non-annoying ways; but it cannot be done without some violence to the original. Bicubic interpolation will look a whole hell of a lot better than nearest-neighbor; but there simply is no 'correct' way of mapping N pixels into some non-integer multiple of N pixels.

Windows tends to work even less well than the ideal case would suggest; but even if you completely discard all issues of legacy widget sets, horrible retro UI designs, etc. and sit down with a bitmap image in photoshop, resizing it by a non-integer multiple is going to be a matter of compromise.

Comment Re:Non-IPS panels (Score 1) 78

'4k' is glorious; but unless you have truly impressive eyes, it's hard to justify on any laptop you'd be willing to carry. Going from 2560 x 1440 to 3840 X 2160 on the desktop was even better than I expected; but that was on a 27-28ish inch display; and if details were any smaller they'd be actively uncomfortable.

On a display of half the diagonal size, in a situation where GPU and battery power are at a premium, it just seems a tad excessive.

Comment Re:Input devices (Score 1) 78

It'd be beautiful if you could get aftermarket keyboards with a trackpoint added. Probably not possible for most models(I would have expected the falling cost of silicon to make embedding the controller into the keyboard FRU and being able to use a lower pin-count USB/serial/i2C/whatever connection to the motherboard; rather than leaving the keyboard passive and running all the lines from the switch matrix more common; but most laptop keyboard connectors continue to be matrix-type with the actual keyboard controller on the motherboard, so you couldn't just add the pointing device without cooperation from the motherboard, unlike what the situation would be if the connector were just a USB port with a nonstandard connector).

Between Lenovo attempting to bring their own, inferior, ideas to the Thinkpad line's design, and their fuckery with assorted terrible preloaded crapware; it's a lot harder to get excited about a new Thinkpad; but going without a trackpoint would hurt.

In fairness, though, Dell appears to have really upped their game on design of late. These models aren't even Latitudes, and they are genuinely nice; rather than merely endurable and attractively priced(though the price isn't bad). Tempting.

Comment Re:A perfect example of why tech is cyclical.... (Score 2) 86

For sufficiently latency-insensitive operations I don't think that it has every really gone away; but my impression(based on hazy memory and anecdote, though I'd welcome anyone with actual numbers) is that, unless you live in an atypically favored location, the delta between the storage you can afford and what the ISP will sell you, much less at a price you can stomach, has actually increased over time, thanks to HDDs massively upping their game while ISPs have improved; but rather more slowly(especially on upload).

Comment Re:Theft waiting to happen (Score 2) 86

Unless they really screw up the encryption(in which case the value of some of the data being transported might be worthwhile); these things seem like they'd actually be pretty dubious theft options. Even new, 50TB worth of consumer grade disk(I assume that Amazon is using some redundancy; but probably isn't splurging on SAS or fancy-enterprise-SATA for disks that will spend most of their life with Fedex, not actually spinning) isn't actually all that valuable(4TB drives are ~$150 retail, 6TB ~$250) and 'used' is not a happy word when trying to sell a hard drive. Plus it's essentially certain that Amazon has every serial number, MAC address, etc. of every component in the box on record, so you are SOL if anyone ever checks.

It also wouldn't be too surprising if the case has some level of active anti-theft reporting. Given that sub-$100 cellphones have GPS, a cell modem, one or more accelerometers, and are built on SoCs with enough GPIO to connect a bunch of tamper switches/sensors to; it wouldn't be particularly impractical for the box to report its location, integrity, orientation, and vibration levels every 30 seconds for the entire trip. Not impossible to defeat; but you'd need to nab it in an area of no service and silence it(by force or RF-blocking container) before moving it elsewhere.

I'd certainly encrypt my data carefully before consigning it to either the post or the internet; but I'd be surprised if hitting these boxes would be a good risk/reward for postal employees(though I know I'd like a look at what is inside, a nice rugged network attached storage module is likely to be a neat piece of gear).

Comment Re:Hmm... (Score 1) 34

I'd honestly be a lot more optimistic/sympathetic to this sort of research(psychology and psychiatry certainly need all the help they can get at actually being useful, and minimally-invasive neuron-scale intervention is a likely avenue of research, both for better data gathering and for possible treatment); if we hadn't just learned about the whole "Lets get some hack psychologists to design our torture program; and then subvert more or less all the relevant parts of the American Psychological Association, including its ethics director, just for giggles, in order to provide an appearance of legality!" program.

I'm not one to be scared by the 'ooh, it looks sci-fi and icky; it must be evil!' school of medical ethics. We are creatures made of meat, subject to a wide variety of more and less ghastly medical problems; and there is no use pretending that some of the solutions are not going to involve ugly, meaty, hacking. It's just that the DoD would need to clean a lot of house(an exercise that they haven't even shown much interest in pretending to undertake), before I'd be capable of pretending that having them seeking to do it is anything but really creepy.

Comment Re:Well if this is true (Score 1) 99

The problem with 'regulations' is that it suggests that the activity is fundamentally OK; just needs a few little safeguards to prevent specific abuses.

In this case, they really need to just be declared hostis humani generis, and their extirpation made the duty of any person or nation that has the occasion to undertake it. Proposals involving salting the earth, skull pyramids, and flaying are also to be considered.

Comment Re:Hubris and Self-Interest (Score 1) 281

If you actually have a problem that decomposes nicely into lots of little, neatly contained, problems it might work really well. It's just that if you have that, you are among the blessed and probably don't need any fancy consultants in order to do just fine. You'd need somebody who is actively capable of pulling defeat from the jaws of victory to screw it up.

The sticking point is whether or not snake oil can dissolve seemingly insoluble problems into lots of little, neatly contained, problems.

Comment Re:Not a hard and fast rule... (Score 3, Interesting) 281

I don't know how broadly it can be applied(if it in fact works as well as they claim at all); but it would appear that the whole point of these 'microservices' is to produce smaller 'projects' so that you have more room to scale before complexity eats you alive. It's not so much a disproof of the 'mythical man-month'; but an adaptation to cope with it.

Getting purely linear scaling without some sort of zero-latency hive mind is unlikely to be possible; but it seems fairly obvious that the amount of overhead you incur by adding 20 extra people to a five man project is going to be rather higher than adding a second person to a one man project(though the jump between 1 person and 2 people might actually be pretty big, if helpful in terms of producing documentation that somebody other than the 1 person understands). If you can break your projects down into smaller pieces, with complexity better contained, and well defined interaction between the pieces, you have teams small enough that you might actually be able to make them faster by making them somewhat larger.

If your project is already a screaming heap of interlocking complexity, there simply isn't as much work that can be done in parallel. Aside from people stepping on each other's toes, there will just be a lot of "Part X can't be done until the guy doing Part Y finishes".

Not so terribly different(if likely to be even less predictable because humans are involved) than deciding how a problem will scale if you throw more computers at it. If your problem is actually a large number of mostly unrelated problems, it'll scale nearly perfectly. If your problem consists of lots of somewhat interconnected problems it will scale; but demands on interconnect will become increasingly expensive. If it's a purely serial problem, and each step depends on the prior step, it may not scale at all.