Wandering Thoughts archives

A peculiarity of hardware at universities

One of the peculiar things you find at universities (and probably almost nowhere else) is expensive hardware without any sort of maintenance contract. This is usually not the result of insanity combined with cheapness, but instead the result of companies donating hardware to researchers.

Computer makers love to donate hardware, because the cost to them is just the manufacturing cost (at most), but they get credit (probably including tax writeoffs) for the full list price, even if no one is actually buying the hardware at that price. But throwing in maintenance would cost them actual real money, so you almost never get that included in the deal.

(The cost to the company can be less than the manufacturing cost if the company is sitting on slow-moving excess inventory.)

Now, in theory researchers could then buy maintenance with their own money. In theory. In practice, maintenance on this sort of donation is quite expensive and few researchers have that kind of spare cash floating around. And no researcher or university can afford to turn down a large hardware donation without an ironclad reason.

(In practice universities will go to pretty great lengths to accept donations, even at not insubstantial costs to themselves for things like site preparation.)

So researchers and universities accept the hardware, plan to make do, and hope nothing breaks. Then when things do break, we get to fish around deep in the depths of expensive hardware and try to kludge things together using anything that works. This can be horrifying to ordinary rational people, who don't understand why the graduate student is using twist-ties and eBay'd parts to patch up a million dollar IMoon machine instead of calling in the professionals.

(Such machines also often lack real system administrators and/or real backups, because there wasn't money for them either.)

The really irritating thing about voicemail

The really irritating thing about voicemail is that unlike email, there's no random access to voicemail messages. Unfortunately this is really because it's being presented to you in audio, which is a fundamentally linear medium.

(Audio menus are a not very successful attempt to get around this.)

This means that the real solution is to have your voicemail emailed to you, or put up on a website, or in some other way available through your computer, where you can use all sorts of good random access technology on it. Ideally you would have your phone hooked into this system, so you could use the phone for playback and the computer for control.

(Many years ago I saw a Northern Telecom prototype office thing that integrated a computer with a phone in much this manner, courtesy of knowing someone who worked there. Even back then it was stunningly compelling, although presumably something sunk it since I've never heard of it since then.)

With Asterisk on the right hardware and some work, you could presumably build this today for free. Using VoIP phones would make it even easier, since they're closer to merging computer audio and conventional phone audio.

(I like phones for this because they are a good way of avoiding the speakerphone effect you would otherwise get from people using computer speakers to listen to their voicemail. While you could get the same effect by persuading people to use headphones instead, plugging in and putting on headphones is a lot more hassle (and thus more likely to be skipped) than just picking up the phone.)

A thought on iTunes and similar online services

What iTunes and similar online services are really selling is convenience and guilt-avoidance, with a small side order of helping out the artists. They're manifestly not selling the music itself (or videos or etc), since you can find them online yourself if you look hard enough.

The 'helping out the artists' bit is currently only a small side order because most everyone understands at some level how little of the money involved actually makes it to the artists.

The guilt-avoidance is not a good long term thing to base your business on, because there's an increasing amount of evidence that the current generation of teenagers feels very little guilt about this stuff to start with. (Partly, I like to suspect, because very few people feel very much guilt about not giving money to visibly rapacious record companies.)

On a legal front, the highest payoff is from closing down non-authorized services that make it convenient to find this stuff on the Internet, because they are your primary competition in the long term.

(Not only is suing individual downloaders time-consuming for a low payoff, but young people famously feel that they're immortal and that the odds don't apply to them, hence I think it has a low deterrent value.)

Odious DRM is thus cold-bloodedly counterproductive because it reduces the convenience factor.

(And iTunes et al can be viewed as a grand experiment in how much pain consumers will put up with before you drive them into the arms of more obscure, non-authorized sources. The early music company online music store attempts have already given us some data points.)

I hate hardware (AMD CPU edition)

I have been trying to spec out a new machine lately, which is reminding me all over again how I hate hardware. This time around, the target of my particular hate is AMD CPUs, especially the new AM2 ones, where the performance picture has become so complicated that you need a large chart to understand it.

Choosing CPUs used to be simple: within a given CPU family, the only thing that changed performance was the clock speed, so you could just buy the fastest CPU your budget and desires afforded and be done with it.

Athlons are no longer like that. Within the Athlon 64 X2 AM2 family, there are now three variables: clock speed, L2 cache size per core, and achievable main memory speed (the clock multiplier, as explained by AnandTech). Models with increasing nominal clock speeds zig-zag in the other attributes, to the point where I had to consult the large Wikipedia page of Athlon 64 processors to keep things straight.

(Thank god for Wikipedia. Good luck finding AMD discussing this anywhere you can conveniently find it; I'm not sure they even have a comparison chart of L2 cache sizes on their website.)

Then, once I'd worked all this out, it turns out that the supply of 1MB L2 parts seems to have dried up around here; local computer shops can't even get the Socket 939 versions with 1MB L2 caches, much less the AM2 ones. (Rumour has it that AMD has starved the distributor pipeline in favour of redirecting most of the supply to certain large computer vendors.)

I could try to view the 1MB L2 part drought as a way of simplifying my life, but instead it just irritates me that I can't spec the CPUs I really want.

(I care about the cache size and main memory speed because I tend to think that they dominate performance for the kind of CPU-intensive things I'm likely to do with my machines. Not that I've actually measured this to find out for sure, which makes me some sort of fool.)

How fast an LCD refresh rate is going to be fast enough?

One of my small peculiarities is that I'm an old-fashioned CRT person. By now this is an unfortunate affliction since it's pretty clear that LCDs are the future, whether or not I really like them. (If the prices for good CRTs haven't already bottomed out and started climbing due to decreased production, it probably won't be long.)

Part of the reason I haven't warmed to LCDs is that they have scary and visible failure modes, like stuck pixels, that the manufacturers blandly claim are not bad enough to replace the unit (I beg to disagree). But a good part of it is that for me, LCDs have almost always flickered.

(Okay, technically LCD flickering is actually ghosting.)

I admit that I'm a worst case for LCD designers, since I use black text on a white background. But for years, LCDs had achingly bad flicker issues whenever I tried to use them, sufficiently bad that I was amazed my co-workers could put up with it. I did once find an LCD that seemed OK on first inspection; unfortunately it was a very expensive SGI widescreen unit.

I've recently been spending a chunk of time in front of an almost current generation LCD, in this case a BENQ FP931. If Google can be believed, this has a 16 ms response time. Unfortunately, it still flickers. Not as much as LCDs did in the old days, but it's still there. (Possibly I am using it in an unfavorable situation, since there is indirect sunlight at my back, but I still expect my monitors to be acceptable there.)

The current generation of fast response time LCDs seem to be 8 ms, with some 'ultra-fast' models pushing to 6 ms. I'm hoping that that's fast enough, partly because I suspect that most people are not as picky as I am, so there won't be much demand for even faster LCDs that will drive their prices down (or even cause very many of them to be made to start with).

(There seem to be some faster LCDs; for example, the BENQ FP93G X claims to be 6ms black/white with a 2ms 'grey to grey' time. Of course, part of the whole fun is that all these are marketing numbers, so I'm never sure exactly what they're going to mean in real life and some source suggest that they're outright misleading and prone to be cooked; one discussion of this is here.)