OpenCL on the Rise

September 30, 2009

Looks like hardware makers are remembering what they are selling: hardware. You’d think they always knew that, but apparently they didn’t.

Hardware makers always liked to control their customers, be it end-users or developers. They make proprietary ports and closed interfaces that only work with their stuff. The thought process is something along “If we make sure our hardware is incompatible with whatever runs on our competitor’s hardware, then customers will come and be forced to stay loyal to us.”. Unfortunately these kinds of locked-in devices never really catch on for some reason, whereas open platforms (given they have some good traits) thrive immensely. Try to get absolute power over your customers, and at some point – no matter how great your product is – they’ll get fed up and leave.

What’s mind-boggling is how recent history is littered with failed attempts at binding hardware users. Since it’s closest to what I want to write about, let’s look at PhysX. First it was an attempt to sell physics acceleration hardware – you’d get great effects if you bought a PhysX card and a game that supported it, but without the hardware it was just another bad physics engine. Somehow game developers weren’t as enthusiastic as they were supposed to be. Then along came nVidia and ported PhysX to their GPUs. Now that a major GPU manufacturer is behind this, every game should use the engine, right? Well, somehow not every gamer trashed their Radeon to switch to nVidia and not every developer immediately wanted to make a game that would only run (well) on half the systems out there.

Because as a developer, you want a broad user base. It should be obvious, but somehow it isn’t. Why do MMOs look like they are lagging behind in graphics quality like 3-4 years? Because they want to reach as many users (with old computers) as possible. Imposing any artificial limits on your hardware makes it worse hardware, no matter what your strategy was.

But, almost a year after the release of the OpenCL specification, OpenCL ports of physics engines are being announced – by Intel (Havox), AMD (Bullet) and nVidia (PhysX). So as a developer, you can finally choose an engine and go with it – it should work on all platforms that support OpenCL!


Custom Firmware

September 25, 2009

In the previous post I ranted about poorly designed gadgets and the lack of openness screwing up any chance at user-driven software improvements. But I should have pointed out that while few companies get what community involvement could mean to them, some users take the matter into their own hands.

Custom firmware projects exist for many types of devices. Tomato and dd-wrt are popular, linux-based firmwares for dsl routers – I’ve been using tomato for quite a while now. And while there are routers that support custom firmware openly (like the legendary Linksys WRT56G series), others can be hacked in order to overwrite the crap they are running out of the box.

Another popular category of moddable devices is mp3-players. Usually it starts with some linux geek asking the age-old question: “…but does it run linux?”. He then starts hacking away, sometimes bricking several players while reverse-engineering the way firmware gets uploaded and how it interacts with the device. Since it is much too expensive to develop a new ASIC for a cheap mp3 player (or almost any device that doesn’t require extremely efficient computing), the hacker is usually able to find out what known components it is made of and use the (known) specifications to learn more about how it works. Once hobby developers have access to the device – usually after one keen individual paved the way – they’ll fantasize about what functions they always wanted their gadget to have, what functions it could have (not necessarily useful but fun), and how they would do it so much better than the people who build it. Then comes the next wave of developers who just want to make minor modifications (like a new theme or a little app or just some experiment) but won’t meddle with the low-level or even mid-level stuff. Before you know it, you have a thriving community of programmers (and sometimes artists) who will spend a lot of time improving the software and porting it to more devices.

Rockbox is such a project, and I’ll install it as soon as the port for my player is stable.

Crappy Gadgets

September 22, 2009

It’s unbelievable how bad the average gadget is. Everything that does a little more than save data in flash memory is most likely flawed, sometimes to the point of being barely usable.

My MP3 player just died. I haven’t used it for a few months (but it wasn’t very old… maybe two years) ever since I bought a Sony W890i cellphone. But the (proprietary) connector of the phone broke – the phone side, so buying the a cable wouldn’t help – so I started using the old MP3 player again. It was fine for some time, until it stopped booting all the way up. Too bad, other than for the crampy 2GB space it was a great device, with good sound and an uncomplicated UI.

So I got a new one. I didn’t spend much time looking because I figured some player with 8GB and good reviews would be sufficient. I would have gotten an ipod, but it’s twice as expensive as other devices with similar capabilities yet I don’t get to transfer data without itunes. What I got was a player with mostly good reviews and the ability to play some formats other than mp3, like ogg and flac. It also plays videos and has a radio, but I don’t really care that much about these functions. Oh, and you can use a microSD card to get even more space.

Sounds great, doesn’t it? Well, so far it works good enough – I’m not returning it (yet?). But I’m still disappointed. When I wanted to try out video playback (other then the one music video that came with the player), nothing worked. Okay, so ipod/psp-compatible mp4 won’t work – let’s feed those videos to the media converter that is supposed to squeeze them into a compatible format. The converter failed to load the videos. So I started looking on the internet and what do you know – it’s a known problem. The company doesn’t seem to care, because apparently it’s enough being able to play one video in order to advertise “plays back videos” and the converter software is mostly an alibi. Nobody really knows what kind of videos the player or the software accepts other than it’s AVI with some weird restrictions.

It’s obvious that this device was only made to be “good enough”, and only for its main function. Watching videos on such a tiny screen isn’t really something most users will do, but it probably helps sales – so they crammed it in and made it barely work. There were other issues as well – the voice recording function made the system freeze up. Obviously that’s another function they thought nobody would use. It worked better after a firmware upgrade, but it’s becoming clearer how this device could be so cheap.

What’s really sad is that few people seem to care (hence the good reviews). When I buy something, I expect it to work almost flawlessly. If there is some bug, it should better be fixed with the next firmware update. I know, there will always be some bugs left. But if you manage to find several the first day you are using it, it’s clear that the manufacturer does not really give a shit about testing. Interface design is also pretty horrible anywhere but in Apple products (where you pay a hefty price for it). And it’s not like there aren’t people out there who wouldn’t be glad to help! Just release the firmware as open source (you are not really selling the software anyway – the hardware is much more important than the crap it is made to run) so more competent programmers can take over. Or don’t even do that, just release the specs. There won’t be a good replacement for a while – open source projects usually don’t have (strict) deadlines – but ultimately the free version will come closer to perfection than the proprietary one. Because once you have sold a piece or hardware (or worse, have stopped producing it), there is no incentive for you to keep supporting it as a manufacturer. It’s sad, but that’s how it is. The “free market” is not going to solve this problem, because even brand-awareness doesn’t seem to impact the long-term memory of customers. My only hope is that some day there will be enough tech-savvy people to make free software an important factor even in gadgets. I know there are some attempts to make gadgets with open specifications, running open-source software, but right now this usually means expensive hardware and bulky, ugly design for some reason. I hope this will change soon.

Life-Long Movies

September 21, 2009

So here’s a little Sci-Fi scenario:

It’s 2050 and the singularity happened. Unenhanced humans still exist, but the earth is becoming more like a “nature reserve” controlled by AIs. Virtual humans are colonizing the solar system – with most of them having long forgotten about the state of the real world. Many choose to work in science and focus on exploration and even more technological development – thus staying informed about the current state of affairs – but the rest tries to create their own worlds inside what was once the internet and is now very much reality for any virtual being. External (real) time isn’t much of a concern anymore if you are a conservative human who resists too big alterations to your neural pattern. With your decision to go purely virtual you might have been one the “bleeding edge” like a decade ago, but nowadays the progressives tend to enhance themselves in ways that make you shiver. And you do, because you also chose to have a “retro” virtual body. Inside “realistic” simulations it looks and feels like a very healthy physical human body. During the last few days (in the real world) it has become overwhelmingly clear that people like you are being kept in another reserve, just like the primitive human on real earth. The progressives have once more decided that they are so much more intelligent than you that you having any unpredictable impact on the world would be unacceptable and thus stuck you in a virtual cage. You get to do almost everything you did before and you continue do benefit from advances in technology, but you don’t get to make any decisions in the real world. If you insist, you’ll probably go into a simulation where it looks like you have power. If you find out the truth, they’ll find a way to change your mind quite literally – although they won’t touch it unless you are really really desperate.

But like most people, you gave in to that higher power. You also get some satisfaction knowing that a good portion of the progressives won’t keep up with the rest and end up in a higher-level reserve. Life isn’t bad – it’s just hard to cope with eternal life when your mind isn’t that all that different from a traditional, physical human. There’s a new form of entertainment that was once all the rage with progressives and has now been handed down like an old toy to a younger sibling. It’s the concept of experiencing entire lifetimes on old earth – completely simulated. It’s not like the usual full-immersion movies or games that take your mind on perfect journeys. It’s a very serious simulation of a life on earth, some time before the singularity. It’s very detailed, very realistic and it uses a good portion of the computation power allocated to your reserve. There are several subscription models that give you limited control over your destiny in the simulation. But all have this in common: you forfeit your memories at the start, your life starts a short time before birth, you get kicked out of the simulation when you die and that’s also the only way to get your old memories back. You experience a life like “modern” humans did for hundreds of thousands of years, only with the privilege of being secretly immortal – the downside being that you don’t really know that during that time.

I know, it sounds a little like The Matrix… but at least there are no “human batteries”.

Invisible Progress I

September 10, 2009

I wonder how mathematicians would feel if a theologian started criticizing their proofs. Since he doesn’t know much more about mathematics than the average religious person – but to whom he is the more trustworthy authority – they might get pretty mad. Even more so considering that he is able to talk in a way that sounds reasonable to most people. I’d still listen to the mathematicians because they are much more likely to know what they are doing.

Now what if a professor of philosophy and history started ranting about a complex, controversial theory spanning multiple branches of science which only in some of its implications for the future touches philosophy? You’d get this.

I know, this is turning into a ad-hominem attack. Just because this person has no certified knowledge of the subject doesn’t mean he isn’t right. Just like you can’t rule out your garbage-man making a good argument concerning hidden variables in quantum physics (especially if he’s like the one from Dilbert), a professor of philosophy could have a good reason to think the singularity is bullshit. But: he got into IEEE Spectrum whereas I doubt that the average forum idiot who argues that 0.999… != 1 would have a chance. Why? Because that is an easy subject, and even if you fail to understand the proofs or why a particular counter-proof is invalid you can be sure that mathematicians are the authority on this subject. With the singularity – not so easy. It concerns a wide spectrum of developments, makes assumptions about what drives these developments, shows a counter-intuitive view of the present and the future and probes into “forbidden” areas like human consciousness. That’s why it’s already very controversial, and why representatives of individual fields feel like they need to comment on it. There is no authority on the subject since there tend to be very few multi-disciplinary scientists in today’s specialized world. So we’ll just listen to anybody with a degree in anything.

Okay, so much for the irrational ranting. Let’s see what the main arguments of the article are:

  1. The technological singularity is too weird / too good to be true and I don’t like the way it is presented. Let’s keep in mind that it is obviously nonsense, shifting the focus to the people supporting it.
  2. Technological progress has not increased. It is even slower than ever, because I have not seen ground-breaking inventions like the car, the plane, electricity, the computer or manned spaceflight in the last few decades. This is how we should measure progress: number of inventions that seem amazing to the average person, have easily understandable applications and do not look too much alike.
  3. Singularitarians extrapolate Moore’s law indefinitely, even though there are physical limits.
  4. Single aspects like communicating directly from brain to brain at rates faster than current speed of thought are infeasible with today’s technology and knowledge.
  5. No (concerning exponential progress in the medical field).
  6. Advances in medical analysis will only bring us more information about diseases we can’t cure.
  7. Positive technological progress is positive and therefore improbable.
  8. No scientist can properly investigate the claims someone made in another field, therefore people are pushing cross-disciplinary phenomena where they can easily add their bad research without anyone being any wiser. This network of misplaced trust is inherently flawed and doomed to fail.

It is sad to see that even “certified intellectuals” resort to mindlessly throwing around concepts just to support their view. The idea of a technological singularity is much older than that article, so there has been some significant amount of discussions already. Most of the presented arguments have been countered a long time ago, so it would have been much more interesting to read responses to the counter-arguments. And the rest is rambling and ranting, presented as arguments.

So apparently all I have to do is mindlessly repeat the answers to those arguments, which is not particularly challenging.

  1. During the past 200 years the present always turned out to be the unbelievable future to people from the past. How a particular prediction makes you feel is not a rational argument and will have absolutely no credibility with supporters or the (rational) undecided.
  2. The reasoning behind why technological progress is accelerating is complex and difficult to prove or refute. I am not going to do that in once sentence and you shouldn’t be using a little anecdote about how you get the impression that today’s inventions are boring and insignificant. However, there are books on the subject (with traceable evidence) and those must be dealt with before dismissing the entire notion.
  3. Nobody is saying that silicon structures will shrink indefinitely. Singularitarians argue that every time one technology reaches its limits, a new one takes over. Like the time vacuum tubes were replaced by transistors. The current paradigm still has a few generations to go and there is a whole bunch of new ones being developed.
  4. Back when computers were used primarily for simple calculations and highly specific tasks nobody would have seen the usefulness of a personal computer capable of then unimaginable numbers of calculations per second. But the price/performance sank drastically and along came the applications. Never make an argument from embedding a future technology in the present world – it just does not make any sense. Especially when dealing with information and communication technologies.
  5. Not really much of an argument.
  6. Diagnostic capabilities will always stay ahead of curative capabilities. You can’t cure what you don’t know. I can see the appeal of not knowing you have an incurable disease until you die, but what exactly is the argument here? That we should stop diagnosing diseases? That we have stopped finding cures for diseases? Whatever the argument is, it begs further evidence (and explanation).
  7. This is more of a recurring theme in the article than an actual argument. I simply don’t know what to do with pure pessimism. “There, there”?
  8. This is interesting, but a very bold claim. Especially because there hasn’t been a real polymath for a long, long time and the argument says that scientists from different fields can only effectively work together (without “just” trusting each other) if they are experts in all fields involved – or at least know enough to spot errors. Since most scientists are experts and many cross-disciplinary projects exist and work even though peer-reviewing is limited to peers (that is – people from your own field of study) – doesn’t that mean that trust-system is good enough? I don’t think scientists blindly trust other scientists. They trust the peer-reviewing system, not individuals. Anyway, very bold claim to make without presenting any evidence whatsoever.

I’d like to dub argument #2 “The Invisible Progress Effect” and study it a little more before making another post about this topic.

Defying Defying Gravity

September 6, 2009

In an earlier post I compared the Sci-Fi-Series Defying Gravity and Virtuality to each other. In short, while there were lots of similarities, I really liked Virtuality better. The pilot make me curious about what comes next (we’ll never know). With Defying Gravity, it’s more a train-wreck type situation. I can’t look away. What makes it especially gruesome is that I really love science fiction! But this series is definitely not about science (although it features a group of scientists on a scientific mission), and the fiction part is pretty much “What would happen if we put a group of juvenile, self-absorbed and increasingly psychotic idiots in a spaceship? Oh, and there is some kind of brain-controlling alien.”, with no actual depth whatsoever. It’s almost funny seeing the “immature nerd” character get scolded for acting like a child – by characters who by no means should be the ones casting the first stone.

Now the latest episode showed almost all the astronauts being immobilized by alien-induced hallucinations. It was (probably unintentionally) hilarious. And not just that, it took some time away from the stupid drama! If this persists in the next episodes, I’ll finally have an excuse for watching this show other than to have something to criticize or to act smug about. Who knows, maybe from now on everybody just runs around the ship, seeing completely different things than the other crew-members, making the spaceship look like an interplanetary loony bin. Now that’s entertainment.

That Rejuvenation Guy with the Beard

September 4, 2009

Aubrey de Grey is an interesting person with some interesting views. What he has to say about evolution and aging is a very important piece of knowledge, for anybody interested in either topic. But every time I see the guy get interviewed, one question pops up: “Why do you look so weird?”. And it always tells me more about the interviewer than the interviewee. A setting where this question seems important enough to be asked is probably not very serious. The other questions/arguments will be along the lines of “But everybody has to die at some point!”, “If longevity was that great, we’d have evolved it by now!” and “So, how much mercury are you using?”. So let’s assume the reason for choosing this particular style is not something trivial like “Fan of a Russian mathematician or 19th century style” or “Doesn’t like shaving or getting a haircut”. What if the reason (or one of the reasons) is being able to quickly judge the person you are talking to? Most people who are really interested in the science behind his project and who are willing to question the inevitability of death will ask many, many questions before (if ever) running out of more interesting topics than beard length. I’m not saying that that’s true every time, but I’m sure a strong correlation exists.

In order to keep this post consistently irrelevant, let’s also speculate about why de Grey seems to be the spokesperson of SENS. It’s probably because he’s one of the founders and because he seems to be an expert on all the topics involved. But that’s a rather boring explanation, so let’s try another one: Maybe the foundation – being completely aware of the controversial nature of their research – decided it was best to put forward somebody who looks and talks like a stereotypical genius. Ever since scientific progress has switched from infrequent but huge breakthroughs to small but steady improvements in all fields, there have been fewer iconic geniuses. If some scientist makes a discovery that can only be properly appreciated by other scientists working on similar problems, the general public doesn’t really care what he/she looks like – or about the discovery for that matter. But if someone claims (soon) to be able to abolish “death by natural causes” and even eventually rejuvenate people – he’ll better look as crazy as his ideas… for him to be taken seriously. See, people need to see some balance between mind and body. If your mind is different, then your body should be too. Would Einstein have been so popular if not for his nutty-professor looks? Among scientists, sure… but the rest of the world? Because there are others, like Feynman. He was a great scientist, a beloved professor and after hearing his lectures or reading his books one can hardly help admiring the guy. But he’s far from being as well-known as Einstein or Stephen Hawking (by the way, I really hope he lives to see the day when there’s a cure for his disease) – probably because there’s nothing special about his appearance.

Anyway, silly theories aside… I sincerely hope de Grey gets all the funding he needs.