While you are right about the limited applicability of Abrash's programming techniques, I think it is unfair to reduce his collective contributions to a "book of tricks".

I think the challenge was not merely optimization but also optimization within the limited realm of graphics programming, which had different challenges. You sound like someone who understands the basics needed to successfully optimize hardware and software performance, so I am sure you can appreciate how using a couple of otherwise vacant registers or figuring out the order of correct order of stacks/heaps could play a huge role in performance, at least back in the day.

Abrash's contributions were a combination of old-school tricks (especially his stuff from DDJ), an understanding of graphics programming from an algorithmic standpoint (and how to optimize them within the limitations of the hardware available), and were geared specifically towards optimizing game engines (and corresponding hardware recommendations). Sure, it's not quite the scope and scale of K&R's contributions, but that's like saying Feynman's work pales in comparison to Einstein's and Bohr's, so he was a hack.

That just shows how much the demographics of Slashdot have changed.

You must be kidding me.

When I was in high school, I discovered Abrash's Zen of Graphics Programming, filled with all kinds of gems. And then, Quake came out and there was his Graphics Programming Black Book.

Between x86 optimization, BSP trees, and assorted C/C++ tricks, Abrash's books were bibles at a time when graphics programming was just taking off.

I remember writing my own ray-tracer and 3d engine based on what I learned in his books.

Then there was his book on Zen of Code Optimization, which was amazing and filled with all kinds of computational optimization techniques for a time when not using a memory register effectively meant your render would stop halfway.

Michael Abrash and John Carmack were legends -- their techniques in optimizing rendering engines and their efforts in making graphics programming accessible to wider audiences were instrumental in enabling high end graphics. In fact, makers of graphics cards were known to design features based on optimization techniques that were used in Quake and other rendering engines.

And there was also something called "demo scene", where people built amazing programming snippets of graphics, media, and art. Between that and Abrash and Carmack's work, graphics got to where we are today.

So, yeah. Your question shows an unfortunate level of ignorance on the origins of the graphics programming industry.

It'll never come, but the year of the Linux Palmtop is upon us.

+1.

I wish I had mod points to mod you up.

As someone who's in fairly good shape and athletic, I have often wondered why people don't follow this simple dictum. In fact, diet is infinitely more important than exercise. And there's a very good reason it's said that six packs are made in the kitchen.

At the end of the day, someone who eats healthy and does not work out is often in better shape than someone who eats junk and "works out" for half hour a day. Most of those people just use their momentum to do some crazy exercises with piss poor forms, and eat unhealthy crap afterwards because they've worked out (think middle aged man with flabby biceps and a beer gut trying to bench press, when he probably has 30% body fat).

Ultimately, I have found that three things work for me:

1. Tracking what I eat like a hawk to ensure that I eat less than is needed
2. Eating a reasonable amount of protein (usually that entails eating adequate fat and fewer carbs)
3. Working out 3 times a week (2 days of rock climbing + 1 day of full body compound lifts - squats, bench, deads)

You can track your calories and protein on websites like LiveStrong or My Fitness Pal, or you can go old school with a scale and play by ear -- either way, ~3500 calories = 1 lb. Eat more than that, you gain weight. Eat less, you lose weight. Sometimes, you retain water weight and it may take a little while, but as long as you are consistent, you will see results.

I have met a number of people who are rock solid programmers and have a deep understanding of technologies. People who can program device drivers in their sleep and have implemented a godawful number of systems over the years. People who have licked networking or embedded systems or whatever (take your pick).

Naturally, they assume that CS is the same as IT, and enter CS programs to get a degree.

And then, I have seen them fail miserably as they realize that programming does not equal discrete math, graph theory, or computational complexity. Usually, it's been a while since they've been out of school, so even simple things like Graphics 101 with vector math and basic physics isn't quite a cakewalk. Plus, I have found that they are quite limited by their own experience and biases (mostly because they've had a lifetime to learn bad habits) and find it quite hard to reconcile real world experience with the academia.

You can especially see this with older, more experienced folks in a class teaching, say, Operating Systems, Architecture, Data Structures, or Compiler Design. And it is not necessarily their fault -- their real world experience sometimes does contradict what's recommended in the "ivory tower" world. Networking is often quite the opposite, though -- it is one of those fields where real world experience proves valuable, and the experienced folks learn a little something about the math behind network routing and such.

Honestly, whenever I see someone with experience wanting to study CS, I just recommend that they get a degree in something like MIS simply because it is a way for you to move up, and it is a lot easier -- handing computer science at a later stage in your life is usually significantly harder unless you've been keeping yourself mentally challenged in math and related subjects. You are in a very different place mentally in your early 30s than you are in your late teens.

You're probably not high enough on the totem pole -- my take from reading that article was that the collusion targeted poaching of high-value employees whose loss would hurt the company in question.

Individual contributors, by their very nature, are usually not worth the concern (except in rare cases).

I think there may be two elements to it -- one is the criminal aspect (i.e. it is illegal) and the other is the civil (i.e. it has other consequences that could result in a civil class action lawsuit).

Ultimately, I think that even if it is not illegal per se, the affected employees could still file for a civil suit citing any number of reasons. Now will that happen? Probably not.

Institutional memory... hmm.

I once stayed on a farm where livestock is raised for organic pork and beef. The farmer's view was that, in order to eat healthy in this day and age, a relationship is necessary between the consumer and the farmer.

I agree, and think relationships are beneficial in other domains too -- healthcare, education, law advice, between employer and employee, even packaged goods...

Nassim Nicholas Taleb talks about it in his books, "Fooled By Randomness" & "The Black Swan".

The idea is one of scalability. An athlete paid for a game or a lawyer or a doctor or a consultant are all financially similar to hookers -- i.e. you trade your skills for money for a given period assuming a given performance. No matter how great a hooker or a doctor or a lawyer or an athlete you are, there is a finite limit to how much money you can make. What more, there is a finite time during which you can monetize your performance, before you have to charm new clients. And even then, you stop because your performance stops to fall.

However, a lot of other professions are not limited by this. They are much more scalable. Imagine JK Rowling or Bill Gates or Dr. Dre. You can write one book or one song that keeps on selling millions. You can have a product that's only limited by your target market, which could keep on growing. Sure, you can write more books or make more songs or update your product, but that only serves to increase your incoming revenue.

The difference is, an athlete's skills are inherently not scalable. But an engineer's skills are. However, your chances of scaling are much lower, but not any lower than that of getting picked to be an NFL quarterback.

The other aspect to consider is also the average salary of all engineers (i.e. graduated with an engineering degree) and their unemployment rate over time compared with the average salary and professional employment status of anyone who's an "athlete" (played varsity or any sport in college). Not sure how the numbers would pan out, but I have a feeling that over time, the engineers would probably do better, with some athletes doing exceptionally well (i.e. you were on the Oxford Crew team, went to grad school at Harvard, and worked in private equity - such as this guy - while the rest wouldn't do so hot).

What lack of diversity? Just because it's not diverse in the way he wants it (blacks and Latinos) doesn't mean it's not diverse. It's got plenty of racial diversity -- Indians, Chinese, Koreans, even South Americans and Eastern Europeans.

Ummm, how about American trademark law?

To ascribe the same level of brand awareness and value to a multimeter as a car is disingenuous. As any behavioral economist will tell you, they're not even in the same ballpark.

Did you seriously just compare buying a car with buying a tool? There's a order of magnitude difference. Even so, yes, they can. If there's a feature that's considered essential to the brand and is trademarked, they would too.

And you know what? Apple can sue other companies for making products that look like theirs. And they have.

That's a silly argument. A knockoff is still a knockoff -- while some people may not accidentally buy it, they are certainly using the reputation of an established brand (Fluke).

While I may not personally accidentally buy it, there are many who would. When I was in college, you'd refer to popular gizmos by their name and description -- you went to the store and looked for the "yellow" multimeter. Well, if you were lucky enough to remember the name, you'd buy the right thing. Otherwise, you'd buy the cheap thing.

Fluke makes damn good products. Making something cheaper that looks like it could be damaging to their brand.