Author: Dann Corbit
Date: 13:02:22 02/18/04
Go up one level in this thread
On February 18, 2004 at 15:46:06, Bob Durrett wrote: >On February 18, 2004 at 13:42:37, Dann Corbit wrote: > >>On February 18, 2004 at 11:15:28, Bob Durrett wrote: >>[snip] >>>Anthony, sometimes programmers look down their noses at hardware people like me. >>> But if there were no hardware, programmers would be as helpless as a newborn >>>baby. There is more that is worthy than just programming. >> >>Will Rogers said it best: >>"Everyone is ignorant, only in different areas." >> >>Something like "What caused the space shuttle explosion?" Probably can't be >>answered in one sentence. Similarly, the contrasts and similarities between >>hardware and firmware and software development are intricate issues and >>difficult to explain. >> >>A lot of people are going to respond to the hype of the whole Hydra thing, which >>is fine. It *is* an interesting chess development (even if the chief developer >>is a bit of a loon). I think the real frustration is that explaining 'why the >>hype is hype' is not comprehended. >> >>Here is what we have with Hydra: >>A very interesting chess machine that could very well be the strongest chess >>playing engine that money can buy ( if you have a big enough pile of money ). >>But in a few years the investment may look foolish as the software approach will >>be faster. >> >>So, if you have a big pile of money, you want to win all the time now, and you >>are not afraid of some "early development issues" go ahead and buy a cluster of >>PC's and a load of Hydra boards. > >Let me put a more positive spin on the Hydra "thingie": > >Hydra is a "wake up call." [Is this hype?] Hydra is not a wake up call. It is another hardware chess program. These things have been around since the early 80's if not before. It is nowhere near as strong as Deep Blue. If that did not wake us up, how will this weaker attempt? >Hydra shows us that there is more to "computer chess" than writing code for >general purpose machines, like the PC. Hydra is a "special purpose machine." >It is not a "general purpose machine." There is a big difference because much >of the information about general purpose machines, and their architectures, may >not be applicable at all to the special purpose machine. Hydra has chess instructions, so it is only useful for playing chess. It has always been possible to do things like this. >Hydra promises the future because it shows the way to produce extremely strong >"research grade" machines which can be used for chess research. [More hype?] Definitely more hype. As soon as you make it, it is the present, but it fades rapidly into the past. Any Hydra machine will be outclassed in a few years. Even Deep Blue will be overtaken in less than ten years, and many millions of dollars were spent on that project. >I can see that a super-strong chess-playing machine might be useful even if it >might be too expensive for an ordinary individual consumer to purchase. There >are many extremely expensive and huge computers in existence today, some of >which are used for research. It's not like it hasn't been done already, except >just not for chess. Belle, Hitech, Deep Blue. Sound familiar? >Incidentally, not all of the big machines are "general >purpose" machines. Not sure what you mean by this >One of the first things a programming student in College must learn is the >"architecture" of the machine he/she is to use for programming. That's why >Computer Science Departments in universities offer courses in computer >architecture. Later, when the student graduates and goes into industry [or >starts a new business] to do programming, it is again necessary to become >familiar with the hardware on which the programs will run. The programmer will >likely not be expected to create a new machine from scratch and is typically not >trained to do so. > >It is therefore very common that programmers everywhere are very familiar with >the hardware which affects them. > >Most Computer Science students are trained to design hardware. No. Most EE majors who go into electronics design are trained to design hardware. Most computer science majors couldn't even program an EPROM. >The Engineering >Department in a university is preoccupied with training students in the skills >of design. Nowadays, with hardware, firmware, and software integrated in >systems, the design task is not just limited to hardware anymore. Never has been. >When we are talking about machines whose purpose is highly specialized, much or >all of the information about architectures of run-of-the-mill general purpose >computers is irrelevant. Hydra makes is an example, although the Hydra >designers still allowed themselves to be dependent on PCs. If you know what the instructions are before hand that you plan to execute, you can spend some work and make those faster. But then, you can't type letters with that FPGA or do database searches with it. So it has far more limited value in the general sense. Do not imagine that hardware systems are not programmed. They are. Instead of a subroutine library, you use your VHDL library. And instead of an optimizing compiler, you run it throught something like SIS. Here are some tools used for this sort of thing: http://www-cad.eecs.berkeley.edu/~pinhong/scriptEDA/ When you get down to the bottom layer, you are still doing programming. But the instructions are implemented in hardware or firmware rather than software. This makes the instructions better suited to a single task, but unsuitable for other tasks. Why not read a book about it at your local library?
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.