I was reading Slashdot this morning and saw a link to a Geek.com story entitled The x86 evolution will lead the revolution. Now I'm not exactly sure what "revolution" this is talking about. There are so many these days. But one thing that I have learned in all my time in the industry is that technology ALWAYS changes, and trends can NEVER be predicted with 100% accuracy. How many of you bought a PhysX by ageia? How many of you would have 5 years ago predicted that AMD would get consistently spanked in performance by Intel's LOWER end processors? For many years in this industry people try to make predictions that they are SURE are going to come true. I can't tell you how many predicted that ATI would dominate the graphics card market. But they ended up in trouble and got bought by AMD. Now neither AMD or ATI has a leading product in their market (and it is not for lack of trying).
Or 3dfx? They REALLY dominated the graphics market in the late 90's. EVERYONE said that the 3dfx cards were far better than that Riva or Rage products (by Nvidia and ATI respectively). After the company released its Voodoo 3 card it seemed to disappear into a puff of smoke. Nobody predicted that with 100% accuracy. This industry is one of the most dynamic industries I have ever observed. Even in the retail support market, which I heavily work in, technology turns around at the drop of a hat. First its capacitive touchscreen technology. Then it is resitive. Then, infrared touch rings around the screen. Then back to resistive. And every time the new tech is touted as the answer for all of the design problems in the units. And it will be something else next month. I can't tell you how many times the IT and computer industry has flip/flopped. Its the nature of high tech.
Now don't get me wrong. It is a fantastic feat of the x86 architecture to have survived as an industry standard for so long. Heck, I remember when "x86" architecture was known as "IBM compatible". And even that nomenclature lasted for a long time. So lets give the platform its due. The above article cites the use of increasingly smaller nm production processes, thinner substrates and better power efficiency like in Intel's new mobile chip the Atom. He mentions the 64-bit x86 technology extensions. However, he uses that for evidence of the platform's dominance. Personally, though 64-bit capable chips are well dispersed through the market, MOST of them are not taking advantage of the 64-bit extensions i.e. running a 64-bit OS. I'm sure they will, when they discover that they can't see all 4GB of their system memory in 32-bit mode, but that has yet to happen.
From a distance, one could see how the author made the assumption that x86 will continue to dominate. It has a track record of success, and it was even recently adopted by Apple for its Macs, abandoning the PowerPC market. (Who could have predicted THAT 5 years ago?) But believe me folks, it can all turn around in an instant. Let me lay out this scenario. What if I bring to the market a new, RISC based processor? It doesn't run any native x86 apps. It DOES, however, have a complete Virtual Machine. Virtualization is new and hot, and especially in server environments, it is the answer to alot of issues. Many people are using it. It runs virtual machines far better and more efficient than an x86 based machine. It would be very easy for a product like this to dominate the server market, especially if that customer is using virtual machines as an exclusive infrastructure. They could care less what kind of instruction set the physical cpu has, just so long as their virtual machines are solid. Virtualization may make all hardware platforms as we know them totally interchangeable. You could then run the VM on any hardware you chose. The second you make something faster and cheaper then -- Boom. Server x86 use goes out the window. Now this hasn't happened but it EASILY could.
Also, the author states that x86 will dominate in the mobile and embedded market. Again I have to say this is MASSIVE speculation by the author, almost to the point of being laughable. The future does look bright for embedded x86. But it is difficult at best to hold on to that market. Just ask Motorola. Or Transmeta. Or Palm for God's sake! Or any of the people who were going to change mobile computing for years to come. Perhaps x86 will evolve into a standard instruction set for all VM's in the future, leaving its legacy not in silicon chips, but in an idea, a multi-platform programming standard. That does not invalidate the author's point, but it isn't anywhere near the hardware dominance that he predicts.
The moral of this story is that its ok to speculate, but don't be surprised if those far touted theories crumble. Other than the history lesson, which I might add is very good work by this author, this article is more speculative than the psychic hotline. I encourage people to think about the future, but the main thing about it is that no person on this planet knows for certain what the future holds. Especially in the computer industry.
No comments:
Post a Comment