You read that wrong. The very last line of the article says:
When we ran the Intel-compiled version of OBLcpu on the Athlon-powered system the percent
improvement [emphpasis mine] was virtually identical to the results run on the Omnibook.
IOW, icc-compiled code works damn well on AMD chips as well.
The real story is probably that 1) Microsoft had little to lose in rewriting, and 2) they have vast resources, and the confidence that they're going to win in the end, by copying the competition, catching up, and surpassing. Once they had established themselves as a standard, and forced themselves onto the desktop, they probably would have won even if Netscape hadn't taken a break. With IE standard on everyone's PC (and more and more sites being optimized for IE), Netscape doesn't have the same opportunity to win back that share.
In short, you can't compete against Microsoft, sad to say. If you want to sell software, make sure it's in an area M$ isn't interested in, or at least connect it with a service, like AOL or Napster.
Seems like it's more than this. Note (ii): not using Potentially Viral Software to develop Recipient software which includes the Software... This could mean that I can't use Emacs (which meets their definition of "Potentially Viral Software") to indent the code I write, if the result is to be distributed with part of the SDK, even if I don't distribute Emacs or any of its source along with it. IANAL, so it may actually not mean this exactly, but they may want end users to think this.
Turns out the article I referred to did not contain stuff about NT's high Total Cost of Ownership. Sorry. I originally ran across this article in Washington DC Computer User, which has the Userweb URL on the front. The newspaper had an adjacent article on how great Unix is compared to NT, but by the time I got around to posting to Slashdot, I'd forgotten these were two separate articles, and it turns out only the feature article is shown on their website.
Slashdot Mirror
Unfortunately, seems that Microsoft site doesn't work on browsers not made by Microsoft.
IOW, icc-compiled code works damn well on AMD chips as well.
The real story is probably that 1) Microsoft had little to lose in rewriting, and 2) they have vast resources, and the confidence that they're going to win in the end, by copying the competition, catching up, and surpassing. Once they had established themselves as a standard, and forced themselves onto the desktop, they probably would have won even if Netscape hadn't taken a break. With IE standard on everyone's PC (and more and more sites being optimized for IE), Netscape doesn't have the same opportunity to win back that share.
In short, you can't compete against Microsoft, sad to say. If you want to sell software, make sure it's in an area M$ isn't interested in, or at least connect it with a service, like AOL or Napster.
Seems like it's more than this. Note (ii): not using Potentially Viral Software to develop Recipient software which includes the Software... This could mean that I can't use Emacs (which meets their definition of "Potentially Viral Software") to indent the code I write, if the result is to be distributed with part of the SDK, even if I don't distribute Emacs or any of its source along with it. IANAL, so it may actually not mean this exactly, but they may want end users to think this.
Uh, hello, this was supposed to be funny, people!
Turns out the article I referred to did not contain stuff about NT's high Total Cost of Ownership. Sorry. I originally ran across this article in Washington DC Computer User, which has the Userweb URL on the front. The newspaper had an adjacent article on how great Unix is compared to NT, but by the time I got around to posting to Slashdot, I'd forgotten these were two separate articles, and it turns out only the feature article is shown on their website.