In a discussion on the merits of various processors, Torvalds wrote that Intel had made the same mistakes "that everybody else did 15 years ago" when RISC architecture was first appearing. Itanium tries to introduce an architecture that is clean and technically pure, something that just doesn't seem to work in the real world. He claims that Intel "threw out all the good parts of the x86 because people thought those parts were ugly. They aren't ugly, they're the 'charming oddity' that makes it do well."-- Linus Torvalds, Itanium "threw out all the good parts of the x86"....
Clever architecture is something that has trapped others in the past. The Alpha processor team spent years learning that many of the architecturally correct ideals they had held needed to be thrown out when it came to the real world. According to Torvalds, "And all the RISC stuff that tried to avoid it was just a BIG WASTE OF TIME. Because the _only_ thing the RISC approach ended up showing was that eventually you have to do the hard stuff anyway, so you might as well design for doing it in the first place."
Itanium has felt like a long, drawn out car crash moving in slow motion for about five years now. Looks like it'll be the biggest waste of time, money, effort and hype ever seen in the industry when it finally does die the death it so rightly deserves.
Reminds me about the design of our favorite scripting language....an architecture that is clean and technically pure, something that just doesn't seem to work in the real world...
Re:Itanium eqv Ada ?
inkdroid on 2003-02-25T18:38:19
And the merits of the X86 kind of reminded me of another language:
[Linus] claims that Intel "threw out all the good parts of the x86 because people thought those parts were ugly. They aren't ugly, they're the 'charming oddity' that makes it do well."I had a compsci prof once tell me that Perl looked like modem noise. At the time I figured he was talking about regexes. Those "charming oddities" sure are useful! It's funny how some so called pure languages (Java) have learned to copy them.
Re:Pity it took down so many actually good chips
ziggy on 2003-02-25T18:52:37
I don't know about that. SPARC isn't going away anytime soon. There really isn't anything that can take its place at the high end (StarCat and StarFire), and a lot of apps running on SPARC are the kinds of things that very conservative organizations deploy with a five+ year horizon. Beowulf clusters are nice, but they're not a mainframe, and not easily partitioned yet.The huff and hype about Itanium scared nearly everyone from CPU R&D, but there's still stuff in the pipeline. AMD's x86-64 (or Intel's skunkworks clone) will probably see the light of day this year, and start to get adopted next year. Sun is making some announcements about the future of the SPARC architecture this week.
And let's not forget that there are some serious shortcomings to the x86 architecture, whether you get it from Intel or AMD: It's hot, and it's power hungry. This is why Sun's 650 MHz UltraSPARC blade systems use less than half the power of their AMD-based designs. These considerations are important if you're going to be deploying a lot of servers in a closet...
Of course, there's still a lot of life and interest in Power/PowerPC, as well as boutique "embedded" architectures like ARM/XScale. It isn't an all x86 world after all.
:-) The only really sad footnote is that DEC couldn't do more with Alpha back when they had the chance. Maybe Sun (and AMD) really did it the right way: take an existing 32-bit architecture and add 64-bit extensions... x86-64 has good chances
jmm on 2003-02-25T18:53:18
The x86-64 line has one really strong advantage - it is a progression from the x86-32 line and will presumeably be compatible without being slower than the contemporary x86-32 chips available, while providing new benefits (addressing beyond 4 GB without using segmentation kludges).The x86 compatibility has been the killer for all potential processor families. Some have died, others are still around but limited to low volume niche makets (compared to the generic PC market that supports the x86 processor).
A "replacement" that has to use emulation and thus ends up with lower performance, does not capture the market.
