Hello audience.
This blog has a new collaborator.
I have a friend who has been contrarian to most of the opinions posted in this blog. He felt enthusiastic about the possibility of posting the criticism himself, so I agreed to grant him the permissions.
Welcome, Howling.
Sunday, June 25, 2006
Hello world
Subscribe to:
Post Comments (Atom)
6 comments:
So you see a future for FSB and Itanium
Not really a future, but I do not think they spell doom, massive extintion or total disaster for Intel either.
Don´t you think that if eddie allowed me to be a coauthor in his blog is because he respects my oppinions, and thinks I have something to contribute? Please, do not insult him!
Sarcasm will not get you too far around here, at least not with me. At the very least, wait until I make a few posts, and then give us your opinion.
Salud!
"So you see a future for FSB and Itanium'"
first of all, welcome howling into this blog. in my opinion, this blog is a lot more logical and constructive compared to sharikou's or MMM's. i sincerely welcome your contribution.
i think FSB should be retired before 2010, as it is becoming a clear bottleneck for intel. however, from the results of NGMA, FSB can still hold its ground. just as mr. sood once said, customers don't care if the technology is obsolete or not. if it still can perform, and perform well, there is no need to replace it.
Technologies may become obsolete, but ideas may not.
The BUS idea has its uses. But in this time, buses for memory/inter processor interconnects is inadequate. A FSB is like an ethernet hub, existing the switches, it is stupid to use them.
That doesn't mean that the concept of "Bus" is stupid. Wireless networks necessarily uses buses, conceptually.
"Technologies may become obsolete, but ideas may not"
So true, If you look into the past one can see the future but in a faster form, I was a keen follower of the Transputer from Innmos with its serial links "bus" to connect to other transputers and its parallel computing languages that support it,
AMDs HT is the modern equivilant, I can see a time when 1000's of fibre optical
connections will be the future HT's so the idea of parallellism for scaling performance is a given thats where my comment on the Itanium was pointing too ie the WLIW architecture offering a way to parallel execution of code and data A good idea who's time will come!
I also remember the idea of the transputer. I didn't study it enough to find the fatal flaws of its initial implementation.
But the important thing is that you recognized that the key attribute of increasing (rather than decreasing) the system bandwidth every time you put an extra processor, and the "linked routing" mechanisms intrinsic in AMD's multi-proc arch.
Post a Comment