Re: Can anythink kill x86-64? (was Re: IPF's future?)

"David Kanter" <dkanter@xxxxxxxxx> wrote in message news:1146771950.603912.116730@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
There is little reason to think that the ISA has much, if anything, to do
with those differences. If you put a POWER (or PPC) decoder on the
front of an x86/AMD64 chip and measured the difference, I'm certain
the various SPEC scores would come out nearly identical to the
native ones. There's simply too many other factors in system design
to consider any "ISA" comparison to be meaningful these days. Core
design, caches, memory bandwidth, etc. will hide any possible
difference the ISA can make.

That is to some extent true. See my response to Bill about the
point I was making, both of you guys were responding to a subargument.

Indeed I was; I agreed with your main argument but had one nit with how you backed it up.

My point was that x86 chips are not the fastest in the world by a long
shot, even if they are clocked the highest. They are seriously
outclassed by the POWER5+ in server benchmarks, which is lower
clocked. My point was that clock speed != performance all the time.

I think anyone who's used a wide variety of architectures would agree with that. Clock speed is relevant only within a family of processors sharing the same design. Comparing NetBust to the P6/PM or Hammer proves this even within the same ISA.

S

--
Stephen Sprunk "Stupid people surround themselves with smart
CCIE #3723 people. Smart people surround themselves with
K5SSS smart people who disagree with them." --Aaron Sorkin


*** Posted via a free Usenet account from http://www.teranews.com ***
.

Relevant Pages

  • Re: WMATA crash & track circuits
    ... So what are you trying to prove, that ISA is the future? ... If PCs are needed to interface, then it does not matter, what PCs. ... If you design embedded solutions you need no ISA either. ... If you design for 10 years, yeah, then stuff will go kaputt. ...
    (sci.electronics.design)
  • Re: WMATA crash & track circuits
    ... I wonder if you ever designed for industry, those things never were an issue. ... There is not much a difference between the cosmic rays at that altitude and those on top of a high mountain. ... So what are you trying to prove, that ISA is the future? ... ISA was easy to design, I designed ISA cards for industry for a living, hehe. ...
    (sci.electronics.design)
  • Re: 8086 powered IBM PC compatible
    ... ISA 16 and 8 bit slots, ... as I design and build my own ISA boards. ... support the old ISA bus. ... Supports DOS. ...
    (sci.electronics.design)
  • Re: Perimeter Dare!
    ... This may have been the preferred methodology in the past, but ISA is designed specifically to protect published servers. ... If you've got the resources using a DMZ/Perimeter network is preferred ... If you have a very secure network design ... Should I just Publish my web server from my Internal Network? ...
    (microsoft.public.isa.enterprise)
  • Re: What am I missing? Routing Troubles
    ... topology design which is absolutely crucial to undstanding routing issues. ... and at least one location has an ISA. ... Internal destination: RemoteClients ...
    (microsoft.public.isa)