From: leebr@ecf.toronto.edu (LEE BRIAN) Subject: Re: WP-PCF, Linux, RISC? Date: Sat, 24 Apr 1993 04:15:44 GMT
In article <C5w9J7.JLL@Nyongwa.CAM.ORG> angcl@Nyongwa.CAM.ORG (Claude Angers) writes:
>In article <C5rx8B.Kzp@ecf.toronto.edu> leebr@ecf.toronto.edu (LEE BRIAN) writes:
>>In article <1qu8ud$2hd@sunb.ocs.mq.edu.au> eugene@mpce.mq.edu.au writes:
>>>In article <C5o1yq.M34@csie.nctu.edu.tw> ghhwang@csie.nctu.edu.tw (ghhwang) writes:
>>>>
>>>>Dear friend,
>>>> The RISC means "reduced instruction set computer". The RISC usually has
>>>>small instruction set so as to reduce the circuit complex and can increase
>>>>the clock rate to have a high performance. You can read some books about
>>>>computer architecture for more information about RISC.
>>>
>>>hmm... not that I am an authority on RISC ;-) but I clearly remember
>>>reading that the instruction set on RISC CPUs is rather large.
>>>The difference is in addressing modes - RISC instruction sets are not
>>>as orthogonal is CISC.
>>>
>>>--
>>
>>Theoretically supposed to be reduced.... not any longer. That's why everyone
>>is arguing about RISC v.s. CISC. Personally, I think CISC will win out.
>>Just take a look at the Pentium! (Not that I like Intel architectures either,
>>but that's another story...)
>>
>>bye!
>>
>
>Do you mean that the Pentium is better than a Risc? or that it will outsell
>them all? If the first, you have to remember that intel CISC (like the
>pentium) are a always a generation away from the best riscs... also Riscs
>cpu are more costly because they are not sold in the same quantities (not
>even on the same order)... but I remember reading about 3 years (maybe 2)
>about a T800(?) from hypercube that did a 100 mips, was superscallar AND
>reordered its instruction itself so
I'm not 100% sure, but I think the T800 was a 25MHz transputer? so ya tie
a gazillion of them together to get 100mips. (The newest is the T9000
which kicks anyone's butt :)... haven't seen them used much though).
Anyway, to respond, I think the Pentium (CISC) is better than the more advanced
RISC (e.g., like the alpha, etc. the 66MHz Pentium has approximately the
same "performance" as the superduper 133MHz Alpha - here, performance is the
weird Specint92 that everyone refers to? - this is what I *heard* - the
Alpha still kicks in the P5's butt in fp - again, this is what I *heard*).
and in the computing world, if you sell lots of chips (like intel), and
make it faster (like intel), you are the winner (like intel), even though
you have a sucky architecture from over 10 years ago (like intel :0).
If you can make a "CISC" chip (superscalar, superduperpipelined, superfast)
with the ideas behind the "RISC" ideology, you got a CISC chip. And then
I admit I can't see the advantages of RISC over CISC...
If the latest technology is a generation behind, then it sucks (relatively
speaking).
Now I may sound like I like intel, but I'll have to say that the P5 is some
real kick butt pile of Si and SiO2...
But I hope that Motorola really catches up with the 68K line... or I'm gonna
start crying...
brian
===============================================================================
Brian "Hojo" Lee | "Hey, excuse me miss, could I have a .GIF of you?"
leebr@ecf.toronto.edu |
leebr@eecg.toronto.edu | (try Linux... the best and free UN*X clone!)
===============================================================================