How are AMD & Intel doing muti-core-wise?

  • Thread starter Thread starter sillyputty
  • Start date Start date
S

sillyputty

Been busy and haven't had a chance to check into the 'processor race.'
Last I heard, AMD was putting out a quad core to challenge Intel's
quad.
 
sillyputty said:
Been busy and haven't had a chance to check into the 'processor race.'
Last I heard, AMD was putting out a quad core to challenge Intel's
quad.

Phenom from AMD. TLB bug, BIOS fix for now, and new stepping someday soon.
http://en.wikipedia.org/wiki/Phenom_(processor)
http://www.theinquirer.net/gb/inquirer/news/2008/01/11/barcelona-b3s-fine

AMD's 45nm will be a half-step, by the sounds of this.
http://www.fabtech.org/content/view/3868/

Intel is currently at Wolfdale/Yorkfield dual/quad 45nm processors.
Nehalem will be a socket change.
http://en.wikipedia.org/wiki/Nehalem_(microarchitecture)

Paul
 
Multi-core, a pathetic fall-back path for the cpu companies that can
no longer make faster cpus because of physical (molecular/atomic)
limitations. The average user needs quad-core or even dual as much as
they need AWD in a car. Nice to brag about but unused and unneeded.

What I wanna know is, what's happening on the speed of cpus. Are we
really stuck at 3.X GHz until fundamentally new materials are used in
cpus? What's happening to speed improvements?
 
Multi-core, a pathetic fall-back path for the cpu companies that can
no longer make faster cpus because of physical (molecular/atomic)
limitations. The average user needs quad-core or even dual as much as
they need AWD in a car. Nice to brag about but unused and unneeded.

What I wanna know is, what's happening on the speed of cpus. Are we
really stuck at 3.X GHz until fundamentally new materials are used in
cpus? What's happening to speed improvements?
Well, seeing as you seem to think that raw Mhz is the be all and end
all, it shows how little you know. Obviously the thick end of a decade
of AMD matching Intel performance at lower clock speeds taught you
nothing.

--
Conor

As a Brit I'd like to thank the Americans for their help in the war
against terror because if they'd not funded the IRA for 30 years, we
wouldn't know how to deal with terrorists.
 
Con> Well, seeing as you seem to think that raw Mhz is the be all
Con> and end all, it shows how little you know. Obviously the
Con> thick end of a decade of AMD matching Intel performance at
Con> lower clock speeds taught you nothing.

Clock speed is important for lots of numerical problems that can't be
parallelized, or for numerically intensive realtime apps. The
multi-core model is a joke since very few if any programs for consumer
use have been parallelized or could even make use of
parallelization. Multi-cores are just a gimmick to keep selling
hardware to twits.
 
from:

http://www.labwithleo.com/transcripts/LWL-transcript-episode143.pdf


Caller:

I'VE GOT A PROBLEM.
I'VE GOT A COMPUTER THAT'S
RUNNING A PRESCOTT 3.4 Pentium 4 IN IT.
AND IT RUNS EXTREMELY HOT.



Answer (with snippage):

THAT'S ONE OF THE
REASONS THE PRESCOTT IS A FLOP.

YOU KNOW WHO WAS
REALLY DISAPPOINTED WAS Intel.

THEY WERE TALKING ABOUT GETTING
THAT CHIP UP TO 4 GIGAHERTZ
UNTIL THEY REALIZED THAT THE
THING WAS PUTTING OFF SO MUCH
HEAT YOU COULD BOIL WATER ON IT.

AND THEY ACTUALLY KIND OF
CHANGED THEIR WHOLE CHIP
STRATEGY BECAUSE OF THAT ONE
Pentium 4 CHIP.

THEY SAID WE'RE GOING IN THE
WRONG DIRECTION AND THAT'S WHEN
THE Core 2s CAME OUT.

THE Core 2s RUN AT
SLOWER CLOCK SPEEDS.
YOU'RE NOT GOING TO SEE ANOTHER
3.4 GIGAHERTZ CHIP PROBABLY.

WHAT Intel FOUND, AND THIS WAS
THE BIG DISAPPOINTMENT WITH
PRESCOTT - PRESCOTT WAS A VERY
BIG CHIP WITH LONG PIPELINES,
LOTS OF THEM, A LOT OF
TRANSISTORS IN THAT CHIP, AND
WHAT THEY FOUND WAS THEY HAD
GONE KIND OF DOWN THE RABBIT
HOLE WITH THAT CHIP DESIGN.
THEY PUSHED IT TOO FAR AND IT
WAS GETTING TOO HOT, IT WAS TOO
INEFFICIENT.

SO, THEY WENT BACK TO THE
DRAWING BOARD AND FORTUNATELY
THEY HAD A LITTLE SKUNK WORKS
PROJECT GOING ON IN ISRAEL.
THEY HAD DESIGNED THE LITTLE
PORTABLE CHIPS FOR Intel THAT
WERE SO SUCCESSFUL AND THEY
SAID, GUYS, WE NEED A NEW
DESKTOP PROCESSOR AND THAT'S
WHEN THEY CAME UP WITH THE
Core, FIRST THE Core Duo AND
THEN THE Core 2.

BELIEVE IT OR NOT, THESE CHIPS
ARE ACTUALLY CLOSER TO A
Pentium 3 THAN THEY ARE A Pentium 4.
THEY HAVE SHORTER PIPELINES,
BUT THEY RUN COOLER AND THEY
RUN MORE EFFICIENTLY.

AND MOST IMPORTANTLY, THEY GET
MORE WORK DONE IN FEWER CYCLES.
SO, YES, THEY'RE NOT RUNNING AT
3.8, 3.4 GIGAHERTZ.
THEY'RE RUNNING, YOU KNOW, 2.4,
2.8, 3 GIGAHERTZ.

BUT EVEN AT THOSE SLOWER CLOCK
SPEEDS THEY'RE MORE EFFICIENT
AND THEY'RE A HECK OF A LOT COOLER.

SO, MY RECOMMENDATION, IF YOU
WANT TO UPGRADE, IF YOU'RE
FINDING THAT THAT PRESCOTT IS
TOO HOT, MY RECOMMENDATION
WOULD BE GET A NEW MOTHERBOARD
THAT SUPPORTS THE Core 2, GO
WITH A Core 2 Duo; I THINK
THEY'RE VERY GOOD.

IF YOU WANT TO STAY COOL, GO
WITH THE SLOWER CLOCK SPEEDS
BECAUSE THEY -- YOU KNOW, WHEN
THEY'RE WORKING, THEY'LL GET
HOT JUST AS MUCH AS ANY OTHER ONE.
YOU KNOW, THESE COMPUTERS ARE
HOT NOWADAYS, BUT THEY'RE MUCH
MORE EFFICIENT AND MUCH COOLER
THAN THE PRESCOTT.

THE PRESCOTT WAS JUST AN
UNFORTUNATE CHIP DESIGN AND
Intel BACKED OFF ON THAT COMPLETELY.
THAT WAS AN EMBARRASSMENT,
FRANKLY, FOR THEM.

BUT WHAT'S INTERESTING IS THEY
WERE ABLE TO TURN ON A DIME
THANKS TO THIS LITTLE ISRAELI
SKUNK WORKS AND THEY WERE ABLE
TO COME UP WITH A NEW DESIGN
 
<lots of cutting>
geoff> AND WHAT THEY FOUND WAS THEY HAD GONE KIND OF DOWN THE
geoff> RABBIT HOLE WITH THAT CHIP DESIGN. THEY PUSHED IT TOO FAR
geoff> AND IT WAS GETTING TOO HOT, IT WAS TOO INEFFICIENT.

No doubt, hence my question. What new semiconductor technology, if
any, is in the pipeline to continue single-cpu throughput? Because
the multi-core model is going to have a very limited life.
 
The only thing they have is dual/quad core. I also remember reading in 1998
or so an article in forbes magazine that Intel would not be able to continue
with the current technologies because of costs.

They showed a graph with sales and a graph of costs. The big cost was
retooling a factory to make the new chip. The lines crossed each other in
about 2002 and the conclusion was at that point, Intel would have to find a
completely new technology.

However, they kept going with the same technology. During the 90s, there
were also articles about using bio-chips but that hasn't seemed to happen
either.

If I had to guess, I would say they have another skunk works going on
somewhere that is looking at the cutting edge technologies to see if
anything can be used/produced mainstream.

--g
 
Back
Top