INTEL is the best

marauder

Originally posted by Syxpak
Get 4-in-1 socket Alpha EV8 setup and then come back and say that intel/amd are "fast" cpus...

LOL

The EV8, previously covered in a three part series (Part 1, Part 2, Part 3), was a fourth generation Alpha processor core that targeted a 0.13 um SOI CMOS process with copper interconnect. ...Unfortunately Compaq canceled this project last June , more than a year before first silicon was planned. [/B] .... Ironically the former EV8 team is still nearly intact but now works for Intel designing future products in the IA64 family.

from http://www.realworldtech.com/page.cfm?AID=RWT021802145442

SouperComputer

i find it amusing that apple also sell current imac's that have clock speeds similar to my 2 year old pentium 3 system

never said clock speeds had anything to do with power?

then whats amusing?

Woden

that intel was selling chips with clock frequencies like that so long ago, perhaps due to increased competition from amd

cerebus

Originally posted by Dataisgod
that intel was selling chips with clock frequencies like that so long ago, perhaps due to increased competition from amd

Hmmm... don't take this the wrong way, but you seem to have been suckered by the relentless marketing hype that proclaims higher clock speed is automatically better. Don't feel bad - a lot of people fall for it, which is why incremental improvements allow Intel/AMD to sell higher frequency chips at a large premium to slightly slower ones.

As some other posters have pointed out, the PowerPC architecture is pretty good at what it does... it is a bit unfair to use clock speed as the basis for comparison with Intel/AMD products.

Bringing it a bit further, some of the more elegant IC engineering I've seen doesn't use a clock at all... I can even point you to a processor example or two. How would you classify these? Does that fact that a device effectively has a "clock speed" of zero make it infinitely worse than than the latest whiz-bang x86 based architecture? Does it make it infinitely worse than a 4.77MHz 8086?

[rant]
As an aside, fast clocks are getting to be a real problem in the industry - not sure if you know much about how a clock topology (tree/mesh/whatever) is structured, but really fast clocks in really big devices cause all kinds of nasty things to happen.

[historical part of rant]
Seeing as how the Alpha has been mentioned already in this thread I'll throw it out there as an example - back in 1998 the Alpha 21264 had an average power consumption of 60W, running at 700MHz. This was due to their clocking approach, which required all kinds of brute force to get true single-phase clocking all through the die.

Scaling up linearly with frequency, the power of parts running today at 3GHz+ would be >240W - the only reason it isn't is that designers have had to put huge amounts of effort into reducing this power. Even with all this effort, recent P4s have an average power consumption of ~80W I believe (they quote 81.8W as the thermal power on the latest datasheets I've seen)
[/historical part of rant]

I guess the point of my rant is that a fast clock speed is not the greatest thing since sliced bread. There are more efficient ways to do things which can provide the same throughput/performance with slower frequencies.

[/rant]

Sorry if it sounds like a personal attack, certainly isn't meant to be one - just my opinion. Consider it an educational experience (There, look a smiley - everything is okay)

Gerry

ultimate performance to power ratio has to be the centrino. The original p6 design team in israel basically souped up the p3. It has the same number of execution units, but supports SSE, has 1 meg of L2 cache, and a major focus on power management which gives you a 1.6ghz chip that is faster ( at desktop apps at least ) than the p4 2.4ghz ( around a 2.66ghz p4 apparently ), and 25 - 30w power consumption. It is of course being pushed only at laptops, but major demand from the big blue chips may see it in servers before too long.
With regard to the eternal amd vs intel debate. Theres still a perception that more mhz = more performance. As plenty of people are trying to explain, this depends on how much work is done in each clock cycle. There is also the misconception where people regard a low speed chip clocked up to be architecturally inferior to the official chip at the higher speed. For particular cpu's, manufacturers go through a number of steppings. These steppings mean differences in design to enable higher yields, and/or higher clock speeds. Usually theres little or no performance difference. ( very rarely is there any difference, the only one I can think of is the p4 C1 stepping where intel added extra entries to the TLB cache ).
All chips of the same stepping are identically architecturally, they are built on the same line, to do the same thing, and are only separated after being tested to see what speed they are good for ( speed binning ).

po0k

Originally posted by marauder
LOL

The EV8, previously covered in a three part series (Part 1, Part 2, Part 3), was a fourth generation Alpha processor core that targeted a 0.13 um SOI CMOS process with copper interconnect. ...Unfortunately Compaq canceled this project last June , more than a year before first silicon was planned. .... Ironically the former EV8 team is still nearly intact but now works for Intel designing future products in the IA64 family.

from http://www.realworldtech.com/page.cfm?AID=RWT021802145442 [/B]

I know.
Since HP aquired Compaq they saw what a big mistake that was, anda - afaik - are still developing the chip, or at least maintaining it.
I threw that out as an example of what a really powerful processor is like. IBM's Power4s are apparently quite good too

Woden

cerebus despite your post dripping with pure patronisation i don't consider it a personal attack (look a smiley for you too so you don't lose sleep over it). I don't consider myself the be all and end all with computers and will freely admit that most of the posters here have more knowledge then myself. However imho i've more computer knowledge then joe bloggs and not been suckered into the "relentless marketing hype". I'm relatively familiar with the strategy that amd have ensured using more work per cycle and less cycles to achieve the performance of higher clocked intel chips etc... and hence there naming strategy and also recently with the barton core with lower clock speeds then some throughbreds but increased cache and such for improved performance. Bascially what i'm trying to say is that i don't think that higher clock frequency necessarily makes a faster chip however with the caveat that if all things are equal bar clock frequency the chip will be faster perhaps as an example (correct me if i'm wrong as i'm sure someone will) the
the 2.53Ghz pentium and the 3.06 one?