New iMacs Get Good Compliments and Superior Rating From PC Magazine

[Q Classic]

too bad imacs are pug-fuggly.

that and are overpriced. and not expandable, not in any true sense.

[FlameFlash]

And thank you for that enlightening comment...

[monolith94]

Flame Flash: I'd kill for that pic. Any chance I can find it?

[FlameFlash]

monolith... drop me a PR (with your e-mail) or e-mail through 'poly and I'll forward it to ya. I just don't want to get 'poly into trouble for them having it on their severs.

And that goes for anybody that might want to get a copy of it.

Personally, I'm still a fan of how the old graphite G4 towers looked, even these newest ones just look too metallic for my tastes.

[Mr. Nice Guy]

Mac vs. PC

In my opinion of Mac vs. PC, I say it really depends on what you want. I'm not discrediting the PC, but I feel that now more than ever Apple stands a good chance of hitting it big in the personal computer market. I'm not sure how big, but big enough to effectively compete with PCs. Don't ask me why; it's just my gut feeling.

But hands down, if you prefer performance and affordability over stability and ease of use, then get a PC. But if you prefer stability and ease of use over performance and affordability, then get a Mac. It's that simple.

[Asher]

It's not that simple, the stability comment is jibberish. And if you really have problems using a PC still, one has to wonder just how bright one is.

[Urban Ranger]

Originally posted by Asher
About Prescott... Prescott is being designed to be 64-bit (using x86-64), but Intel doesn't want everyone knowing about it.

So how do you know?

Originally posted by Asher
Intel doesn't want the market to adopt x86-64, and instead wants them to wait a couple years and ditch x86 completely and use Intel's new IA-64 (which the Intel Itanium and Itanium 2 chips use). So depending how "vital" they see 64-bit consumer chips at the time it comes out, they may or may not enable it.

Prescott, as far as I can tell, is your basic FUD. Intel is basically saying, "Hey lookie here, we have this CPU here that is way better than that pathetic AMD Hammer. Way, way better. What? You want specs? None can do, it is top secret. Sorry."

Originally posted by Asher
The main problem is AMD's upcoming Hammer series of chips run x86-64, which would mean Intel would have a 32-bit consumer chip for 1-2 years while AMD had a 64-bit on the market. (Note that 64-bit isn't inherently faster than 32-bit, it just supports larger system RAM and integer precision, which Intel argues isn't needed for the desktop yet)

What have you been smoking? Given that all else is the same, with a data path twice the size twice as much data can be moved in the same amount of time, giving better peformance at the same clock speed.

[monolith94]

vaporware then, UR?

[Asher]

Originally posted by Urban Ranger
So how do you know?

Because it's been leaked. It's called Yamhill. Intel is now trying to cover it by officially "canceling" the project, but it doesn't take much to see they'd continue doing it because it's not difficult to add to the existing core (minimal increase in die size), and if they gambled incorrectly they'd give the market away to AMD. They'll still keep it as a backup plan.

Prescott, as far as I can tell, is your basic FUD. Intel is basically saying, "Hey lookie here, we have this CPU here that is way better than that pathetic AMD Hammer. Way, way better. What? You want specs? None can do, it is top secret. Sorry."

The specs have been leaked, just like Hammer's. Hammer is far closer to release (by at least 4 months), but Prescott's specs are generally accepted to be 800MHz system bus, 1MB L2 cache (or more), and Yamhill, as well as SMT (which is already in all P4s, just not enabled, and is allowed to be enabled in P4 Xeons).

What have you been smoking? Given that all else is the same, with a data path twice the size twice as much data can be moved in the same amount of time, giving better peformance at the same clock speed.

The data paths have already been wide enough for up to 80-bit numbers in the CPUs. SSE2 can handle 64-bit floats, x87 can handle up to 80-bit floats. Adding "64-bit compatibility" to a chip just means increasing memory addressability and increasing integer precision from 32 to 64-bits. In this case they also did some other modifications to try to fix lots of x86's shortcomings (ie, more registers), but those are all trivially added and don't result in performance increases in and of themselves.

Just because a couple integer data paths go from 32 to 64-bits doesn't change the fact that the ALUs don't process 32-bit numbers in pairs, so when dealing with those 32-bit ints you're only using half of the data pipe. It's a fact of life.

[Urban Ranger]

Originally posted by Asher
The problem is using x86-64 is the cheap/lazy way to go about getting 64-bit chips. x86 is way too old, and huge chunks of modern x86 processors exist just to get around the limitations of x86 itself. It was never meant to last this long, and technology has advanced so much that we need a clean break from x86 (IMO).

It's not that the x86 is way too old but it is an inherent problem with CISC machines. AFAIK, there is no special problems to this particular instruction set.

Originally posted by Asher
Intel and HP spent the last ten years developing something called IA-64, which is a totally and completely new instruction set and design philosophy that actually significantly outperforms Apple's G4 at the same clockrate (the 1GHz Itanium 2s score 8-10x higher in the SPEC benchmarks than Apple's G4).

Correction: HP has abandoned development of EPIC for something like 5 years.

Correction 2: VLIW is not a new design philosophy. It really is just modified RISC architecture.

Just curious though, by supporting EPIC, are you admitting that RISC is superior to CISC?

Originally posted by Asher
The problem is, the technology is still being "perfected" in the high end server and workstation market, and it's not expected to trickle down to the consumer market for a couple more years.

There are two big problems. One, Intel created a monster that not even it can control in the form of x86. There is just way too much invested in software and hardware to break away from it.

Think COBOL. It was supposed to last for 5 or 10 years only. Five decades later, it is still widely used.

The second problem is with the ubercompilers that Itanic relies upon. They don't exist.

Originally posted by Asher
It is also pretty slow emulating old x86 programs right now, which is a main problem for consumer PCs as well.

This is not a big problem. Most people just won't notice because of the excessive power current Wintel boxes have.

Originally posted by Asher
AMD lost $9M in that same period.
Intel made $936M in that same period.
AMD doesn't have much of an idea on how to run a business. It's a shame, really, since they finally have decent engineers.

Poor form. Most businesses lose money during lean periods. That doesn't mean anything.

[Mr. Nice Guy]

Originally posted by Asher
It's not that simple, the stability comment is jibberish. And if you really have problems using a PC still, one has to wonder just how bright one is.

Ah, that's gibberish! Just listen to what all the Mac users out there have to say about the stability of Macs.

[Urban Ranger]

Monolith,

Entirely possible.

Originally posted by Asher
Because it's been leaked. It's called Yamhill. Intel is now trying to cover it by officially "canceling" the project, but it doesn't take much to see they'd continue doing it because it's not difficult to add to the existing core (minimal increase in die size), and if they gambled incorrectly they'd give the market away to AMD. They'll still keep it as a backup plan.

Sure it has been "leaked," that is one way of making disinformation look official. We have all seen this haven't we? It is not unusal for a company to arrange for certain information to be carefully "leaked."

Certainly it doesn't surprise me that Intel has a backup plan just in case EPIC doesn't work out.

Originally posted by Asher
The specs have been leaked, just like Hammer's. Hammer is far closer to release (by at least 4 months), but Prescott's specs are generally accepted to be 800MHz system bus, 1MB L2 cache (or more), and Yamhill, as well as SMT (which is already in all P4s, just not enabled, and is allowed to be enabled in P4 Xeons).

Hammer is now officially out. You can find out the specs easily.

As for SMT, it can be interesting but the benefits are highly theoretical for Wintel boxes since the systems weren't designed around it. Neither was P4, as a matter of fact. Glueing on extra pieces isn't going to work well, if at all.

Windows needs also be modifed.

Originally posted by Asher
Just because a couple integer data paths go from 32 to 64-bits doesn't change the fact that the ALUs don't process 32-bit numbers in pairs, so when dealing with those 32-bit ints you're only using half of the data pipe. It's a fact of life.

You are right if all the ALU's are fully utilised every clock cycle. Unfortunately for you, that is not the case. The main problem with fast CPUs nowadays is the ALU's sit idle waiting for data (both opcodes and oprands) to arrive. Going with a 64-bit wide data bus goes a long way of solving the problem.

[Mr. Nice Guy]

Mac Gabbery

Here's a link for ya, Ash. Despite a few rants and raves about Windows, there is plenty more more rants and raves here about the Mac. After clicking on this link, type 'Mac vs PC' (without the quotes) in the search box, then choose 'Mac Gabbery' in the 'Search Forum' field. Don't change anything else.

http://www.macworld.com/cgi-bin/ubb/ultimatebb.cgi?ubb=search&search_forum=40

[Asher]

Originally posted by Urban Ranger
Correction: HP has abandoned development of EPIC for something like 5 years.

They did?

http://www.hp.com/products1/itanium/

Correction 2: VLIW is not a new design philosophy. It really is just modified RISC architecture.

Just curious though, by supporting EPIC, are you admitting that RISC is superior to CISC?

RISC is superior to CISC, it's time to move away from CISC. CISC had its heyday. Modern CISC CPUs become RISC after they spend 1-2 pipeline stages decoding the CISC instructions into RISC-esque micro-ops anyway, that's the only reason they're fast.

There are two big problems. One, Intel created a monster that not even it can control in the form of x86. There is just way too much invested in software and hardware to break away from it.

Think COBOL. It was supposed to last for 5 or 10 years only. Five decades later, it is still widely used.

This would be a problem if IA-64 could not run x86 code. Alas, it can, just not very fast (for now). When Intel moves it to the desktop, you can bet on them having equivalent x86 performance to the top end x86 chips, as well as supporting IA-64. Otherwise no one would buy it. There's even speculations of Intel shrinking down the P3-T core with improvements even further and integrating it inside the IA-64 desktop version. CMP.

The second problem is with the ubercompilers that Itanic relies upon. They don't exist.

Ah, this still is a problem, but not as big as one two years ago. Or one year ago. Until last year there wasn't any IA-64 hardware to compile to, now there is. Compilers are definitely becoming smarter and better as more applications and more fine-tuning is going on for the compilers. Look at the Itanium 2 SPEC benchmarks for how a good compiler can work.

This is not a big problem. Most people just won't notice because of the excessive power current Wintel boxes have.

It is for the enthusiast market, which is the first group of people who embrace new technology such as this.

Poor form. Most businesses lose money during lean periods. That doesn't mean anything.

I was making a point to Gianny who was talking about how much of an example AMD was compared to Apple, when AMD lost money during that period while Apple still turned a profit.

[Asher]

Originally posted by Cloud9
Ah, that's gibberish! Just listen to what all the Mac users out there have to say about the stability of Macs.

I'm not saying Macs are not stable (OS X is quite stable), I'm saying PCs are just as stable as Macs are these days.

Here's a link for ya, Ash. Despite a few rants and raves about Windows, there is plenty more more rants and raves here about the Mac. After clicking on this link, type 'Mac vs PC' (without the quotes) in the search box, then choose 'Mac Gabbery' in the 'Search Forum' field. Don't change anything else.

http://www.macworld.com/cgi-bin/ubb...search_forum=40

Can you explain to me why I'd be interested in reading some posts from fanatics who really have no clue what they're talking about, Cloud9?