Go Back  DVD Talk Forum > General Discussions > Tech Talk
Reload this Page >

Interesting email about MAC vs. PC

Tech Talk Discuss PC Hardware, Software, Internet and Other Technology

Interesting email about MAC vs. PC

Old 08-03-05, 07:12 AM
  #1  
DVD Talk Godfather
Thread Starter
 
DVD Polizei's Avatar
 
Join Date: Jan 2002
Posts: 52,603
Received 1 Like on 1 Post
Interesting email about MAC vs. PC

Usually, I don't publish personal emails, but this I had to. This person is, in my opinion, the best in his industry. However, he constantly is wanting me (and others on his list) to purchase a MAC.

In addition to MAC's being free of viruses and trojan horses, one other tidbit of info that some of you may not be aware of is the "Megahertz Myth". This has to do with the difference in the pipeline architecture between MAC processors and the processors used on PCs; mainly AMD and Intel.

MAC uses a 7 stage pipeline architecture. This means information is being fed to the processor, down a 7 stage pipeline. When a "hiccup"
occurs, (information that kicks out and must be sent again) which is inevitable in the computer world, the entire pipeline must clear itself, before any new information is sent. With PC processors the architecture is a whopping 20 stage pipeline. This means when information is sent that is not understood and must be sent again, the pipeline must wait while 20 stages clear out - vastly slowing down the computers ability to recover quickly.

For this reason, MAC calculates their processor speed quite differently than PC's do. This means when choosing a MAC computer, you must take the MAC's listed processor speed and times it times 2.25% to see what the Pentium IV equivalent would be.

For example, if you see a MAC with a 1.45 Gig processor, that would be the equivalent to a Pentium IV 3.25 Gig processor. A MAC with a dual 2.75 Gig processor (like those on the new MAC G5's) would be equivalent to a Pentium 6.25 GIG processor - if Intel could even make anything that fast - which they can not yet.

About four years ago, at a Mac World convention in New York, Steve Jobs and his cohorts gave a very interesting speech on what is referred to in the MAC community as the "megahertz Myth". This is what PC retailers do NOT want you to know (and most do not even know themselves), but most MAC users are aware of (mainly through me of course). This keynote address explains in detail the pipeline architecture and why PC users think they are getting a faster processor than a MAC, but in reality are not.

Apple had this speech stored on their web site when it first was given, but has since moved on to bigger and better things. This weekend, I stumbled onto a web site that has the entire 8 minute keynote speech, which you can watch. It is quite fascinating. I thought some of you might enjoy learning a little something about the differences between MAC's and PC's - other than the obvious, which is MAC's do not have viruses and trojan horses. Here is that link -
ENJOY!:

http://eshop.macsales.com/tech_cente...=MHz_Myth.html

NOTE: In order to watch this movie, you will need to make sure you have Quictime installed on your PC. If you have an iPOD, then it installed with Itunes. If you do not have an iPOD, then you will need to go to this link and download the player first;. Yes, it's free! <grin>

http://www.apple.com/quicktime/

That concludes today's computer lesson. <lol> Now back to work getting this web site updated with the new releases and their audio samples which will go online later today..
Old 08-03-05, 08:01 AM
  #2  
DVD Talk Legend
 
AGuyNamedMike's Avatar
 
Join Date: Jul 2000
Location: (formerly known as Inglenook Hampendick) Fairbanks, Alaska!
Posts: 15,275
Likes: 0
Received 0 Likes on 0 Posts
This means when choosing a MAC computer, you must take the MAC's listed processor speed and times it times 2.25% to see what the Pentium IV equivalent would be.
"Times it"? Don't get me wrong, I think Mac hardware is superior to the vast majority of the intel/AMD clones out there, and with OS X, they finally have a real operating system. I'm just a bit wary of advice from a person who uses the word "times" in place of "multiply", and in a logically and mathematically incorrect sentence.
Old 08-03-05, 08:16 AM
  #3  
DVD Talk Legend
 
Join Date: Oct 1999
Location: |-|@><0r [email protected]|)
Posts: 17,214
Likes: 0
Received 0 Likes on 0 Posts
For this reason, MAC calculates their processor speed quite differently than PC's do. This means when choosing a MAC computer, you must take the MAC's listed processor speed and times it times 2.25% to see what the Pentium IV equivalent would be.

For example, if you see a MAC with a 1.45 Gig processor, that would be the equivalent to a Pentium IV 3.25 Gig processor. A MAC with a dual 2.75 Gig processor (like those on the new MAC G5's) would be equivalent to a Pentium 6.25 GIG processor - if Intel could even make anything that fast - which they can not yet.


First, 2.25% is not the same as 2.25, which is apparently what the author intended. Anyone who confuses this simple concept should not be in the business of providing technical advice - about anything.

Second, a dual-processor configuration is only fully utilized when both processors are completely busy 100% of the time. Consider:
  • For normal use - applications, web surfing - a single-processor machine is rarely busy 100% of the time - if you watch your process scheduler, you'll see that 80+% of its time is spend idle (wasted cycles.)

    A dual-processor configuration helps here when you have one CPU-intensive process, like encoding a movie or applying a digital filter in Photoshop, and a lot of background processes; one processor can handle the big task without getting interrupted by the others. But in normal practice, if one processor is idle 80+% of the time, what's the advantage of two processors idle 80+% of the time?

  • For gaming, the situation gets even worse. Gamers don't benefit at all from the second processor, because (1) the game engines are never written so that the work can be shared, and (2) even if they could be so written, a lot of the work has to be performed sequentially anyway, so it can't be shared. Essentially, during a game, the second processor sits around and handles background tasks. It doesn't contribute 100% to framerate speedup. Also, game performance in today's machines is much more closely tied to graphics processor performance and I/O pipelines (AGP8X, etc.), which have nothing to do with your processor.
Third, yes, lengthening a processing pipeline has self-limiting returns for performance, because conditional instructions ("if" statements) can require the CPU to throw away some of the pipeline work. This doesn't happen on every instruction, though. And for every branch, you have a 50/50 chance of predicting right, and those odds are much improved by things like branch prediction and monitoring the history of previous jumps for each instruction. In other words, these chips rarely have to do a full pipeline flush, which is why Intel implements such a long pipeline. Look - why would Intel build a 20-stage pipeline CPU if a 7-stage pipeline performed twice as fast? That's just silly.

Fourth, think about automobiles: do people rate them solely on 0-to-60 speed tests? No, that's just one factor in the overall performance of the car; you also have to look at shock absorption, cornering radius and torque, etc. Similarly, computer speeds are no longer compared apples-to-apples on megahertz. Instructuction set efficiency, power consumption considerations (Intel's M-series of processors), hyperthreading, L1/L2 cache size and performance, memory size and performance, I/O subsystems like AGP and SCSI and Firewire, the all-powerful graphics processor, the efficiency of your OS - even the quality of the compiler used to build your application contributes to how fast your computer performs.

So the goal here is: buy the computer that works the best for you. Your happiness with a machine is not likely to hinge on its MHz rating, regardless of how you calculate it. Do you like the OS? Can it run the applications you want? Is it cost-competitive for you? Etc.

- David Stein

Last edited by sfsdfd; 08-03-05 at 08:22 AM.
Old 08-03-05, 08:17 AM
  #4  
DVD Talk Legend
 
Join Date: Oct 1999
Location: |-|@><0r [email protected]|)
Posts: 17,214
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by AGuyNamedMike
"Times it"? Don't get me wrong, I think Mac hardware is superior to the vast majority of the intel/AMD clones out there, and with OS X, they finally have a real operating system. I'm just a bit wary of advice from a person who uses the word "times" in place of "multiply", and in a logically and mathematically incorrect sentence.
I noticed that, too - but the message is so substantively flawed that I didn't think it was necessary to focus on the kind of juvenile language mistakes.

- David Stein
Old 08-03-05, 08:40 AM
  #5  
DVD Talk Gold Edition
 
Join Date: Feb 1999
Location: HB, CA
Posts: 2,601
Likes: 0
Received 0 Likes on 0 Posts
The "megahertz myth" is the error of oversimplifying processor performance by simply looking at clock speed. However, his explanation makes a similar mistake by oversimplifying pipeline stages.

For example, it's true that longer pipelines incur a greater penalty in terms of clock cycles whenever the branch predictor guesses incorrectly. However, he's assuming that Intel and IBM's branch predictors are equally good. They're not. Intel's is much better, which means relatively fewer wrong predictions and thus fewer cycles lost.

Also, it's a bit misleading to look at the penalty purely in terms of cpu cycles. If a P4 takes twice as many cycles to refill its pipeline than a PPC, but it runs at twice the clock rate of the PPC; it would take the exact same time for both to refill their respective pipes (in actuality, the P4 takes more than twice the cycles to refill its pipes, but I'm using easy numbers here for clarity).

Anyway, I've never been a big fan of the "netburst" architecture, and I think even Intel is now realizing that approach is a dead end. However, while trying to educate people about how CPU performance is more complicated than just MHz is fine, it's a bit patronizing to try and replace it with something just as oversimplified.
Old 08-03-05, 09:05 AM
  #6  
DVD Talk Godfather
 
Michael Corvin's Avatar
 
Join Date: May 1999
Location: Louisville, KY
Posts: 58,108
Received 19 Likes on 11 Posts
The email aside, did anyone actually watch the video? It is actually very interesting. I watched it a couple years back.
Old 08-03-05, 09:10 AM
  #7  
DVD Talk Limited Edition
 
Join Date: Feb 2002
Location: On the penis chair
Posts: 5,169
Likes: 0
Received 0 Likes on 0 Posts
If Intel chip is much faster than Mac, then there would be no reason why Apple would want to switch their chip to Intel starting next year.
Old 08-03-05, 09:18 AM
  #8  
DVD Talk Gold Edition
 
Join Date: May 2002
Location: waiting for forum.dvdtalk.com ...
Posts: 2,755
Likes: 0
Received 0 Likes on 0 Posts
apple's forthcoming switch to intel renders this argument somewhat unimportant. i think apple has basically reduced its competition with microsoft to a battle of the OS. while apple will still charge a premium for their boxes, their overall price should come down somewhat with intel bulk pricing. customers will be left to choose between os x "jungle cat de jour" vs. windows vista (horrible name).

and i probably will still end up with both.
Old 08-03-05, 09:59 AM
  #9  
DVD Talk Godfather
Thread Starter
 
DVD Polizei's Avatar
 
Join Date: Jan 2002
Posts: 52,603
Received 1 Like on 1 Post
Yup. I emailed the guy back with a simple email:

"That was a lengthy explanation, but Apple is switching to Intel."

Old 08-03-05, 10:41 AM
  #10  
DVD Talk Legend
 
Join Date: Oct 1999
Location: |-|@><0r [email protected]|)
Posts: 17,214
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by kms_md
i think apple has basically reduced its competition with microsoft to a battle of the OS.
I think that any chip on the market these days will meet essentially all of a desktop user's needs for the foreseeable future. We've reached an interesting point in computing history - we have a surplus of computing power, memory, and storage, and now we have to figure out new uses in order to sell them new machines.

For about two decades, the holy grail has been a better GUI-style OS - nice visual styles, good performance of windowed applications, and strong ties to multimedia. Windows XP and OS X fully meet those needs. So... what next? Microsoft can't think of good ways to sell Longhorn, and even the ambitious features that might have had some value have been yanked.

So it looks like we're entering a period of technological stagnation on the OS front. The interesting developments are likely to happen at the application level - more interesting web technologies, improvements to email as a communications tool, etc.

- David Stein
Old 08-03-05, 01:00 PM
  #11  
DVD Talk Godfather
Thread Starter
 
DVD Polizei's Avatar
 
Join Date: Jan 2002
Posts: 52,603
Received 1 Like on 1 Post
Stagnation is a good thing in this case. Instead of racing from one OS to another, leaving severe gaps in security and features, maybe current engineers can create something that won't need to be upgraded for a very long time.
Old 08-03-05, 02:02 PM
  #12  
DVD Talk Legend
 
Join Date: Oct 1999
Location: |-|@><0r [email protected]|)
Posts: 17,214
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by DVD Polizei
Stagnation is a good thing in this case. Instead of racing from one OS to another, leaving severe gaps in security and features, maybe current engineers can create something that won't need to be upgraded for a very long time.
That would be great, but is unlikely. "Better security" is not a strong selling point - in fact, that's why it's so abominable today. Apple had it right by touting their Aqua skin.

Feature inconsistency is a very serious problem, but it's also pretty subtle and would require significant engineering to resolve.

- David Stein
Old 08-03-05, 04:38 PM
  #13  
Banned
 
Join Date: Sep 1999
Location: Memphis, TN
Posts: 11,544
Likes: 0
Received 0 Likes on 0 Posts
Along with all the other flaws in his argument, he also doesn't take advances like HyperThreading and dual cores into account, which can give the equivalent of 2 processors in one chip, in some (but not all) cases.

Furthermore, AMD processors also have a shorter pipeline, so you can't simply compare "mac" to "pc". You need to compare the various chips against one another on a more substanial level to be able to gauge speeds.

Furthermore, each chip has pluses and minuses. Some do certain things faster than others. Each has it's own advantages.

His basic argument (that megahertz is not a reliable indicator of speed) is correct. His reasons why that is true are badly flawed, and his numbers are just jokes.
Old 08-03-05, 05:09 PM
  #14  
Senior Member
 
Join Date: Oct 2003
Posts: 627
Likes: 0
Received 0 Likes on 0 Posts
Since OSX all of the cool kids have switched to macs. And by cool kids I mean nework engineers. It's much better for the work we do than XP.

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is On
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Thread Tools
Search this Thread

Archive - Advertising - Cookie Policy - Privacy Statement - Terms of Service - Do Not Sell My Personal Information

Copyright 2018 MH Sub I, LLC dba Internet Brands. All rights reserved. Use of this site indicates your consent to the Terms of Use.