Intel iMacs really that much faster than PPC?

[macphisto]

Okay, so I was mulling this idea and let me know what you think.

Apples to apples (no pun intended...well, maybe a little) it seems to me that the Intel chips are on par horsepower-wise with with current G5 PPC chip iMacs. The only benefit that you get with the Intel iMac is the fact that it is a dual processor iMac with a slightly faster chipset and a few other bells and whistles.

Am I thinking correctly with this?

[Dont Hurt Me]

No

[crazzyeddie]

Usually, a Pentium M (single processor mobile chip) would perform better per clock than a G4, which performs better per clock than a G5. Since the Core Duo is two chips, it should perform even better... but only benchmarks will show for certain.

[bugfaceuk]

Well you know, those stats were largely bull shoe.

They were integer and float operation through put, and the base line for equivalence would be 2x faster. I mean you would hope that with two cores, you could do approaching 2x as much (yes yes, approaching). Of course, if it (or the CPU intensive operation) is single threaded, then we might expect similiar or possibly slower performance with 2 cores.

What is clear that a "very good at multi-tasking" OS like OS-X will benefit greatly from dual-core CPUs to become a "very very good at multi-tasking" OS. I have a dual core windows machine and a dual core linux machine, I noticed the extra core on Linux, on XP... not so much.

[Little Endian]

http://www-128.ibm.com/developerworks/library/pa-fpf970mp/

According to SPEC rates the 970MP and Intel Duo are more or less equal clock for clock.

A single 2.5Ghz 970MP Dual core chip gets 32.3 on integer and 42.8 on Floating Point.

A single Intel Duo Dual core chip @ 2.0Ghz gets 32.6 on interger but only 27.1 on Floating Point. SPEC scores don't translate into real world performance though but I think the G5 will perform better clock for clock and core for core until more Universal binary Apps are launched and better optimizations for the intel platform are made. Remember Apple's benches are the usual twists and they are putting Intel Duo performance with it's best foot forward, we will have to wait for more detailed benchmarks.


http://www.anandtech.com/mac/showdoc.aspx?i=2520

Anandtech also goes into detail about PowerPC vs. X86 based raw perfromance.

[Morn]

A 2.4ghz dual cpu opteron
SPECint_rate2000 37,5
SPECfp_rate2000 34,4

A single dual core 2.5ghz G5
SPECint_rate2000 32.3
SPECfp_rate2000 42.8

3.2ghz Pentium D
SPECint_rate2000 32,6
SPECfp_rate2000 32,5

Intel Core Duo 2ghz
SPECint_rate2000 32.6
SPECfp_rate2000 27.1

To put it in perspective with other CPU's around today. (And yes the 2ghz G5 and 1.67G4 are a lot slower than all of these). And it's the opteron that tends to win in most real world tests.

Here's what the original powermac 2ghz G5 gets.
SPECint_rate2000 17,2
SPECfp_rate2000 15.7

[maya]

Is it my imagination or did no one notice that Steve Jobs compared a single core iMac G5 to a dual core x86 iMac. :confused:

If that were the case, then yes the Core Duo would outperform a G5. :p

Maybe I have to re-read and watch the keynote again, to make sure. :o

[Morn]

What other imac would he compare it to?:confused:

[maya]

What other imac would he compare it to?:confused:


Well he is comparing a dual core to a single core processor. Plus the Core Duo has 2M L2 cache, while the G5 only has 512k cache. Any processor with a 1M or larger cache will see better performance gains. ;)

If Jobs was going to compare he should have done it with a Core Solo or something. This was just sneaky, as usual.

[MacRumorUser]

Exactly.... Why would he compare it to a dual G5 wheh it's not in the imac?

Same as the macbookpro compared to the G4 powerbook..

We will see what they do with the powermac's towards summer/fall I guess. I'm hoping 2x duo core all around... That would give the performace boost all around on the powermac models..

[Morn]

The powermacs won't have Core Duo.
http://en.wikipedia.org/wiki/Intel_Next_Generation_Microarchitecture
Conroe will be in the powermac most likely. Which will be necessary as the Core Duo is not faster than the 2.5ghz dual core G5. And we can pretty much guess from intel's estimated launch date that it's going to be out after the wwdc.

[AtHomeBoy_2000]

Exactly.... Why would he compare it to a dual G5 wheh it's not in the imac?

Same as the macbookpro compared to the G4 powerbook..
I totally agree. A lot of people forget about that part.

[rot@ti.org]

Apple's web site has some more numbers comparing the 1.83 dual core MacBook with the 1.67 PowerBook:

3D rendering 4.1X faster
Gaming 2.2X faster
HTML load 1.9X faster

So not everything is 4 to 5 times faster. I wonder how much of the speed increase is due to the faster bus.

See http://www.apple.com/macbookpro/graphics.html for above numbers.

[Koodauw]

What I would like to see is how much faster is it for everyday apps. Floating point and integer operations mean very little to me. Does it help on the load time of apps? Does Photoshop render faster? Will safari be faster? Will it perform better with 8 apps open? 10-12? than its PPC counterpart? These are true tests of speed if you ask me.

[Morn]

Gaming? That'll be like 80% due to the faster graphics chip. Games don't yet take advantage of multiple cores.
Koodauw, yes, if the apps are native. Though loading time is hard drive speed dependant. I think that safari and photoshop (when it's universal) will be much faster. Multitasking performance shouldn't change that much, it's mainly about ram.

[maya]

Apple's web site has some more numbers comparing the 1.83 dual core MacBook with the 1.67 PowerBook:

3D rendering 4.1X faster
Gaming 2.2X faster
HTML load 1.9X faster

So not everything is 4 to 5 times faster. I wonder how much of the speed increase is due to the faster bus.

See http://www.apple.com/macbookpro/graphics.html for above numbers.

You are also comparing 5 year old technology to current technology. The G4 has been out since 1999+, while the Core Duo has been out since late 2005.

Of course the Core Duo will outperform on the notebook side, and also the desktop side, as the G5 was released mid 2003.

One has to remember that you cannot compare directly a G5 or G4 with a Core Duo unless you have a dual core G4 or G5 in the same cat.

Sure the Core Duo does more work per watt, however that is expected as its considered new technology. Its all marketing. ;)

[andiwm2003]

..........................Sure the Core Duo does more work per watt, however that is expected as its considered new technology. Its all marketing. ;)

no.

the only relevant measure is how much faster my apps are on the new mac compared to the old mac. if the new mac has 2,4,8 cores and the old mac has only 1 core doesn't matter at all.

theoretical comparisons of a non existant single core 1mb cache g5 powermac to a nonexistent intel core duo single 1mb cache are a waste of time.

performance of the apps is important to the user not techical justifications why the old mac is slower. who cares?

[Flowbee]

You are also comparing 5 year old technology to current technology.

And yet, until this week, Apple was selling the G4 Powerbooks as "current" technology.

[Morn]

The Core Duo is based on the Pentium Pro which was out in 1995. Although it's been tweaked an awful lot since then. :) The G4 is just a G3 with an added altivec unit. And the G5 is very similar with tweaks and additions, like an extra FPU unit.
The newest microarchitectures around today, the Athlon. And the Pentium 4 which Intel is giving up. New is not necessarily better.
And the G5 is faster than the Core Duo. But the G4 was even slower than the Pentium M, Freescale was not able to keep up with Intel.

[neocell]

no.

the only relevant measure is how much faster my apps are on the new mac compared to the old mac. if the new mac has 2,4,8 cores and the old mac has only 1 core doesn't matter at all.

theoretical comparisons of a non existant single core 1mb cache g5 powermac to a nonexistent intel core duo single 1mb cache are a waste of time.

performance of the apps is important to the user not techical justifications why the old mac is slower. who cares?
Completely agree. Who the hell cares about hypothetical situations/configurations. Open the box, boot it up and which one is faster. Don't care what's inside, just how well does it work and how fast does it go. It's perfectly fair to compare the PowerBook to the MacBook and the intel iMac to the old one. What are they replacing? The old ones. What are you doing? Buying a new one and using it? Crap, how could anyone bitch about this. Is it faster, yes or no, if so how much. Period.

[Meyvn]

It's also fairly significant that the G5 is 64-bit, and this first generation of Core Duos are not.

[lssmit02]

It's also fairly significant that the G5 is 64-bit, and this first generation of Core Duos are not.
Yeah, I was wondering about that. Most apps don't benefit from the G5 being a 64 bit processor, in OS X, though,

Myth #5:
Myth: My application will have much faster performance if it is a “native” 64-bit application.
Fact: This is true for some other architectures because the number of registers and the width of registers changes between 32-bit and 64-bit mode. However, the PowerPC architecture does not have either of these limitations. It was designed for 64-bit computing from the beginning, and supports 64-bit arithmetic instructions in 32-bit mode. Thus, on PowerPC architectures, software does not generally become faster (and may actually slow down) when compiled as a 64-bit executable. According to Apple (http://developer.apple.com/documentation/Darwin/Conceptual/64bitPorting/index.html link)

But the same might not be true with a 32-bit Core Duo.

[Meyvn]

I never bought the 64-bit on 64-bit is a huge difference over 32-bit on 32-bit anyway. What I was arguing is: what happens when significant applications begin to be released in only 64-bit format? What happens when significant features in future OS releases and applications (which is more probable than simply completely disabling 32-bit support any time soon) are only available for 64-bit computers? THAT is the real concern.

[Morn]

It's also fairly significant that the G5 is 64-bit, and this first generation of Core Duos are not.

64bit is mostly hype, it doesn't make much different in 90% of apps

[Catfish_Man]

The Core Duo is based on the Pentium Pro which was out in 1995. Although it's been tweaked an awful lot since then. :) The G4 is just a G3 with an added altivec unit. And the G5 is very similar with tweaks and additions, like an extra FPU unit.
The newest microarchitectures around today, the Athlon. And the Pentium 4 which Intel is giving up. New is not necessarily better.
And the G5 is faster than the Core Duo. But the G4 was even slower than the Pentium M, Freescale was not able to keep up with Intel.

The G5 is almost totally unrelated to the G4. All it shares is the instruction set (including Altivec). Different number and types of execution units, different pipeline lengths, cacheline size, out of order depth, number of rename registers, dispatch width, bus type and speed, basically everything. The G5 is descended from the POWER4 line, not the G3 line.

One thing I've been wondering is this:
Why does the 970MP (dual core G5) score SO much better on SPECcpu (which is single threaded) than the 970FX does?

I believe that the key point is that it used version 8 of IBM's compiler, rather than version 6 as the 970FX did. XlC 8 can automatically multithread and vectorize programs in certain situations, so what claims to be a single threaded non-vector benchmark may have ended up being a dual threaded vector benchmark, at least to a certain extent, allowing the 970MP to use some of its second core to help the first core out.

For reference:
2.5GHz 970MP SPECint: 1428
2.5GHz 970MP SPECfp: 2076
2.2GHz 970FX SPECint:1040
2.2GHz 970FX SPECfp: 1241
2.26GHz Pentium-M SPECint: 1839
2.26GHz Pentium-M SPECfp: 1375

note: Pentium-M here refers to Dothan, which is the predecessor to the Core Duo (Yonah).

Sources:
http://www.aceshardware.com/SPECmine/
http://www.heise.de/ct/05/24/018/
http://www.llnl.gov/asci/platforms/bluegene/papers/10mendell.pdf

<edit>
in response to the 64 bit posts: the main point of 64 bit is to allow for more than 4GB of ram.
</edit>

[Nermal]

What I was arguing is: what happens when significant applications begin to be released in only 64-bit format?

Developers will make their software compatible with whatever is available. If Apple's systems are all going 32-bit, then the apps will be 32-bit too.

[eXan]

G4, which performs better per clock than a G5

Maybe true, but only at lower clock (1.6 ghz). I saw benchmark comparing 1.9 ghz iMac G5 with 1.92 GHz iMac G4 (upgrade of course ;)) and G5 won hands down.

[BlackDan]

Considering a dual CPU system only has a 30% speed gain on a single CPU system (in general) and seeing Apple claims speed doubled, I'd say the intel's are about 30%-40% faster per core than the G5's. (or even faster considering the frontside bus has slowed down)

But then again, benchmarks have NEVER been reliable or representative of real-world performance. (e.g. you could design a machine that has 10x performance over a G5 on SPEC_INT, but is extremely slow in other areas, so it's real-world performance may be in fact slower than the G5). The same goes for the perfomance per watt Jobs showed in his keynote. How do they define this performance and how is it measured?

Though I am sure the Intel is indeed faster and uses less power it's very hard to say HOW MUCH faster it is.

[ijimk]

so on the new intel macs can you put windows on them as well? like for playing an occasional non-mac game?:confused:

[Nermal]

You should be able to in the future, once someone makes a loader or virtualiser.

[Little Endian]

Apple's web site has some more numbers comparing the 1.83 dual core MacBook with the 1.67 PowerBook:

3D rendering 4.1X faster
Gaming 2.2X faster
HTML load 1.9X faster

So not everything is 4 to 5 times faster. I wonder how much of the speed increase is due to the faster bus.

See http://www.apple.com/macbookpro/graphics.html for above numbers.

Comparing Dual Core to single core is bad enough but it is also interesting that Apple Chose to use the 1.83Ghz core Duo against the 1.67Ghz G4. I think it would have been a little more reasonable to compare a 1.67 Core Duo to the 1.67 G4 both entry level 15 inch PowerBook and MacBook Pro are clocked at 1.67. Comparing 1.83 Duo to the G4 gives a clock speed gain of almost 10% over the G4 besides the Duo having two cores, 4X the Level 2 Cache, faster memory, faster GPU, and faster Bus.

If you think about it the speed gain that the Duo has over the G4 while seemingly impressive should be expected given the improvements. I was thinking about it though and Intel Duo does not seem to perform as well as I had hoped even with Apple's best foot forward benches. Let me explain below.

3D rendering (Modo) 4.1X faster= 300% gain
Gaming (Doom 3) 2.2X faster= 120% faster:
Garage Band 2.1x faster= 110% same with Final Cut Pro.
Translating performance gains into percentages looks less impressive than saying 2x and 4x faster etc. but that's why apple uses those terms much better for marketing.

First off both 3D rendering and gaming gains could easily have 50-100% of the gains brought on by the Faster GPU especially depending on what Graphics setting were used for Doom III. Settings are very important on Doom III performance especially Resolution and FSAA performance which can vary wildly from GPU to GPU. It is also important to point out that while DOOM 3 is not SMP aware Modo is which could explain another 30-70% gain in performance for Modo. Modo also utilizes the GPU for rendering, I really don't know to what extent but the faster GPU in the MacBook Pro helps for sure.

Garage Band and Final Cut are also SMP aware which would explain at least half of the 110% performance gains they have over the G4. Take that all into account and Clock for Clock core for Core the Intel Duo only seems to be have perhaps a 20-50% performance gain which is nice but one would expect more with 10% faster Clock speed, faster memory, 4x FSB speed, and 4X more cache.

I wonder if the second core in the Intel Duo chip can be disabled with CHUD tools like the 970MP? If possible disable the second core in a 1.67Ghz Duo and compare to a 1.67Ghz Powerbook doing so would eliminate the gains from SMP and should prove quite interesting.

Even more interesting would have been comparing Intel Duo performance to the PowerPC (7448) G4 instead of a 7447A. Too bad Apple did not choose to use the 7448 in the last Power Book Update. The 7448 G4 could have given clock speeds as high as 2Ghz or so plus Double the L2 cache up from 512k to 1Megabyte, faster FSB up from 167Mhz to 200Mhz, better Altivec performance, and even lower power consumption than the current G4. I think Apple most likely chose not to use 7448 because it would have slimmed down the performance gains that Intel Duo could show. The 7448 despite what some people believe as being a white paper chip is actually in production.

http://www.powerlogix.com/press/releases/2006/060110.html

oh well enough with my rant it's all speculation anyhow we will see for sure once detailed benchmarks of real world performance can be done.

[Little Endian]

Considering a dual CPU system only has a 30% speed gain on a single CPU system (in general) and seeing Apple claims speed doubled, I'd say the intel's are about 30%-40% faster per core than the G5's. (or even faster considering the frontside bus has slowed down)

But then again, benchmarks have NEVER been reliable or representative of real-world performance. (e.g. you could design a machine that has 10x performance over a G5 on SPEC_INT, but is extremely slow in other areas, so it's real-world performance may be in fact slower than the G5). The same goes for the perfomance per watt Jobs showed in his keynote. How do they define this performance and how is it measured?

Though I am sure the Intel is indeed faster and uses less power it's very hard to say HOW MUCH faster it is.

I disagree that Multiple CPU machines only have a 30% (in general) speed gain. 30% is about right given an app that is poorly optimized for Mult-threading. Heck Apps that are not even Multiprocessor aware can see gains as much as 10% just because OS background tasks are being run on the other processor or cores. Generally I would say Apps that are multi threaded can see gains of about 50% to even close to 100% in best case scenarios on a Multiple CPU or core system. In garageband on my Quad G5 I can utilize up to 250% of 4 cores same with idvd MPEG encoding. Quicktime Encoding Utilitzes as much as 350% of my available CPU power. I also had a Dual 2.5 and saw an average of 50% speed gain in real world performance with SMP apps.

As far as real world performance goes.

http://barefeats.com/quad02.html

http://barefeats.com/imacg52.html

[punkmac]

Completely agree. Who the hell cares about hypothetical situations/configurations. Open the box, boot it up and which one is faster. Don't care what's inside, just how well does it work and how fast does it go. It's perfectly fair to compare the PowerBook to the MacBook and the intel iMac to the old one. What are they replacing? The old ones. What are you doing? Buying a new one and using it? Crap, how could anyone bitch about this. Is it faster, yes or no, if so how much. Period.


Totally right on!

[BlackDan]

Totally right on!

Well, it's 0x faster for word processing, 0x faster for filling in numbers in Excel, 0x faster for surfing the web, etc. etc.

Satisfied? ;)

[tivoboy]

Well, it's 0x faster for word processing, 0x faster for filling in numbers in Excel, 0x faster for surfing the web, etc. etc.

Satisfied? ;)

I think this is actually, probably not accurate. It is most likely even with Rosetta going to be faster using office and as well, faster on page display and processing while browsing.

But, I do get your point.

[Little Endian]

Originally Posted by neocell
Completely agree. Who the hell cares about hypothetical situations/configurations. Open the box, boot it up and which one is faster. Don't care what's inside, just how well does it work and how fast does it go. It's perfectly fair to compare the PowerBook to the MacBook and the intel iMac to the old one. What are they replacing? The old ones. What are you doing? Buying a new one and using it? Crap, how could anyone bitch about this. Is it faster, yes or no, if so how much. Period.

Totally right on!

Who Cares? People who are more technically inclined perhaps (Hardware Junkies) You are posting in the Hardware section of MacRumors you know. People who care about where there money goes? People who care if it makes sense to spend $1500 or $2500 on a new Computer that Apple Claims to be 2-5 times faster? when it may be much less than that depending on your software and needs.

I understand that most of the things in this thread are speculation and hypothetical situations and configurations however if you don't like it don't read it. This site is called Macrumors and most of the content is just that "speculation".

Perhaps you should go to the ipod forums and go ranting about how why should anyone care how a case looks like. Hell if the case fits and looks like it will protect who cares right?

[jimN]

I wonder if the second core in the Intel Duo chip can be disabled with CHUD tools like the 970MP? If possible disable the second core in a 1.67Ghz Duo and compare to a 1.67Ghz Powerbook doing so would eliminate the gains from SMP and should prove quite interesting.

Even more interesting would have been comparing Intel Duo performance to the PowerPC (7448) G4 instead of a 7447A. Too bad Apple did not choose to use the 7448 in the last Power Book Update. The 7448 G4 could have given clock speeds as high as 2Ghz or so plus Double the L2 cache up from 512k to 1Megabyte, faster FSB up from 167Mhz to 200Mhz, better Altivec performance, and even lower power consumption than the current G4. I think Apple most likely chose not to use 7448 because it would have slimmed down the performance gains that Intel Duo could show. The 7448 despite what some people believe as being a white paper chip is actually in production..

Why compare it with only one core running when you are buying a machine to have both cores. That's like me comparing myself to an olympic sprinter and then saying he has to hop to make it fair. Sometimes one thing has an advantage over an other, why cripple it to make the comparison. Similarly why compare it to a moto chip we never got. You heard the rumours, they couldn't get the 7448 into the powerbook - something that probably helped bring forward thne switch to january. No sense comparing a horse to a unicorn, one doesn't exist.

[whooleytoo]

Infoworld have pretty much dismissed Apple's SPEC results - on the basis that the specs chosen are primarily to benchmark multi-processor systems, and thus generate a good many threads. This immediately skews the tests in favour of a multi-processor (or multi-core) machine.

I.e. the benchmarks accurately rate the Intel Macs, but make the PPC Macs worse to make the comparison more favourable.

[Chip NoVaMac]

Well he is comparing a dual core to a single core processor. Plus the Core Duo has 2M L2 cache, while the G5 only has 512k cache. Any processor with a 1M or larger cache will see better performance gains. ;)

If Jobs was going to compare he should have done it with a Core Solo or something. This was just sneaky, as usual.


Not sneaky at all IMO. He was comparing the old model to the new and improved model. Both at the same price points. More bang for the buck. Go Apple.

[Little Endian]

Why compare it with only one core running when you are buying a machine to have both cores. That's like me comparing myself to an olympic sprinter and then saying he has to hop to make it fair. Sometimes one thing has an advantage over an other, why cripple it to make the comparison. Similarly why compare it to a moto chip we never got. You heard the rumours, they couldn't get the 7448 into the powerbook - something that probably helped bring forward thne switch to january. No sense comparing a horse to a unicorn, one doesn't exist.

True true, both Machines are what they are and the MacBook Pro is overall superior but disabling one core would give a better picture of how Applications that are not SMP aware would perform. More importantly it would give a clearer picture of clock for clock core for core performance which would give a better picture of how well the Universal Binary Apps run on X86 vs. the PowerPC machine it is replacing. As far as comparing "MacBook Pro to a non-existant product" that is out of curiosity I would rather had Apple give us one more PPC update if it meant real world better performance for some Applications and possibly better battery life. I say possibly better battery life because Apple still has not Published that info yet.

Performance is important to me since I own a Quad Core G5 and I really want to know if spending the $2500+ for a MacBook pro would give me a viable alternative to purchase as a second Machine to use as sort of a Desktop replacement as well as travel notebook. If performance is not nearly as good as Apple states then I'm better of getting the imac Intel Duo and purchasing a Powerbook G4 refurb or on closeout.