interesting!

yaaahoo

1st post

yaaahoo

I don't think so!

Math Benchmarks

Since OSX Beta was introduced 1.5 years ago, I have been seeking math benchmarks -- integer (INT) and double precision floating point (FP) on various Mac systems. I would like to see G3 vs. G4 and OS9 vs. OSX, at a minimum. Whenever I've asked one of the Mac speed testing sites, all I get is more Photoshop and games. I believe Apple recently admitted that Sines (///) are twice as slow on OSX vs. OS9...

That's it?

I thought the Pentium 4 was faster than that. I mean, it has the power to burn CD's (literally) and create MP3's. That to me says enough!




Blue Man Group = traitors

Good!

Now maybe Apple will get their @$$ in gear.

That's it?

I thought the Pentium 4 was faster than that. I mean, it has the power to burn CD's (literally) and create MP3's. That to me says enough!




Blue Man Group = traitors

That's it?

I thought the Pentium 4 was faster than that. I mean, it has the power to burn CD's (literally) and create MP3's. That to me says enough!




Blue Man Group = traitors

Tripple Post?

Don't use the Back Button!

And I'm pretty sure these tests are flawed

Let's put it this way..

I buy a lot of macs and PCs for work, typically, the only time I see a newer mac out perform or out bench a P4 or Athlon is when Apple is doing the testing.

I doubt these guys tests were the best tests I the world... however they are just telling me something I already know. Apple's hardware and OS are not the fastest things on earth... for damn sure.

Who's surprised by this?

Not me. Apple should sort the whole AIM mess out once and for all.

yawn...

boring...

back to work, a*******.

...so what

Who cares, as long as the computer does what you need? I've got a dual Athlon MP 1600+ and a G4/867 happily coexisting here. I use the PC for 3D, and the Mac for After Effects and FCP. Both are fine machines, and besides my 3D software isn't available for Mac whereas Final Cut Pro isn't available for PC, so the comparison is moot.

Somehow

I am not convinced. I believe I read that the test were made for a Pentium III. How can you use the same test on a totally different processor? I thought these tests were supposed to be specialized. Also I think I read somewhere that Apple has never been a member of SPEC anyway so that there are no tests for Apple's processors.

Where is the standard anyway? I mean it seems like with all the contradictory testing going on its like you just pick which test you would like to believe and go with that. Until there is an actual "standard" that everybody agrees upon all this stuff is bunk --good or bad.

Not real world

Luckily this is not real world comparisons. Of course a SPEC test is better on a P4 than any other system. The SPEC is so small that it can all fit in the P4's tiny cache thus eliminating memory latency (a HUGE problem in the cheap P4 systems that use PC133). More importantly, you can never suffer a branch mis-predict if you only ever branch backwards (looping) so this doesn't expose the Achilles heel of the P4. By far the most important point is that this is NOT VPU OPTIMIZED! It would be absurd to not use AltiVec to do real-world FPU operations! This is why everything that is FP-intensive tends to be vector-parallelized with AltiVec instructions.

I bet that the compiler they used can compile for SSE or MMX. If the Pentium gets to use its pathetic VPU why can't we use our outstanding one? I bet if you pre-compile the SPEC test with Vast/C-AltiVec we would smoke them!

Another pathetic test trying to uphold the out-dated status-quo. I wonder if they ever discuss how much mileage the PPC gets on Motorola's CPU R&D budget which is dwarfed by Intel's. Don't you think that is more of an issue? Image how amazing our chips would be now if we could concentrate on good/new designs rather than kludging old ones.

But I guess that is seeing too big of a picture,
Jeff.

AltiVec

The good performance of a PPC is heavily dependent on use of AltiVec. I have some experience with this myself. Take a look at here for some details:
http://floyd.liunet.edu/~gkhanna/autovp.html

For these tests they did not use any AltiVec enabled application.

Original article

Since I couldn't reach the article on The Register, I found the original C't article. The English version is here (http://www.heise.de/ct/english/02/05/182/). The tests were not of the floating point unit per se, but of floating point operations associated with the C++ and fortran compilers, which are used for scientific applications rather than typical consumer or commercial applications. The article implies that a better compiler would improve the numbers.

Real World Tests

I own a Dual 800mhz G4 and an 867 G4 and I recently had the pleasure to sit down with a brand new Dell 1.9ghz P4.

1) It's ugly...there new design looks worse than my a**
2) In real world use such as sitting there opening and closing things its god damn slow. and I'm being serious, for example, running software update and then trying to use your machine is not possible...so whether it beats out the mac or not, at least my mac is always useable at the same speed whenever I want!!! MULTITASKIN WOO! I measure speed in useability and Apple wins hands down.

Remember ByteMarks?

I remember several years ago when apple was touting some new processor or the like, trying to convince people that MHz didn't matter (which it doesn't insofar are trying to compare two chips with the same MHZ number), Apple touted their chip being faster than the current Intel offerings using ByteMarks (At least I think it was ByteMarks). And all the Intel folks jumped up and down and said they were unfair tests for like 16 different reasons ranging from compiler to internal code to optimizations to the color of the tester's eyes.

There was also no mention in the Register article about whether the SPEC benchmarks use the SMD instruction set on the Pentiums (simialr to the Altivec stuff on the G4s). I'm also not too sure on this "old" version of gcc they used. Since OS X and the BSD layer are fairly new to the G4 platform, isn't it more than likely that the compiler they used would be slow (lacking code optimazations and perhaps including such things as debug info and the like)?

And if you also happen to remember the hoopla over some ATI vs. nVidia benchmarks, the ATI card was seemingly optimized to do well in the Quake benchmark.

Hot Chip

My chip is hotter than your chip!

Nener nener neeeeeneeerr.

(really it's hot, really really hot. I mean red glowing hot. Hot hot hot. fire an egg on the heat sink hot)

This is great news!!!

I'm glad the G4 gets toasted by the P4. I never liked this chip since its basically a G3 with an improved floating point unit and Altivec. Now, I only hope that these results are enough of a reason for Apple to get the G5 ready.

- Mark

Honesty

The G4 is designed for embedded and telecoms work, as are the more recent IBM G3s. As a result, they are designed with very different performance characteristics in mind than desktop and workstation processors. That is a fact. So, Apple either needs to take a substantial part of desktop PPC development in-house. There are indications that they might already have done so. If they have not done so, than some form of drastic action is needed.

If the latest Moto rumours are to be believed, all we have to look forward to in the next 18 months are modestly improved G4s. Faster clockspeeds (up to 1.6ghz), 266mhz FSB, 4mb L3, and 512k on-die L2 this year, and more Mhz (perhaps even 2ghz…Intel will be at 3ghz+ with SMT and other substantial enhancements by then) and RapidIO (an exciting, if late development) nest year. This is not nearly enough.

Unless the "G5" is close at hand (and has the Itanium-like specs some rumours have suggested), ACTION IS NEEDED. Apple needs to somehow accelerate this G4 schedule (unlikely), take control of some aspects of PPC development (maybe leverage some IBM workstation PPC technologies from the wickedly fast Power3 and Power4) or at least introduce SMP (even with cheap SMP-altered G3s) throughout the line. They certainly need to have motherboards with the performance and features expected of PCs (DDR, ATA/133, USB 2.0, etc.), as Mac motherboards are 2 years+ out of date. The last big upgrade was the Yosemite board of 1999, which in upgraded form still is the basis of the newest G4s.

Apple needs to do SOMETHING to bring Mac hardware performance back up to the industry norm. Heavily AltiVec optimised niche apps like FCP3 can still impress, but AltiVec-oriented benchmarks dangerously mask glaring integer, FPU, and day-to-day performance lags. OS X is a performance laggard, and this too must be swiftly addressed.

I proudly advocate, use, and will continue to use the Mac. The (perceived) performance of my decked-out 400mhz Yosemite can still compete with, or best most PCs. However, we are falling behind as a platform in terms of performance.

We can't pretend that PCs do not exist, that all comparisons are unfair, and lie to ourselves. Apple needs to bring performance, and price/performance back up to where it deserves to be.

My Yugo

My Yugo can beat your BMW hands down. Sure, you'll have to race without tires, but it'll be a race and I'll win!!
Testing a DP G4 with a benchmark that ignores AltiVec and the second processer is the same level of BS.

Look at the chipspeeds on the upcoming Itanium and you will see that chipspeed is only useful for comparing processors in the same line.

And as far as benchmarks go, comparing cross-platform apps is all that really matters. Any company (including Apple) can define benchmarks to make their chip better.

Your homework (once again)
Read/ Watch
http://www.apple.com/g4/myth/

what does it prove?

Proves that the P3 is GREAT at this SPEC benchmark.
Whoopee flipping doo.

Now, can I go and get some work done? On my G3.

I swear, idiots abound.

Fools. The day OS X and Final Cut Pro and iApps are available on a x86 is the only day I'll give a damn about benchmarks. Asses.

Oh, and have a good day.

Silly

I agree across the board that what really counts in *most* applications is productivity, not speed, and the Mac wins hands down there, regardless of processor architecture. Making a comparison based on an unoptimized benchmark is about as close to worthless as it gets.

Besides, although I'm not pretending that the PPC (NOT per clock) is faster in most areas--Barefeats, among others, has shown that it's competitive with a P4 but so so up against an Athlon--these results are just silly. I always look to the distributed.net RC5 project as an illustration of potential scientific speed (as always, it only applies to specific circumstances, of course); their client is optimized thoroughly to perform the same calculation on a variety of architectures, and the G4 whups every i86 chip by a wide margin.

I'm not at all anxious to get away from the PPC architecture, since it seems to offer a lot more with a lot less tradeoffs. But, who knows.

selling Macs

I train guys who sell Macs, and you know what? The speed doesn't enter into it anymore. Nobody cares.

We show the customer what you can DO with a Mac (like the iApps) and what you CAN'T do (like get viruses), and they buy one. It's that simple.

Selling Macs

I train guys who sell Macs, and you know what? The speed doesn't enter into it anymore. Nobody cares.

We show the customer what you can DO with a Mac (like the iApps) and what you CAN'T do (like get viruses), and they buy one. It's that simple.

A little info for yall ..

1) SPECmark is a SYSTEM benchmark, not a CPU benchmark, as claimed.
2) SPECmarks are highly compiler dependant
3) They used a dual G4 with one processor disabled. But this doesn't eliminate the overhead of a multiprocessor system.
4) They compared an optimized Fortran compiler vs a non-optimized C++ compiler.
5) They were testing double precision FP's. Altivec is irrelevant.
6) The P3 box was running linux (likely will all the bells and whistles killed). The Mac was running OS X (with all bells and whistles still active).
7) Just like video drivers being optimized for Quake tests, there's much in the Intel worked optimized for SPECmarks.

Essentially, the test is meaningless. Unfortunately, the anti-mac bigots/idiots will trump it to actually mean something.

Ok, ok, but

Ok, i agree that this test is meaningless and in no way suited to show the PPC goods/flaws.

But this test still tells us something. The compilers available are not very useful in doing scientific computing for the people interested, i.e. g77 or f2c. Apple/Motorola should definitely do something about it, if they want to reach the higher education market.

new os, old core

The Apple can draw on a few crumbs of comfort. The SPEC benchmarks disable a second CPU if present, and don't necessarily take advantage of the Altivec and Screaming Sindy dedicated FP instructions - that's down to the compiler. C't noted that it wasn't using the latest gcc compilers on the G4, but then again, it had shunned the latest Intel compilers too, which give a 20 per cent speed hike. And C't records that G4 performance is scaling better than the clock frequency, compared to the Pentium, with a 18.6:15.5 against the Pentium's 7:8. But given that that the Pentium is currently 2.2Ghz, and the G4 1Ghz, that will take a very long time indeed. (We've already disclosed how Apple's forthcoming chips use a larger 4MB cache, which is a much better way of getting there from here).
problem is, one former Apple engineer told us, in serializing the twenty five year old BSD layer with the fifteen year old code of the extensions NeXT began to add in the mid 1980s.

Apple's attack parrots unfailingly point out that Mac OS X is "the most modern OS" you can use. In marketing terms, perhaps that's true. In technical terms, the integration point has been described to us as so crufty that no one wants to touch it. This is very old code indeed, and explains why you see the spinning disc cursor so often in OS X. Serialization is one of the toughest problems for an engineer to solve.

You idiots believe this?

Sigh, we hate it when the rest of the industry can't stay in character.
Look, people, we're the soap opera, right? So when an allegedly technically
astute German computer magazine starts publicly lambasting Apple for calling
the Power Mac G4 a "supercomputer" because its testing using the industry
standard SPEC benchmarks reveals that, for floating point operations, a 1
GHz G4 processor is a total dog on quaaludes compared to even a relatively
ancient 1 GHz Pentium III, our role is to bask in the drama, wring our
hands, and milk it for all it's worth and then some. Meanwhile, it's up to
the "real" technical press to notice and report that said German magazine's
methodology had more holes in it than a colander made out of Swiss cheese.
In this manner, you, the viewer, would get both an entertaining dose of
hysterical melodrama and a serious technical grounding that would let you
enjoy our hysterics while realizing you'd have no reason to run screaming
into traffic. See? It's all about demarcation.

Unfortunately, as faithful viewer Timothy Ritchey pointed out, it looks like
everyone else out there forgot their lines. Even The Register, normally way
better about this sort of thing, surprised us by running a headline of
"Benchmarks demolish Apple speed boasts"-- sheesh. So now, much as we hate
to do it, we're forced to backtrack and mention a couple of obvious reasons
why c't's benchmarking results aren't really a reason to start drafting a
suicide note.

First and foremost, c't itself admitted that the G4 should have mopped the
floor with "the x86 FPU with its antiquated stack structure and eight
registers only"-- so why, when the G4 was shown to be half as fast as the
Pentium III, did the magazine just say "gee, we guess the G4's no
supercomputer" and then saunter away, hands in pockets, whistling a jolly
tune? Doesn't anyone think it's strange that they failed to mention that the
SPEC2000 test, as compiled, utterly ignores the G4's Velocity Engine
registers, which are what gives that chip its supercomputer-class,
greater-than-gigaflop floating point performance? What c't did is tantamount
to forcing you to write with your toes and then telling you that your
handwriting sucks.

What's more, while the industry just loves SPEC benchmarks, faithful viewer
Mark Davis reminds us that they've always been biased towards Intel
processors, in part because the SPEC code just floods the chip with a
constant stream of perfect instructions and let it work at peak efficiency,
which is nothing like how real software is processed. As you may recall from
Jon Rubinstein's "Megahertz Myth" spiel, Intel's recent chips take a speed
hit from the recurring need to clear and refill those extra-long pipe

Above from Appleturns.com

I submitted the *Facts* & Jack Miller in his usual fashion, made it entertaining....formatting got screwed though....yikes!

http://www.appleturns.com/

The MON edition should be up all tuesday ;-) See Reuns for previous posts

Factual Info is:

The G4 "Should have mopped the floor with the P III's ancient 80 register design"
Well, "DUHHH"! But you idiots DISABLED the Altivec unit - the 32 128 bit registers which accelerate Floating Point ( to Supercomputer Status 15 Gigaflops on a Dual GHz G4 Sustained!)
So, how could those "New 32 registers" DISABLED! be of any help??? Altivec also accelerates Integer instructions.....and this "Synthetic Benchmark" which Intel is ALWAYS #1 in, even when say an AMD chip is actually faster points to "Benchmark Tweaking the Pentium by an entire Intel TEAM"!

Also, this test "Implies" any CPU has ONLY an Integer unit & an FP Unit. In the case of the G4 80% of the chip die and 90% of the performance is disabled.


The GCC 2.9.5 compiler was used....strange, when Apple helped OPEN SOURCE GCC Compiler version 3.0 to automatically optimize G4 code ( wait till you see THIS effect on OS X 10.2 err whatever they call the BIG MAY UPGRADE!
MOT has focused on "Parallel instruction Altivec" solutions, which can now excecute 16 instructions per clock cycle! That's in JUST Altivec, not including Int & FP!!!
-----------------------------------------

Some "Moron" implied here ( and Register article there ) that the G4 Photoshop bake off is "Fixed". Even G4 optimized but NOT P4 optimized. That is NOT TRUE!
Apple currently shows SSE 2 & "Velocity Engine" optimized Photoshop. A GOOD Benchmark, because it is carefully optimized by Adobe for BOTH Platforms. Those results show even the slowest SINGLE processor Mac is a LOT faster than a 2GHZ P4

Apple claims, correctly:
"The dual 1GHz Power Mac G4 is an astonishing 72 percent faster than the fastest PC on the market with a 2GHz Pentium 4 processor. Fact is, even the single processor 933MHz and 800MHz Power Mac G4 systems outperform the Pentium 4 by 51 percent and 26 percent, respectively.
http://www.apple.com/powermac/specs.html

---------------------------------------------------
They did NOT want a fair test! Fair is Darwin Intel vs. Darwin for PPC - optimized for BOTH running whatever they run. Linux for intel is HIGHLY Optimized! Perhaps Linux PPC vs x86 Linux? Free BSD for PPC & X86, etc.
Even in Mac OS 9?

It was a set up. They disabled Altivec even in GCC 2.9.5 because the Pentium would have lost, and that was NOT the point was it? Yet, they did NOT disable MMX/SSE 2 in the Pentium. Sound fair?

In reality, this is more like a G3 running Intel code ( Who runs two different code bases? C++ for PPC & FORTRAN for Intel??? How about something people USE! Metrowerks, Apple's Project Builder...like shipping products use? Or just the recent GCC?
---------------------------------------------
Anyway, you who believed it....pathetic! Those who don't care because a Mac is more productive? Buck up a little, we also have the most efficient CP

Statistics

Bench marks are just like statistics. Sorry if you've heard this one, but there are...

Lies.
Damn Lies.
And Statistics.

It always depends on the context and the conditions. They are all fast, but fast isn't the whole story. Since I can't get a Mac with a P4 in it, I just don't care.

gcc 3.0

this gcc compiler is new to me, but sounds interesting (finally!)

I think at least these articles made a fair point, that compliers available are not very well optimised and that number crunching performance has almost been disregarded on OS X.

And even in the non-scientific arena - 4 years after altivec was available (and they were talking it up well before then), very few programs / compilers actually make any use of it - why????

Apple must make up for these shortfalls, so please stop apologising for them.

Re: gcc 3.0

You never heard of GCC 3.0???? You live in a cave, or what! (?)

You "think these articles make a point that compilers are not very well optimied & that number crunching performance has been almost disregarded on Mac OS X"?

Well, I guess you are pretty uninformed then! Metrowerks has not only a small checkbox for G4 optimie, but a new scheduler for the MPC 7450-55 ( as well as older 74xx G4 chips). Metrowerks is the most popular development software for Mac & Windows in the world!
Then adding G4 optimiations to GCC 2.9 - 3.0 by Apple engineers & GIVING it away FREE to Open SOURCE is extra "Giving Back" to the BSD, Open Source, Darwin, community.

OS X has not graphics card support for the OS, and YES, the code is NOT optimized ( on the version 10.1.3 shipping ) as well as it can (is now) and will be.
--------------------------------------------------
"And even in the non-scientific arena - 4 years after altivec was available (and they were talking it up well before then), very few programs / compilers actually make any use of it - why???? "

Talking up Altivec before the first G4 shipped??? The VERY FIRST G4 was announced 9/19/ 1999! In September of THIS YEAR it will be 3 years old! ( Hoe do you figure Altivec was available OVER 4 years ago?? That 4th Q 1999 Motorola only delivered to Apple 80,000 CPU's instead of the 350,000 it promised....so you couldn't even get one until Jan 2000.

NON- Scientific? PhotoShop, OpenGL ( & ALL Games Using it on Mac OS X & 8/9 Soundjam MP, FCP, All Audio applications like Logic Audio, Cubase 5, Nuendo, Peak DV, MOTU Digital Performer 3.x, etc. etc. Cool FX Plug ins like AltiVerb ( Real-time convolution Reverb- ONLY possible on a G4 or a $15,000.00 Sony unit with 48 DSP CHIPS @ 1/2 the performance.....)
I do Audio & I can't find a non-G4 optimized program! How about MPEG or Sorenson Encoding? Speeding up the process of MPEG-2 to 30x encoding via G4 & Real time Video FX Rendering in FCP 3. ( Besides Graphics/Audio/Video/Science who else NEEDS serious acceleration?)
MOST Parts of OS X are G4 optimied, more coming soon.

Compilers with G4 optimization include:

Project Builder - Apple free
GCC - Open source FREE Compiler
Absoft Fortran Toolsets - Absoft
Metrowerks' CodeWarrior -
Macintosh Common Lisp (MCL) - FREE
Object Tools Eiffel - Metrowerks plug in
REAL BASIC
FUTURE BASIC

( Listed here:
http://developer.apple.com/tools/compilers.html )

JAVA G4 optimized Compilers:

Metrowerks CW 7.0
Visaj Personal Edition
AgentSheets
Jake - smith resources
Spin - ZAT
Simplicity for Java
BBEdit
SpotCheck - Genie Works
VOODOO Server -Uni Software Plus -MW Plug In
DashO-Pro - PreEmptive Solutions
Mac PLAF Bundle for Swing -Luca Lutterotti
MacJikes - Fastest Java co

Re: gcc 3.0 cont...

Fancy-API2Mac - Greg Guerin
GL4Java - OpenGL
BrowserLauncher - FREE
InstallAnywhere Now! - Zero G Software
Java Development Kit 1.1 (JDK) - JavaSoft
( Here are JAVA G4 optimized compilers )
http://developer.apple.com/java/javaintro/tools.html

---------------------------------------------------
Hey, can you see this list? You would be HARD PRESSED TO avoid using a non-G4 optimied compiler! I cannot find ANY Compiler for Mac OS 8/9 or OS X which does not do G4 optimizations, anywhere, ever the FREE OPEN SOURCE Compilers for Linux PPC, Darwin, BSD, etc.
So WHAT THE h*** ARE YOU TALKING ABOUT!!!???

-Meanwhile, I'm getting back to work on assembly code & G4 specific optimizations ( and Graphics Card support for G Force 2MX, G Force 4MX (54 Billion Instructions/sec!) & G Force Ti ( 1.23 Trillion/sec ) to handle all the GUI overhead for Mac OS X 10.2 ( or whatever they call it.)

If software YOU use, is not G4 optimized & you want it to be, ask the developers how they managed to avoid using a G4 optimied compiler! ( even though I.E. & Word are Integer based and can't benefit much from AltiVec - it can also speed Integer! So.....LAZY DEVELOPERS who haven't updated CODE or software since Jan 2000 is the ONLY answer for ANY non-G4 optimized software. ( at least a 50% improvement across ANY software program!)

Of course in May, anything running on OS X will be optimized for G4 anyway.....so, one way or another....
---------------------------------------------------

I'm not defending Apple, I just can't believe you people are so clueless, and cannot see TROLL BAIT - and the most rigged benchmark on earth.....and call it some TRUTH you always knew. LOL!

One more thing, I saw an iBook @ Comp USA 500MHz G3. It said next to the CPU:
500MH G3 ~ 1 GHz P III

Right ON the display for Price/ features/ spec's in Comp USA!

Sincerely,
Armas

P.S. Get your facts straight! If Altivec was NOT disabled, while SSE on P III was enabled, the P III would have been squashed like a BUG!

for fun, try ufc-crypt

benchmark statistics

Note that the top two ranks are currently held by a Dual-1Ghtz G4 running MacOSX 10.1.2 -- over an Athlon 1700+ running FreeBSD 4.5....

Have fun, kids :-)

More Benchmarks

http://www.dvformat.com/2002/11_nov/reviews/cw_macvspciii2.htm

It compares a Pentium 4 3.06 Ghz with a dual G4 in adobe photoshop and stuff, are the benchmarks accurate?