Forums | MacLife

PowerBook G4 1.67GHz 1GB RAM took 133 seconds to render the "mendelbrot variation zoomed" setting.

Oh, by the way, very impressive upgrade from 1.0 to 2.0.2. I noticed that in 1.0, it took 50 seconds to render the default fractal, where now that takes 5 seconds. Well done!

For s***s and giggles, I'll post my results on my newly upgraded PowerPC 7500 266GHz G3 later.

EDIT: NVM on the 7500. My copy of Tiger isn't getting here soon enough.

Last edited by Shadowless (2006-01-14 11:46 am)

avkills wrote:

Hey Czach,

Did you do the FCP encodings on the Quad with Compressor or just with Quicktime?  Compressor is a ton faster and you get better results also.

I did it using the 'quicktime conversion' export.  I was not aware of compressor.  I am going to show my lack-of-pro savyness a little here, but fill me in about compressor. Maybe the next time I go up there, I will re-run the test with compressor.

avkills wrote:

Another question Czach, did the iMac Duo pretty much feel the same as the others... mousing and what not?

When I first walked up to theiMac Duo, I thought it felt a little slow when launching apps and the dashboard and stuff.  Mousing around and clicking in programs, especially the new iMovie '06 was snappy and fine - just like you would expect.  True, it only had 512 MB of ram and the performance launching apps and dashboard could be more of a function of RAM and HDs then CPU.  Maybe I should have walked over to an iMac G5 with 512 MB RAM to see if I got the same impression.  I am used to my 1.5 GB RAM at home and I had just walked from the quad with 2.5 GB, so it is entirely possible that my observation is a little biased.

Some explaination on the quad G5 performance in this thread sound spot-on accurate with what I just read in MacWorld this morning:

MacWorld Article wrote:

If you give just a quick glance at the Macworld Lab benchmark results for the Quad, you might wonder what the fuss is all about—on the Speedmark test suite, the new system barely managed to edge out the previous Mac performance champ, the 2.7GHz dual-processor model, and considering that the Quad has twice as much raw processing power as its single-chip, dual-core siblings, its lead over them is surprisingly modest.

A closer look at the results, however, shows why the Quad really is a big deal—for the markets it was designed to serve. On many of the real-world tasks that make up the Speedmark suite, the Quad only marginally outperformed the other models we tested; on six of the 15 tests, it even lagged slightly behind the 2.7GHz dual-processor system.

That’s because many of the tasks timed in Speedmark, such as starting up the system, depend more on other components—particularly the hard drive—than on the available processing engines; other Speedmark tests use applications, such as Apple’s own consumer-oriented iTunes and iMovie, that haven’t been optimized for multiprocessor systems. Even our Photoshop CS2 test, a set of 14 scripted tasks using a 50MB image, showed a speedup of less than 10 percent relative to the 2.7GHz dual-processor Power Mac, even though past testing has shown that Photoshop benefits greatly from dual-processor hardware.

A few of Macworld’s lab tests, however, show the stunning improvements we’d hoped for from the Quad. In our standard rendering test with Maxon’s Cinema 4DXL 9.1, the Quad needed only 37 seconds to handle a chore that took 63 seconds on the 2.7GHz dual-processor Mac, 71 seconds on the recently released 2.3GHz dual-core Power Mac, and 83.3 seconds on the new 2GHz dual-core model. In a separate test (not part of the Speedmark suite) that involves encoding video into MPEG-2 format with Apple’s Compressor utility, the Quad got the job done in just 3 minutes and 23 seconds, whereas the 2.7GHz dual-processor system needed 5 minutes and 12 seconds, the 2.3GHz dual-core required 5 minutes and 35 seconds, and the 2GHz dual-core model, at 6 minutes and 20 seconds, took almost twice as long as the Quad.

Basically, this is validating that the quad is not going to show much speed gain for apps that are not optimized to fully use the multi-processing power.  What's weird is that they are using the 4 processors, though.  Just not to their full ability.  For instance, when the CPU was at 25%, it was at 25% equally on all 4 processors (I watched it in acticity monitor), it was NOT at 100% on 1 processor.  It just seems weird that an app could split the load amongst the 4 processors, but not be "optimized" to get those processors moving at something more than 25% each.  I wanted to see those babies at 90% each and really see what the thing could do!!

MacBoy4139 wrote:

Compressor is included with DVD Studio Pro, Final Cut Pro, or Final Cut Studio.  It is a separate app that just compresses video into a variety of formats.

It gives better quality and faster encoding times than QT Pro even.

I'm curious as to whether or not it will be on that machine.  Having setup the first iMacs, they didn't come imaged with all of the necessary software - as it didn't exist as an image yet.  Store ops had yet to send it over.

Ahhh - that makes sense.  There is an option in my FCE at home that says 'export to compressor', but it is greyed out.  Probably because I don't have the necessary software.

wow.
not what i expected.

MacBoy4139 wrote:

dcalfine wrote:

avkills wrote:

If you want my 2¢ I think Apple should keep universal binaries for a long long time.  Consumers, Laptops and servers = Intel.  Pro Towers = PowerPC.

I agree entirely.  Of course, the best thing imaginable is if we had AMD in the game too for gaming Macs.  But other than that, Apple wouldn't have much use for them.  Maybe higher-end laptops.

Why servers Intel and not PowerPC?  Virginia Tech did choose Xserve because of the PowerPC.

Good point.  I was thinking web servers, for which I would expect the Xeon to be the best.  But for cluster nodes, yeah, the PPC970 is very good.  MACH V is currently the 15th fastest supercomputer in the world.  GO APPLE!

Czachorski wrote:

Basically, this is validating that the quad is not going to show much speed gain for apps that are not optimized to fully use the multi-processing power.  What's weird is that they are using the 4 processors, though.  Just not to their full ability.  For instance, when the CPU was at 25%, it was at 25% equally on all 4 processors (I watched it in acticity monitor), it was NOT at 100% on 1 processor.  It just seems weird that an app could split the load amongst the 4 processors, but not be "optimized" to get those processors moving at something more than 25% each.  I wanted to see those babies at 90% each and really see what the thing could do!!

I'm not sure the technical reason behind this but it was only 1 cpu thread, My dual core pc does the same thing 2 processors both a 50% on sigle threaded apps dual thread apps will max them both or just one to 100%

The iLife suite is a consumer app and will always lack behind the pro apps in both performance and features.

I agree with MacBoy4139  that years of optimizing for the PPC it may take a little time to fully optimize for the Core Duo.

Also Extreme Tech did a great artical on why Intel suck compaired to AMD, and one conclusion they came to was on the PC side some software simply wasnt optimized for the Intel chips and didnt use them to there full potential. This wont be an issue for Apple as there is only one brand chip to code for.

Also keep in mind the switch to intel wasnt just what the cpus could do now but years down the road, The G5's and G4's werent scaling very well. Also over all costs will be lower now that they share the same hardware so it help with profibility of the company. Which I like because I own Apple stock and I'm real happy right now.

jayster wrote:

Czachorski wrote:

Basically, this is validating that the quad is not going to show much speed gain for apps that are not optimized to fully use the multi-processing power.  What's weird is that they are using the 4 processors, though.  Just not to their full ability.  For instance, when the CPU was at 25%, it was at 25% equally on all 4 processors (I watched it in acticity monitor), it was NOT at 100% on 1 processor.  It just seems weird that an app could split the load amongst the 4 processors, but not be "optimized" to get those processors moving at something more than 25% each.  I wanted to see those babies at 90% each and really see what the thing could do!!

I'm not sure the technical reason behind this but it was only 1 cpu thread, My dual core pc does the same thing 2 processors both a 50% on sigle threaded apps dual thread apps will max them both or just one to 100%

The iLife suite is a consumer app and will always lack behind the pro apps in both performance and features.

I agree with MacBoy4139  that years of optimizing for the PPC it may take a little time to fully optimize for the Core Duo.

Also Extreme Tech did a great artical on why Intel suck compaired to AMD, and one conclusion they came to was on the PC side some software simply wasnt optimized for the Intel chips and didnt use them to there full potential. This wont be an issue for Apple as there is only one brand chip to code for.

Also keep in mind the switch to intel wasnt just what the cpus could do now but years down the road, The G5's and G4's werent scaling very well. Also over all costs will be lower now that they share the same hardware so it help with profibility of the company. Which I like because I own Apple stock and I'm real happy right now.

iMovie had 24 threads going.  This was done with someone else on a test to see if a G5 Quad was really worth it.

The end result that we had to come up with?  Apple bastardizes consumer stuff.

MacBoy4139 wrote:

iMovie had 24 threads going.  This was done with someone else on a test to see if a G5 Quad was really worth it.

The end result that we had to come up with?  Apple bastardizes consumer stuff.

That was me. 

I am willing buy into the theory that Apple cripples consumer apps - the data supports it.  And that plain sucks.  I am an in between consumer and pro level, and I am caught in a crappy pinch.  I would like the speed boosts from the high-end machines even for my consumer uses.  I may not ever make the leap to the full-pro software, because I am a pro at something else (my engineering career) and I don't have time to learn the pro software, but I CAN afford the pro hardware and then save even more time.  Very frustratring.

Oh well.    It just means that I will probably just stay with my current machine for a few more years, rather than upgrading.  Why is Apple's marketshare so low again?

dj phat 2000 wrote:

Has that Fractal Bench program been universalized yet?  I'd like to see what the G/Mflops are of the Core Duo is.  I'm almost willing to bet that the G5 is still faster.

The problem with that one is it isn't universal, so it will be emulated on the Core Duo. That's why I recommended Fractographer: it may not tell you the Mflops, but it will give you a good benchmark with raw processor speed, (as long as you're not choked on memory, but 512 MB should be enough) and it's also universal, so it will be native on both systems.

Last edited by akb825 (2006-01-14 1:52 pm)

pcguy wrote:

This is a bit off topic, but what will be interesting is how will Apple "speed bump" their computer from now on? will they continue their tradition of announcement during show, or will they do what the rest of wintel world do, upgrade cpu/product line as newer hw come down from their supplier!

Good question.  We aren't in the dark anymore about what can happen!  It is just a matter of knowing when.

pcguy wrote:

One question I have is why didn't Apple use the 1.83/2ghz duo cpu on the MacBook, 2k is very high end(price wise) compare to what other intel based machine charge, and other company such as Dell is already listing 2ghz laptop part.

Not all Intel Core Duos are the same.  There is a performance variety (used in the iMacs) that use more power and have higher performance & MHz ratings.

There is a lower powered version, which is made for laptops and really really small computers (i.e. Mac mini?)

pcguy wrote:

This is a bit off topic, but what will be interesting is how will Apple "speed bump" their computer from now on? will they continue their tradition of announcement during show, or will they do what the rest of wintel world do, upgrade cpu/product line as newer hw come down from their supplier!

Apple have introduced simple speed bumps in between keynotes before, so I guess that won't change much. Keynote hardware presentations were usually for more than simple CPU speed increases.

And some hardware advancements were public before Apple incorporated them before, too, through Motorola/Freescale/IBM announcements. I hardly think the 970MP came as a surprise to anyone when Apple introduced the dual core PowerMacs.

,xtG
.tsooJ

smeghead wrote:

jayster wrote:

I agree with MacBoy4139  that years of optimizing for the PPC it may take a little time to fully optimize for the Core Duo.

But OPENSTEP was made for x86 and Sun...

OpenStep, originally NeXTSTEP, was made on, Idunno, probably MIPS. It first shipped (commercially) on M68K in the black NeXT hardware, then later (version 3 and up) was also offered for x86, SPARC and PA-RISC.

Apple had to do some tricky stuff to make it run on PPC.

They did? Being based on the Mach microkernel, NeXTSTEP, and thereby OS X, is highly portable. None of this stuff is new to the OS; fat (universal) binaries were available back then, too, using the same mechanism they use now. The potential to quickly port Mac OS X to just about any architecture has always been there.

If the x86 OS X has been under development as long as the PPC version, and especially since they've been working closely with Intel this past year, wouldn't the x86 coders know the hot optimization tricks by now?

You'd think they would, won't you? Then again, I doubt they've been putting much more than a token "make sure it compiles and runs on x86" effort into the parallel development most of the time.

,xtG
.tsooJ