Welcome to the MacNN Forums.

Eug Wanker · May 28, 2006, 11:40 PM

I had some time to run some benches: Core Duo 2.0 MacBook, vs. G5 2.0 iMac.

Basically:

Aperture web export: MacBook 2X as fast as iMac G5.
Cinebench CPU render: MacBook 2X as fast as iMac G5.
Cinebench HW lighting: MacBook 1/2 as fast as iMac G5.
Handbrake H.264 enc.: MacBook is 3.5X as fast as iMac G5.

PowerPoint web export: MacBook 1/3rd as fast as iMac G5.

I also ran Skidmarks GT. The iMac got 202 for the Altivec score. The MacBook got 24.

P.S. If you want the details, I posted it in the link in my sig. (Warning... It's a blog, but no there's no advertising.)

ashesindust · May 28, 2006, 11:53 PM

Thats pretty nice to hear also i find it great that the Macbook is going to support Windows Vista when it comes out.

greenamp · May 29, 2006, 12:24 AM

Neato. You should probably note that PowerPoint is running in emulation though on the MacBook.

Spliff · May 29, 2006, 01:04 AM

Don't Handbrake and Aperture use Altivec? And if so, shouldn't the numbers be much closer?

Spliff · May 29, 2006, 01:17 AM

Okay, I did some reading over at the HandBrake forums. One of the reasons that the PPC Macs are so slow with Handbrake is because the Intel Handbrake libraries are far more optimized than the PPC libraries. So your Intel Handbrake results probably have much less to do with the processor speed and much more to do with the amount of work done on the Intel Handbrake libraries.

In short, the PPC/Altivec x264 code is shite.

Simon · May 29, 2006, 02:58 AM

I observed that using HandBrake to convert DVDs to MPEG4 now runs at ~66fps on my 2GHz MBP when it used to run at ~20fps on my 1.67GHz PowerBook G4. I'm actually surprised the G5 didn't do better.

Eug Wanker · May 29, 2006, 11:03 AM

Originally Posted by Spliff

Okay, I did some reading over at the HandBrake forums. One of the reasons that the PPC Macs are so slow with Handbrake is because the Intel Handbrake libraries are far more optimized than the PPC libraries. So your Intel Handbrake results probably have much less to do with the processor speed and much more to do with the amount of work done on the Intel Handbrake libraries.

In short, the PPC/Altivec x264 code is shite.

That may be true, but whatever the case, I'm happy it's much faster on my MacBook. I had previously wondered if it would take longer to optimize Handbrake for Intel, because the PowerPC version has been around for much longer, but obviously that's not the case.

Originally Posted by Simon

I observed that using HandBrake to convert DVDs to MPEG4 now runs at ~66fps on my 2GHz MBP when it used to run at ~20fps on my 1.67GHz PowerBook G4. I'm actually surprised the G5 didn't do better.

Perhaps it's different for H.264. I note that my G4 iBook sucked much worse than my iMac G5, and more so than would be explained by clock-speed alone. The iMac G5 2.0 was almost twice as fast as my G4 iBook 1.33. (As you probably know, H.264 encoding is much, much slower than MPEG4 encoding.)

P.S. I finally posted my graphs so they can be embedded. Here they are:

Spliff · May 29, 2006, 01:19 PM

Eug,

Damn, those are impressive figures. Makes me want an Intel iMac. But I'll try to wait until CoreDuo2 and Leopard is shipping.

Eug Wanker · May 29, 2006, 01:32 PM

Originally Posted by Spliff

Damn, those are impressive figures. Makes me want an Intel iMac. But I'll try to wait until CoreDuo2 and Leopard is shipping.

Very reasonable.

Core 2 Duo is going to be very, very fast, and it's out soon. Hopefully the iMac will get it sooner rather than later.

It's plausible that 10.5 Leopard may be out by the end of the year, and if so, that's perfect timing for a Core 2 Duo 4 MB cache Leopard iMac for around Macworld 2007. I wonder if 802.11n will be available by then too. I'm may wait for 802.11n, HDCP, and a HD optical drive before I upgrade my iMac. Extra CPU performance is nice but at this point it's actually secondary to the above for me, and I definitely don't want to upgrade the desktop before Leopard and iLife '07 hit anyway.

P.S. Are you and Spliffdaddy related?

Spliff · May 29, 2006, 01:41 PM

Originally Posted by Eug Wanker

P.S. Are you and Spliffdaddy related?

Not at all. He's rabidly right-wing and I'm not. And I'm Canadian. I'm thinking that I need to change my username to eliminate confusion.

Simon · May 29, 2006, 01:53 PM

It really is impressive to see the MB blow away the iMac G5 like that. I'd recommend Eug's plots to those still asking why Apple made the switch to Intel.

stuffedmonkey · May 29, 2006, 04:37 PM

Eug - I am torn between getting a Macbook now and waiting untill August to get a new Conroe Based desktop when they are released... I guess what I want to ask is how happy are you with yours? I keep hearing a lot of wacked out stories, and it is hard to tell what is common.

andreas_g4 · May 29, 2006, 05:12 PM

Thx Eug. The Handbrake speed is outrageous, too bad I almost never need it… I really need something to tax my sweet new MacBook!

Eug Wanker · 02:58 AM

Originally Posted by stuffedmonkey

Eug - I am torn between getting a Macbook now and waiting untill August to get a new Conroe Based desktop when they are released... I guess what I want to ask is how happy are you with yours? I keep hearing a lot of wacked out stories, and it is hard to tell what is common.

Some people are convinced the iMac will get Merom. They may be right, but either way, Merom or Conroe, the new iMac will be very fast. The towers may get Conroe or Woodcrest.

I like my MacBook quite a bit. The keyboard is a bit of an improvement. Firmer feel, and it doesn't bow out at all like the iBook keyboard did sometimes. The screen is improved in contrast if you have nice lighting conditions, but I find the glare very distracting in rooms with strong ambient lighting.

Rosetta apps feel sluggish, but everything else just flies on the unit. I have 2 GB RAM though, so that helps.

P.S. It seems that even Apple apps can show humungous advantages on the MacBook vs G5 Macs:

However, unlike my Handbrake test, QT H.264 video encoding (iPod) is faster on G5 Macs... at least if it's a dual:

Other iMovie tests also show a G5 advantage:

Simon · 03:48 AM

I'd guess these iApps have been heavily optimized for Altivec. Since Altivec is history in the Apple world, I'm wondering if these numbers will improve when Apple starts optimizing for SSE3 (and hopefully soon for SSE4 on the Core 2).

Spliff · Jun 2, 2006, 03:46 AM

Eug,

Slightly off-topic, but what are your settings in Handbrake? I tried encoding a VIDEO_TS folder to h264 at 1000 kbps (1-pass) and I was disappointed with the resulting file (lots of blockiness and distortion around the edges).

Do I need to go higher than 1000 kbps?

Eug Wanker · 08:11 AM

Originally Posted by Simon

I'd guess these iApps have been heavily optimized for Altivec. Since Altivec is history in the Apple world, I'm wondering if these numbers will improve when Apple starts optimizing for SSE3 (and hopefully soon for SSE4 on the Core 2).

Dunno if this is true or not but I'm told that SSE3 isn't that helpful for a lot of stuff, which may be one reason Apple has been targeting SSE2, despite the fact that all Intel Macs fully support SSE3. I wonder how much Apple will target SSE4, but even if they don't any time soon, the good news is that Merom/Conroe/Woodcrest should be much faster per clock on SSE2 code as well.

Originally Posted by Spliff

Slightly off-topic, but what are your settings in Handbrake? I tried encoding a VIDEO_TS folder to h264 at 1000 kbps (1-pass) and I was disappointed with the resulting file (lots of blockiness and distortion around the edges).

Do I need to go higher than 1000 kbps?

Well, I'm no Handbrake expert, but I strictly use 2-pass encoding. I tried 1-pass a few times, but 2-pass was clearly better (not surprisingly). 1000 kbps 2-pass is still not perfect by any means though.

I tell Handbrake to do video deinterlacing, and I use the main (not baseline) profile. I usually let Handbrake set aspect ratio and cropping, but sometimes I may tweak that a little.

The bitrate really depends on the source material and the intended output size though. You can take the rez and frame rate and multiply it by a factor of say 0.15 to get reasonable results. I may use something like 0.18 because there is a little bit of an improvement sometimes.

eg:

640x480 x 29.97 fps x 0.15 = 1381017.6 ~ 1380 kbps
704x400 x 23.976 fps x 0.15 = 1012746.2 ~ 1010 kbps

Note the above numbers are for video bitrate only. Audio is extra.

ie. For widescreen DVDs, 1000 kbps 2-pass is not bad, but if you want to improve things you may want to bump the video bitrate 20-25% (or even more).

P.S. Here is a Photoshop bench to add to the list:

Normally it wouldn't be all that exciting, but the one interesting fact was that while the 1.5 GB iMac never paged out at all, and the 1.0 GB Cube paged out just a little bit, the 2.0 GB MacBook paged out like crazy. Rosetta's RAM hunger really is impressive.

Eug Wanker · 12:52 AM

Hmmm... Toast 7's encoding to DVD takes just about forever on the MacBook. Very, very slow, which I guess shouldn't be surprising, given that the encoder is running under Rosetta. My single-core G5 iMac is considerably faster, and seems to require only about 2/3rds of the memory that is needed to run Toast 7 on the MacBook. I guess the good news is that it's utilizing both cores on the MacBook, so it's not as slow as it could be.

Still, other than the encoding speed, you'd never know that Toast was running on a completely different CPU architecture. Toast's GUI is pretty responsive on the MacBook.

Spliff · Jun 4, 2006, 03:34 PM

Eug,

Have you compared ffmpegX's h264 encoding speed to Handbrake's on both your iMac G5 and MacBook? Someone over at the ffmpegX forum released a G5-optimized h264 library for ffmpegX that is supposed to increase encoding speed on the G5 by up to 30%.

I wish h264 encoding was faster on G5s. I hate having my iMac running at max CPU and at 70 degrees Celsius with the fans a full-tilt for a solid 16 hours or more. That can't be good for the iMac's longevity.

Eug Wanker · Jun 4, 2006, 03:46 PM

I haven't used ffmpegX much because Handbrake is easier for what I do. Thus, the speed difference doesn't really matter to me.

As for longevity, My iMac mobo has already been replaced so I look it as being almost brand new.

(I have the 3-year Apple Care, and I plan on replacing it with a Core 2 Duo iMac before the warranty runs out.)

BTW, my fans are never even close to full tilt, even at 100% CPU usage. They are of moderate loudness at full load, but if you want to hear full tilt, run the Apple Hardware Test, or else install OS 10.3.9 Panther on it. It sounds like a vacuum cleaner.

Spliff · Jun 4, 2006, 03:55 PM

Originally Posted by Eug Wanker

As for longevity, My iMac mobo has already been replaced so I look it as being almost brand new.

(I have the 3-year Apple Care, and I plan on replacing it with a Core 2 Duo iMac before the warranty runs out.)

BTW, my fans are never even close to full tilt, even at 100% CPU usage. They are of moderate loudness at full load, but if you want to hear full tilt, run the Apple Hardware Test, or else install OS 10.3.9 Panther on it. It sounds like a vacuum cleaner.

Yeah, you're right about the fans. I forgot how loud they sound when using AHT. Still, they run at 4500 RPM when under heavy load. Makes me nervous.

You have a Rev. B iMac G5, right? Same as mine. What happened to the mobo? I didn't get AppleCare, so I'm gambling.

Eug Wanker · 04:07 PM

I have the iMac G5 2.0 20" with Radeon 9600. Yes, that is the second iteration of the G5 iMac.

P.S. I was misremembering. It wasn't the mobo. It was the power supply.

I had used it for over half a year when I started getting weird glitches. I tested the RAM, and it was faulty, so I replaced it under warranty, but it turns out it wasn't the RAM. It was the power supply cutting out, and I think it must have taken out the RAM, because I had tested the RAM when I first got it and it was fine.

This is a well-known issue. (I had thought it was the mobo because that's another well-known issue.) See here. Even though it says it's only for 1.6 and 1.8 GHz iMacs, it isn't, and my iMac 2.0's serial fell into that serial number range too.

Maybe I got one of the last batch of flaky power supplies I dunno. (I ordered mine the first day it was announced.) In any case, a replacement power supply solved all my problems. Even 12 hours straight of Handbrake leaves me with a 100% happy iMac.

Spliff · Jun 4, 2006, 04:05 PM

Originally Posted by Eug Wanker

This is a well-known issue. (I had thought it was the mobo because that's another well-known issue.) See here. Even though it says it's only for 1.6 and 1.8 GHz iMacs, it isn't, and my iMac 2.0's serial fell into that serial number range too.

Maybe I got one of the last batch of flaky power supplies I dunno. (I ordered mine the first day it was announced.)

Hmm, my 2 Ghz iMac has a serial number in that range. So far no problems but I better read up on the symptoms of a flaky power supply.

snoopy199 · Jun 4, 2006, 08:21 PM

I'm curious to run the same aperture web export on my G5 vs. my MB.

How many photos did you export, I suspect the larger the population the more refined and accurate the testing will be. I won't have a chance to do it tonight but I'll run it tomorrow.

I expect the G5 to beat out the MB but who knows now. My G5 is a dual powermac with a X800xt so just about everything I do on my MB is done even faster on my G5. I don't think I've come across anything task thats faster on my MB but admittedly I have not scientifically test the it.

Eug Wanker · Jun 4, 2006, 08:28 PM

I just exported a webpage of 12 photos from my local food market: Toronto St. Lawrence Market

My G5 is an iMac so I'm not surprised it's slower, even despite the better GPU. I didn't have tons of heavy edits either, so the limiting factor in most of the photos was the RAW conversion.

P.S. I really hate how Aperture oversharpens everything on web export.

snoopy199 · Jun 4, 2006, 08:59 PM

I really don't use aperture to export to the web so I'm unfamilar with that aspect. I export to a jpg and upload those images to smugmug I recommend them highly. I used to host my pictures on .mac but apple's slow servers where hampering my work and others viewing them.

Because my G5 is a dual powermac and I have a powerful GPU, I really do expect my G5 to beat out the MB, but I'll wait and see. My home office is right next door to the nursery and I have 5 month old twins that I do not want to wake up - which is why I'll have to wait until tomorrow to run the test

snoopy199 · Jun 5, 2006, 08:08 PM

As promised my G5 was indeed faster but that was to be expected. I exported a web gallery containing 41 images.
G5 3 minutes
MB 4 minutes
My MB's temperature spike almost immediately to 75c and steadily climbed to 82c that was when the fans ramped up and kept the temps in check. Once the process completed the temps decreased back down to 54c where its sitting now as I surf.

Did your Mb see a quick spike and then climb into the 80s? How much memory do you have on yours, While I just upgraded to a gig, I'm wondering with aperture it may be alittle too meager.

mduell · Jun 5, 2006, 10:12 PM

Originally Posted by Spliff

Okay, I did some reading over at the HandBrake forums. One of the reasons that the PPC Macs are so slow with Handbrake is because the Intel Handbrake libraries are far more optimized than the PPC libraries. So your Intel Handbrake results probably have much less to do with the processor speed and much more to do with the amount of work done on the Intel Handbrake libraries.

I'm surprised the difference in the supporting libraries is that great. PowerPC with Altivec has been the optimization target for the last 7 years at Apple and the Handbrake devs.
Has Intel actually written libraries specificly for Handbrake, or is Handbrake just using some of the standard Intel math libraries?

Eug Wanker · Jun 5, 2006, 11:33 PM

Originally Posted by darth-vader000

As promised my G5 was indeed faster but that was to be expected. I exported a web gallery containing 41 images.
G5 3 minutes
MB 4 minutes
My MB's temperature spike almost immediately to 75c and steadily climbed to 82c that was when the fans ramped up and kept the temps in check. Once the process completed the temps decreased back down to 54c where its sitting now as I surf.

Did your Mb see a quick spike and then climb into the 80s? How much memory do you have on yours, While I just upgraded to a gig, I'm wondering with aperture it may be alittle too meager.

I have 2 GB RAM in the MacBook. Handbrake spiked the temp up to something like 88, and it dropped to 83-85.

What G5 do you have? Dual 2.3?

Yeah, your results make sense, as your Power Mac is a very powerful one, but still the MacBook's performance is impressive if it's getting 3/4 of the performance of that dual Power Mac with X800 XT.

snoopy199 · Jun 6, 2006, 05:55 AM

Originally Posted by Eug Wanker

IWhat G5 do you have? Dual 2.3?

Yeah, your results make sense, as your Power Mac is a very powerful one, but still the MacBook's performance is impressive if it's getting 3/4 of the performance of that dual Power Mac with X800 XT.

I have a revision b 2.0 GHz G5, I keep kicking around the idea of selling it. Its my main machine for photoshop and aperture and it excels at that. I don't really use if much for anything else since I have the macbook.

I'm not totally surprised by the web gallery export, yes I was expecting the G5 to beat it, but I did think the MB was going to hold its own. I don't think that specific task really leverages the GPU, its generating jpgs and then generating web pages. processor power and disk IO will be the limiting factor in that test imho.

Eug Wanker · Jun 6, 2006, 08:22 AM

Originally Posted by darth-vader000

I have a revision b 2.0 GHz G5, I keep kicking around the idea of selling it. Its my main machine for photoshop and aperture and it excels at that. I don't really use if much for anything else since I have the macbook.

I'm not totally surprised by the web gallery export, yes I was expecting the G5 to beat it, but I did think the MB was going to hold its own. I don't think that specific task really leverages the GPU, its generating jpgs and then generating web pages. processor power and disk IO will be the limiting factor in that test imho.

Disk speed is not a limiting factor. Processing power is very important, but if you've done any edits on the pix then GPU power is going to be a factor too.

snoopy199 · Jun 6, 2006, 08:25 AM

Oh I agree, in my day to day interaction with aperture, I find it much faster on the PM then my MB and most of that is because of two factors processing power and GPU power.