Digital Media | Spotlight: Photography | Inside Aperture | Blogs
I feel the need ... The need for speed
Some of the younger readers might not recognize the title of this post as a line from the 1980's movie Top Gun, and others of us barely remember it because it was so long ago :)
But this post is about speed. If you are like most photographers you are happier behind the camera than you are behind a keyboard. So, when you are processing images on your computer you are usually taking time away from shooting. It doesn't matter if you are a pro or an amateur, I'm sure you'd rather be shooting than editing. Like a lot of photographers I have investing in computer hardware and software that allows me to get in, do my job and get back out again.
But, the question that always seems to come up is; how much faster is one computer than another? Magazines and websites do tests of processors, RAM, graphics cards under a variety of situations and these results are often interesting and informative but don't always tell me how much faster one system will be than another when I'm doing the type of real-world work that I do.
So I thought I'd do a real world Aperture test on 3 different machines and see what the results are.
For the test I copied the exact same Aperture library on to all three machines. The test I performed was to select 42 images from a project and export them out of Aperture to JPG format (as you might if you were sending sample images to a client). The images in the project were all Canon RAW files from an EOS 30D and were aproximately 8MB each. The export settings were same for all test, Export to JPEG and fit within 1024x1024 pixels, use current version name, and save to a folder on the desktop.
The machines tested were as follows:
1) Power PC, Dual Processor 2.0Ghz G5, 3.5GB RAM, Radeon X800 XT with 256MB RAM
2) MacBook Pro, 2.33Ghz Core Duo (2 processors), 2GB RAM, Radeon Mobility X1600 with 256MB RAM
3) MacPro, 2.66Ghz (4 Xeon processors), 2GB RAM, Radeon X1900 with 512MB RAM
So here are the results of the test:
The Power PC G5 machine took 3 minutes and 31 seconds to complete the task
The MacBook Pro (laptop) took 3 minutes and 14 seconds to complete the task
The MacPro took 1 minute and 25 seconds to complete the task
So the bottom line is that my old G5 machine takes about 2 1/2 (2.48)times longer to complete the test task than my new MacPro and my MacBookPro laptop takes about 2 1/4 (2.28) times as long as the MacPro.
It should be noted here that the Aperture library on the MacBook Pro was on an external USB 2.0 drive (Lacie 250GB drive) because it will not fit on the laptop's drive. This fact may have cost the laptop some time on the test, but I'm not sure ... the Lacie drive might be just as fast as the internal 5400rpm drive? All that being said, I think it's fair to test the laptop with an external drive becasue that's the way I actually use it. As I said, I wanted this to be a "real world" test.
So, it seems that the old saying is true .... you do get what you pay for when you invest in an new MacPro, this is truly an amazing machine when coupled with Aperture. The MacBookPro on the other hand is not much faster than my old Dual G5 machine, but it is portable and that's really worth something when you are on location or traveling.
Until next time,
Keep shooting.
Allen Rockwell
Allen Rockwell Photography
Comments (10)
I think the fact that the Macbook Pro is a little faster than a dual G5 says a lot of the laptop! Not too long ago (oh about a year), the G5 in your test was considered one of the best machines out there for the job.
Clearly you were only interested in testing what would give you the fastest results, and not necessarily do a comparison. But I am curious as to how the Mac Pro would perform using the same external drive as in the laptop (connected externally that is).
Thanks for reporting your findings. You do realize that you are not helping our budget right? ;)
While I am no expert at either, I am a Computer Programmer starting to learn photography (my wife has a great eye). So my background is in Computers. I'd have to say that your test here is very usefull for you, and anyone else who has the exact same equipment and work style as you.
Factors that will affect the performance of Aperture: Disk RPM, Memory, Graphics Card processor speed and memory. Physical disk configuration (RAID?), # of processors, location of library vs. export location. (reading/writing to/from the same disk is going to perform very poorly). With the external drive... USB1.1, USB2.0, Firewire 400, Firewire 800, as well as the rotational speed of the drive in the enclosure will all significantly affect the performance.
To speed up aperture, I'd recommend investing in multiple high-performing disks (firewire 800 if you need to go external), extra RAM, and the best video card you can afford.
Performance is not an easy black-and-white issue and is affected by many factors. I agree with Allen... "you get what you pay for" is very true, but by putting your money towards a few key components, you can optimize your aperture performance.
Please remember that the maximum USB2.0 speed on a MBP under Mac OS X 10.4 is more than half of what it should be. You should have better used a FW400 (or even better FW800) drive. Especially with a test which is quite dependent on hd performance.
Yes, I know that my USB external hard drive is slow and that's why the MacBookPro numbers suffered. But as I said, I wanted to get some real numbers based on the way I work, with the hardware I work with. I'm still impressed that my laptop is faster than the Dual G5 machine that has served me so well for all these years.
I really don't care if this kind of tasks takes 2 or 5 minutes at it can be scripted and launched while I'm away. And you can even push your jpgs to a FTP within the same script.
I lose much more time waiting for the "real-time" adjustments to be performed/rendered during an editing session, this is where the real speed penalty is IMHO. There a MacPro performs decently, at best, while a Dual or Quad G5 is really lagging behind.
This roughly corresponds to my gut feel for the performance of my two primary Aperture systems. I've got a 17" Core Duo MBP and a Dual 2.5GHz G5 PowerMac with a X800. And, for the MBP, I've run my libraries on the internal disk, an external FW800 disk, and a External SATA disk (more on this at some point). And, without doing obvious metrics, they run at about the same pace with each other in almost every disk configuration. Since you're running a MBP vs a 2x2.0G5, that seems fairly consistent with your findings.
All that to say, dammit, I should get a Mac Pro sometimes soon, shouldn't I? And dammit, you just provided me with more of a reason to do it. (grin)
Neat test. It appears that the fact that the Mac Pro is a Quad with 512MB of vram gave it the significant edge. I have a dual 2.5Ghz G5 with 4GB of ram and a ATI 9800XT with 256MB of vram driving a 19" LaCie CRT and a 20" Apple Cinema Display. Although I find Aperture to be reasonably fast with this set up, I am willing to bet that if I got another dedicated graphics card with 256MB of vram to drive one of the monitors, my performance would improve tremendously.
Nice test, I'm running a G5 Dual Core 2.0 with 1.5gbs of ram and the basic GeForce 6600LE and 2 320gb Seagate 7200rpm SATA drives with a Lacie 250gb Firewire 400 drive as a backup, Aperture runs OK (got much slower with the recent updates though) but the rotate tool is almost unusuable
So I'm thinking upgrading my graphics card to the Radeon 1900 G5 Edition (available at Newegg or Apple) and maybe tossing in another gig or two of ram will fix it so i can actually use the rotate tool, can anyone confirm that?
right..
MacPro: 10.62Ghz, 4x2.66 (minus some because it cant use it all.)
PowerMac: 4.0 Ghz, 2x2.0Ghz
so 10.62/4= 2.66... which is about the amount of times it is faster.. so what did you prove here?
how about adding a PowerMac Quad G5 to the equation?
have you tried removing half the ram from the powermac and see how much that slows it down?
... ..
153957
153957,
Are you offering to send me a Quad G5? I'll be happy to run the test again after I receive it :)