depends on what you do and how much ram you use…i use both, theres comparisons on the net somewhere, i think its about ghz for ghz between the two, so a dual may be closer than a single.
I don’t have any experience with the G5 or a 2.8GHz P4. However, I do have a Power Mac with dual 1GHz G4 processors from 2002 and a corporate PC with a single 2.4GHz P4 processor from 2003. For Photoshop 7 operations that completed mostly within RAM, I measured the following using a stopwatch:
1a) Image upsize to 720dpi
2xG4: 1.0; P4: 1.2
1b) Unsharp mask
2xG4: 2.0; P4: 2.5
1c) Magic wand
2xG4: 4.0; P4: 6.0
1d) Linear gradient
2xG4: 1.5; P4: 3.5
Reopen original image, then
2a) Image upsize to 1000dpi
2xG4: 4.0 to 2.5 P4: 3.0
2b) Unsharp mask
2xG4: 5.0 to 4.0; P4: 4.0
2c) Magic wand
2xG4: 7.0 to 8.0; P4: 13.0
2d) Linear gradient
2xG4: 6.0 to 3.0; P4: 26.0
Notes:
1) The 2xG4 was running Mac OS X Jaguar, had 1.5 GB RAM (up to 60% free RAM available to Photoshop) and dual 80 gig ATA66 hard disks. The P4 was running Windows XP Pro, had 512MB RAM (up to 85% free RAM available to Photoshop) and a single 20gig hard disk.
2) There was quite a bit of variation in the Mac timings, noticeable with longer operations, hence the range given in the results. Mostly the second time was faster, but not for 2c).
3) It is very difficult to compare performance like this. I deliberately chose operations to minimise the use of the hard disk because I don’t think my PC’s hard disk is particularly quick. For example, upsizing to 2000dpi took 7.5" on the 2xG4, and 1’38" on the P4, and unsharp mask on the same image took 28" on the Mac, and 2’36" on the P4.
4) I also have a PowerBook with a single 1GHz G4. It had the same amount of RAM as the PC but generally was a lot slower. The only operation that was significantly quicker was the linear gradient test. It also exhibited a wide variation in timings, particularly for unsharp mask.
My experience on Macs and PCs tells me to:
1) get dual processors
2) get a second, fast, hard disk
3) get lots of RAM
I suggest a G5 dual 1.8GHz with additional hard disk and RAM.
I would go top-of-the-line G5 Dual with a second internal SATA HD and 4GB RAM.
And, at this time of the year, I would wait until after MacWorld in July because newer and faster machines have a habit of being announced at that time.
If you wait until the end of June you might be able to get a 2.0 dual on sale.
Thanks a lot for your responses!
I am in no hurry and can wait a little longer.
Graham, thank you for your detailed feedback, this really helps me!
Graham,
Instead of using a manual stopwatch, try using the built in timing feature. It’s in the drop down menu at the bottom left of any document window. It’s much more accurate, especially on those short times.
Peter, I have found Photoshop’s timing feature to be inaccurate on the Mac (I haven’t checked the PC). Here are the results again, using Photoshop’s timer, and with stop watch measurement in parentheses (the stopwatch timings are of course dependent on my reaction time, so thay may be up to, say, 0.2s too long):
1a) 1.0 (1)
1b) 2.1 (2)
1c) 3.4 (4)
1d) 7.9 (1.5) (sic)
2a) 3.1 and 2.1 (4 and 2.5)
2b) 4.7 and 4.1 (5 and 4)
2c) 6.3 and 7.1 (7 and 8)
2d) 17.3 and 13.3 (6 and 3) (sic)
Notice the massive difference with the linear gradient test.
Yes Photoshops timing has always been inaccurate.
Photoshop’s timing is absolutely accurate in that it records the amount of time Photoshop spends doing things. But it omits the time it spends waiting for various OS services to complete, so it doesn’t provide a good real-world benchmark of how long it actually takes someone using the computer to complete real-world tasks you can’t continue editing an image until the screen has finished redrawing, for example.
You can use Photoshop’s built-in timing to make narrow synthetic comparisons between two machines. The results will be accurate. They just may not be terribly relevant to actual experience.
