I wondered why was Adobe emphasizing so much the availability of speed boosts for Photoshop CS4 available through a 64bit version and special video drivers associated with various graphics chips ; particularly for the PC version of Photoshop. So I decided to do some quick benchmarks on my Vista and Windows XP laptops using CS3 and CS4 versions. And as Paul Harvey would say, “now you know the rest of the story”.
The Vista machine has Dual Core Centrino 2 2.0GHz processors, 4GB of Memory, 320GB 5400 RPM disk drive which has 70% free space, and an ATI Radeon card with 512MB of dedicated memory and Vista Home Premium with SP1.
The WinXP machine has Dual Core Centrino 1.8GHz processors, 1GB of Memory, 100GB of 5400 RPM disk drive which has 30% free space, and an Intel Video card with 64MB of shared memory and Windows XP Media Center edition with SP2.
Comparison of PhotoShop CS4 and CS3 on Windows Vista versus Win XP
|
|||
Task | Vista CS4 | Vista CS3 | Win XP CS3 |
Load 12MPixel Image |
3 sec
|
5 sec
|
2 sec
|
Smart Blur -30, 25, High |
20 sec
|
23 sec
|
16 sec
|
Surface Blur – 9, 15 |
10 sec
|
11 sec
|
7 sec
|
Filter Gallery Start |
4 sec
|
4 sec
|
2 sec
|
Cutout Filter |
2 sec
|
1 sec
|
< 1 sec
|
Curves adjustment |
2 sec
|
1 sec
|
< 1 sec
|
Exposure adjustment |
15 sec
|
14 sec
|
8 sec
|
Selected area Gaussian Blur |
1 sec
|
1 sec
|
1 sec
|
Save As TIFF-LZW format |
3 sec
|
4 sec
|
3 sec
|
Merge HDR phase 1 |
65 sec
|
61 sec
|
45 sec
|
Merge HDR phase 2 |
32 sec
|
42 sec
|
18 sec
|
Now these readings have been fairly carefully taken but I found there to be a great variability among the timings despite the facts that a)Photoshop was the only app running; b) all the tray apps and services had been stopped and c)identically the same photo was used in all the tests – a 12MPixel JPG taken with the new Nikon D90 camera. But to give an example Smart Blur operation varied between 22 and 17 seconds on Vista CS4 tests and 26 to 21 seconds on Vista CS3. In contrast, the Windows XP tests were relatively consistent with no variation in the Smart Blur timings and the widest range being the Merge HDR tests of 48 to 42 seconds on the first phase.
Microsoft Passes Its Problem to Adobe
What these numbers show is that Photoshop CS3 on Windows XP seriously outperforms both Photoshop CS3 and CS4 on Windows Vista. Even more troubling for Adobe is that PhotoShop CS4 is only marginally better than CS3 in Windows Vista. Now this is the 64bit version of Photoshop CS4. But I have not yet found the time to find, download, and test the drivers for my ATI 512MB Video card which may or may not improve performance of CS4. But the bottom line is that Microsoft has shifted its poorly performing Windows Vista problems onto Adobe because Redmond is forcing Windows consumers to buy Windows Vista. And of course Apple is laughing all the way to the bank by charging at least double if not 4 times for a comparably equipped Mac laptop or desktop.
I keep wondering why Adobe does not develop a Linux version of Photoshop. Adobe has the compiler technology in tow. And many of their high end customers are using Linux machines for video and 3d rendering on huge farms and would certainly welcome a Photoshop presence for direct editing. And of course Linux machines consume 1/4 the memory of Windows Vista and provide much better response time for comparable apps (Open Office for example). Finally, with Redmond lining up competitors to Flash, Photoshop, Dreamweaver and Acrobat – they have already attached a bulls-eye on many Adobe markets. So it is not as if Redmond could threaten them any more. But the bottom line is Redmond is making Adobe eat its Vista dogfood performance problems.