Thats a massive 26.5% increase, so if you are attempting to compare scores between platforms you must be running the same version of photoshop.

********

Most of the regulars coming to Driverheaven.net will be aware of the fact we run a custom Photoshop benchmark in all of our motherboard and memory hardware reviews. Today we are releasing the new overhauled V2, which offers full compatibility with V7 and the newest CS version. The V2 benchmark will be now featuring in our forthcoming reviews and is a fantastic real world test of cpu power and memory bandwidth. I have been using Photoshop since 1990 on the Macintosh and have taught many users how to unleash the versatility of this incredible application.

In case you missed the link above, you can download the action benchmark script and test image here. Before running the benchmark please read the information on this webpage to ensure correct benchmarking. The archive is dual encoded, winrar on the outer shell and a self extracting uharc archive within. Use winrar to extract then double click the EXE.

This script is compatible with Apple Macintosh versions of photoshop and you can download the Macintosh friendly ZIP version here. For those reading this who own a website you are allowed to mirror either version of the file and this page information as long as the contents including the readme, or benching actions script are not altered in any way - a link to this page would also be appreciated.

Most people are tired of running 3dmark and Pcmark and similar benchmarks, while these are visually attractive benchmarks they are not in many cases indicative of real world situations, this is not the case with photoshop. This is a real world application with a myriad of uses appealing to all users from digital photographers to budding graphic designers. It is also a wonderful test of all round system performance, specifically memory bandwidth and CPU processing power. Massive files are also a good test of hard drive access performance as photoshop will rely heavily for its "paging file".

This version of Driverheaven Photoshop Bench as I mentioned is compatible with V7 and CS, but I really have to stress, due to coding alterations between versions of the application, along with rewritten filters, the test results are not interchangeable. Photoshop CS has some improvements to filter coding and performance and many results such as the stained glass filter give very different timing results. In forthcoming Driverheaven reviews all results will be taken from CS but im aware not everyone will have the newest version so ive ensured the suite will work with the older V7, the results will just be slower.

Preparations before benchmarking

Firstly, you are going to need 512 meg of ram minimum in your system and you must ensure before running the benchmark that all cpu hogging background programs are closed and any applications such as antivirus which might corrupt test results are disabled. My rule of thumb is to restart your system and load photoshop fresh into memory, you will see some quite large variations if not. Those of you with 1024 or more of system memory can skip the next memory optimisation section, but those with 512 should follow my guidelines.

After loading Photoshop, go to EDIT/Preferences/Memory and Image Cache.

Once you are into the subpanel, ensure photoshop has a minimum of 300 meg of ram allocated.

You will need to reload the application, however to ensure accurate results, I also advise a restart, this is to verify no memory fragmentation occurs. Once you restart, reload Photoshop, and open our test file. The test file is a JPEG which will decompress to 60 megabytes. For those interested, its a picture of carrickfergus castle, in Ireland, taken with my Canon EOS 20D digital camera and resampled to a larger size.

Open the driverheaven test image "driverheaventest.jpg" file


Once you have opened this, open the actions pallette, making sure first that button mode is OFF.

Within the same section scroll down the options to LOAD ACTION.

This will open a window, drive to the place you extracted the archive and load the DH BENCH.atn file. Once loaded press button mode again and you should see this

Benchmarking

If you have opened the test image and loaded the DH bench action script, now all you have to do is press F4 or click on the red actions bar and follow the instructions, noting the times for each section. After a short while you should have the results of 12 different subbenchmarks which will give a total figure. These benchmarks are varied and will stress different facets of your system CPU and memory bandwidth performance giving an overall time.

To record times, you use the built in photoshop timing feature which is accessible from here simply move the selection until "TIMING" is highlighted. you will now see a readout to the left of this. if the time does NOT say 0.0 before running the benchmark then we need to reset the time (opening the test image will probably have set this counter to around 1 second). Hold the ALT key and click on the arrow then back to TIMING again, you should see the figure reset to 0.0 quickly - this ensures a zero starting point.

The script will explain what you need to do as it is running with stop points so you can record times for the final result and to compare processor and system differences between various sections of the suite.

The suite comprises the following sections:

1: Texturiser Test (1)
A Canvas filter, scaling 106, relief 5, without inverting texture and light direction from the top.

2: CMYK Colour Conversion
Important for all the professionals as this 4 plate (cyan, magenta, yellow and black (k-key colour)) colour mode is used in graphics bureaus and printing presses for colour reproduction to newsprint or magazine. Digital photographs will be converted to this mode for final output.

3: RGB Colour Conversion
Important test for web designers as this colour mode is used for webpages and will be the mode of choice for digital images raw from the camera.

4: Dust and Scratches
A filter which is used quite widely by professionals scanning less than perfect original flat copy. Radius is set to 3 and Threshold is set to 36.

5: Watercolour
An effective and intense filter, set to Brush detail 9, shadow intensity 2, and texture of 2.

6: Texturiser Test (2)
Another run of the texturiser test, this time with different settings which will be applied on top of the watercolour filter above, Canvas filter again, scaling of 109, Relief of 11, invert texture on this time and light direction again from the top.

7: Stained Glass
A very intensive filter indeed, which ran slower on V7, with CS this filter runs in a fraction of the time, Cell size 10, Border Thickness 4, Light Intensity of 3.

8: Lighting Effects
Light source on RGB colour, red 1, green 113, Blue 252, With light on, Intensity of 85, Light Type, Omni, Position 50,926,50, Vector 0: 35 185 94.907, Vector 1: 27 778 68 056, Radius of 19.444, Current light 1, Gloss 71, Material 35, Exposure 46, Ambience -37, Ambient colour: RGB, Red 255, Green 255, Blue 255, Texture Channel, blue, with white is high, height 100, Frame width 68.519.

9: Mosiac Tiles
Makes use of memory bandwidth, Tile Size 9, Grout Width 3, Lighten Grout 9.

10: Extrude
Size 12, Depth 16, without solid front faces, without mask incomplete blocks, type: blocks, and random.

11: Smart Blur
Radius 20.7, Threshold 28.7, Quality High, Mode Normal.

12. Underpainting
Brush size 7, Texture coverage 16, Texture type Canvas, Scaling 100, Relief 4, Invert texture on, light direction, top.

Comparison results 1 (High End Overclocked Intel PC System):
Intel Extreme Edition processor 3.73 ghz @ 4.1ghz (64 bit processor), Asus P5AD2-E Motherboard 1002 bios, 2 gig of OCZ ram running at 800mhz 8-2-3-4. OCZ 520 Powerstream supply. Maxtor 300gig 16 meg Cache HD, Windows XP 64 bit Professional. Photoshop CS.Other test results will be added shortly on other systems, for more results and to compare with our members, please visit this thread on our forums.

1: Texturiser Test (1)
2.5 seconds

2: CYMK Colour Conversion
3.1 seconds

3: RGB Colour Conversion
5.5 seconds

4: Dust and Scratches
6.1 seconds

5: Watercolour
21.3 seconds

6: Texturiser Test (2)
2.6 seconds

7: Stained Glass
9.5 seconds

8: Lighting Effects
5.2 seconds

9: Mosiac Tiles
12.1 seconds

10: Extrude
38 seconds

11: Smart Blur
22 seconds

12. Underpainting
18 seconds

Total time: 145.9 seconds.


Comparison results 2 (Laptop):
1.6 Dothan, 512MB PC2700 (2x256), XP Home, Pshop 7.0, 60GB 4200rpm HDD

1: Texturiser Test (1)
4.0 seconds

2: CYMK Colour Conversion
11.4 seconds

3: RGB Colour Conversion
13.1 seconds

4: Dust and Scratches
51.3 seconds

5: Watercolour
42.5 seconds

6: Texturiser Test (2)
14.8 seconds

7: Stained Glass
339.6 seconds

8: Lighting Effects
14.0 seconds

9: Mosiac Tiles
24.3 seconds

10: Extrude
81.9 seconds

11: Smart Blur
43.2 seconds

12. Underpainting
38.3 seconds

Total time: 678.4 seconds.

Comparison Results 3 (High End Macintosh System)
http://homepage.mac.com/znowdog/ - thanks : Lars Broberg
PowerMac G5, dual 2.5, 2GB ram, WD Raptor, Photoshop CS

1: Texturiser Test (1)
10.1 seconds

2: CYMK Colour Conversion
2.2 seconds

3: RGB Colour Conversion
2.4 seconds

4: Dust and Scratches
3.5 seconds

5: Watercolour
29.9 seconds

6: Texturiser Test (2)
8.5 seconds

7: Stained Glass
10.8 seconds

8: Lighting Effects
4.4 seconds

9: Mosiac Tiles
40.4 seconds

10: Extrude
65.8 seconds

11: Smart Blur
60.3 seconds

12. Underpainting
38.4 secondsTotal time: 276.7 seconds.

Special thanks to Craig "5320" Humphreys and Jon "Tastyweat" McRae for their beta testing during the development stages.