|
|
|
| Store Home | View Cart or CheckOut |
Order Tracking | Privacy Statement | Store Policy | Contact Us | About Us |
Search Click here to Search by Part Number Store Index Memory MacGurus RAM SATA (Serial ATA) Host Cards Burly Enclosures Cables Hard Drives RAID Burly Hardware RAID SuperDrive Burly DVD-R Burner SCSI SCSI-Firewire Adapters FireWire Case Kits and Drives Multi-Drive Enclosures Host Cards FW 400 Cables FW 800 Cables Enclosure Parts Burly Parts Hard Drives SATA Hard Drives Notebook Hard Drives MacBooks Express34 and Cardbus Expansion Cards Momentus Drives Special Deals Specials&Closeouts Rick’s Blogs SATA III Burly Maintenance Why a Burly? Tech Guides Guides to Acceleration Building a Photo Database Storage Acceleration Photoshop Acceleration Storage Guides DeskTop Drive Setup Preparing New Drives Move Users 10.5 Move Users 10.4 Port Multiplication Guide Table - Front Side Bus Giga CPU Upgrade Roll Your Own SATA Hot Swap SATA Guide Roll Your Own RAID Motherboard Layouts Drive & RAID Database |
Photoshop Performance and System Configuration The graph below simulates a realistic workflow environment. This includes file saves along with other commonly-used operations like rotating and cropping. The test file is purposely quite large- 878.9MB to be exact- in order to force heavy scratch disk use on every operation.  This graph immediately illustrates the performance benefits of multiple data channels for Photoshop. Bar #1 represents the "default" configuration; the designated scratch disk is the boot volume. Bar #2 shows a significant gain achieved with the scratch disk on the 4-drive RAID stripe. Bar #3 shows even more acceleration; the image file is saved on the separate User/data drive, and the scratch disk again is on the 4-drive RAID stripe. Bar #4 represents probably the fastest possible scenario. The image file has been saved to the striped RAID scratch. The performance benefit is obvious, but so is the risk. In case you didn’t pay attention earlier, we’ll say it again: saving data on a striped RAID volume is many, many times more risky than saving on a non-striped volume. Use this method only with a frequent backup plan already in place. Is this the end of the story? No, because as usual, the devil is in the details. And as we’ll see, those details become important when planning system configurations for Photoshop use. When we break this test routine down a little- separating file saves from other operations, some interesting comparisons emerge:  Bar #1 represents the "default" configuration; the designated scratch disk is the boot volume. Bar #2 shows a significant gain achieved with the scratch disk on the 4-drive RAID stripe. But it is also evident that TIFF file saves are substantially slower. Yet, the speed for non-save operations is faster and indeed, the overall performance is faster. Bar #3 shows even more acceleration; the file is saved on the separate User/data drive, and the scratch disk again is on the 4-drive RAID stripe. All file saves are now faster, but non-file save operations suffer slightly. Bar #4 is the "work disk/scratch disk" method; by far the fastest all around with no obvious tradeoffs. All types of operations are faster. Ok, so you want to live dangerously? Frequent backups, people. Some Final Thoughts.. Effective Photoshop acceleration results from a combination of many factors, and hardware upgrade plans must be based on a user’s current and anticipated real-world workflow requirements in order to be cost effective. However, some general conclusions can be drawn:
Below are the individual test results charted in the graphs above, and the settings used for the tests. The key for the various directory locations is as follows: A = /Users (the default location of Users in a standard OSX installation) B = /Volumes/DriveB/Users/myhomefolder (the Users directory is on the second internal drive) C = /Volumes/external 4-drive striped RAID array volume D =/private/tmp/501/Cleanup At Startup/ (the default location of Photoshop’s temporary file) Photoshop Preferences Settings:
History Palette Options: first 3 boxes checked Export Clipboard: off Image Previews: always save; icon, mac & win thumbs File Compatibility: none Recent File List: 1 Enable Workgroup: off Displays & Cursors: precise cursors, pixel doubling, color channels in color off Memory & Image Cache: cache levels 8 Plugins & Scratch Disks: no extra plugin folders and either the startup as designated scratch disk or the external RAID array, depending on the test. Only one designated scratch disk was used at a time. Layers and Channels Palette thumbnail options: small (All other preference settings were unchanged from the default installation) 
†the individual settings in the Lighting Effects filter were the default "Spotlight" settings with intensity on 100. The test machine used for this article is a "rev 1" Powermac G5 2GHz Dual Processor (Boot ROM Version 5.14f0) running Mac OS 10.3.3 (7F44).
The application is installed on the boot volume in the standard location of /Applications. The following modifications were made to the standard installation: Photoshop was upgraded from 7.0.0 to 7.0.1. The G5 update was installed. The Photoshop Camera Raw & JPEG 2000 plugin was installed. The AltiVecCore Update was installed. The FileBrowserUpdate was installed. The OS X Keyboard Shortcut Fix was installed. Photoshop Scripting 1.0.2a was installed. Aside from the default plugins and those listed above, no other additional plugins are installed. The original 7200rpm Seagate Barracuda 160GB SATA hard drive was removed. Two 10,000rpm Western Digital Raptors (WD740GD-32FLA0 firmware rev. 21.08U21) 74GB SATA drives were installed in drive bays A and B. Both drives are attached to the motherboard SATA bus with the OEM SATA cables. Each drive is formatted as a single journaled volume with Apple Disk Utility 10.4.2 (v145.3) Drive A is the boot volume and is 69.12 GB formatted as Journaled HFS+ with 53.37 GB free space. Drive B is 69.12 GB formatted as Journaled HFS+ with 57.3 GB free space. The original ATI Radeon 9600 Pro AGP video card was removed and replaced with an ATI Radeon 9800 Pro SE AGP video card (ATY,R350 ROM Revision: 113-A14402-116 256MB VRAM). The display used throughout the testing is a 22" Apple Cinema Display using a 1600 x 1024 resolution. The only PCI card installed is an ATTO ExpressPCIProUL4D (ROM Revision: 1.4.2f1 Revision ID: 0x0008), a two-channel SCSI controller in slot #4. Software for the card is ATTO Configuration 2.6.1 and ATTOExpressPCI4 Driver Version 3.1.0f1. Connected to this card is an external four-drive ’Burly’ RAID array hand-built with GraniteDigital connectors and terminators. Inside the Burly are 4 x 18GB SEAGATE x15.3 Cheetah SCSI drives (ST318453LW rev. 0006). The four drives are configured as one un-Journaled HFS+ 68GB striped volume (68.3 GB capacity and free space) using SoftRAID 3.0.3 as the disk driver with the "Workstation/128k stripe size" setting. During the testing, no other applications were opened or used, and no Finder windows were open. Each test run was completed without interruption. Several background processes were always running and no special effort was made to ’kill’ any processes. Airport was on and continually connected to the Internet, but File Sharing was always off. A Firewire scanner was on and always connected, as were a usb UPS connection, external usb speakers, and an ink-jet printer. Energy Saver preference settings were set to never sleep the computer or display, and never spin down the hard drive. Processor performance was set to ’highest’. Prior to running any test which used the external striped RAID array volume as the designated scratch disk, or as a "scratch disk/work disk", that volume was erased with SoftRAID and the directory optimized with Diskwarrior 3.0.2. The computer was then re-booted. Before any testing was conducted, the computer was maintained with Disk Utility and Diskwarrior. After test 2.2, the User directory was pointed back to the pre-existing User directory on the second internal drive. The relevant Photoshop preferences were changed, and again the computer was maintained with Disk Utility and Diskwarrior. Before each of the 4 test runs, the relevant changes were made to Photoshop’s preferences, and the computer was re-booted. Photoshop was then opened, and it’s preferences again checked to ensure the changes were in effect. All times were recorded using the timer feature in Photoshop 7. Special care was taken to carefully observe the timer, which displays two numbers for each completed operation: the first is elapsed time, the second is elapsed time plus screen redraw time. Only the first number was used. All test results were immediately recorded by hand with pencil and paper. Scoring for the tests was simple and straight forward. The time for each operation to complete was recorded and the results were added upon the test’s completion for the total. The method was repeated for each test configuration. |
| Online Store | About Us | Contact Us | Store Policy | Home | Forums |
| 3D Animated Flags--By 3DFlags.com |
Copyright © 2003-2008 MacGurus All Rights Reserved |
Proudly