Well we spent most of the weekend trying to figure out our benchmarks results - would Windows 7 be able to beat Windows XP in raw speed and performance? Here is that tale of the table:
| Speed Comparison: Windows 7 versus Windows XP | ||
| Test | Windows 7 | Windows XP |
| Copy 800MB file from USB | 29sec | 27sec |
| Copy 800MB file to USB | 59sec | 52sec |
| Copy 800MB file to file | 25sec | 28sec |
| Copy 430MB directory USB to drive | 31 sec | 68sec |
| Create 120,000 line .sql file using PHP | 6sec | 12 sec |
| Insert 120,000 records into MySQL Table SQL | 46 min 05sec | 140sec |
| Select 120,000 records with filterand group by | <1 sec | < 1 sec |
| Insert 60,000 records into MySQL Supplier Table | 24min 11sec | 68sec |
| Insert 60,000 records into MySQL Supplier Table SQL | 275sec | 16sec |
| Insert 60,000 records into MySQL Supplier Table PHP | 291sec | 18sec |
| Left Join between Product and Supplier Tables | <1 sec | <1 sec |
| Premiere Elements Boot up | 29 sec | 25sec |
| Premiere Elements Analyze Scene | 200 sec | 190sec |
| Premiere Elements write FLV 10min clip 640 x 480 | 630sec | 428sec |
| Photoshop CS3 First Boot up | 14sec | 12sec |
| Photoshop CS3-Smart Blur 14MPixel image | 16sec | 12sec |
| Photoshop CS3-Finishing Impressionist 14MPixel image | 9sec | 6sec |
| Photoshop CS3-Finishing Filter Gallery 14MPixel image | 15sec | 9sec |
| Photoshop CS3-Finishing \impressionist 24MPixel image | 31sec | 25sec |
| Photoshop CS3-Crop, Resize, Shape, Color Correct | <1 sec | <1 sec |
| Photoshop CS3-Brighteness, Exposure, Sharpen | <1 sec | <1 sec |
| Photoshop CS3-Topaz Smart Sharpen 26MPixel image | 20sec | 18sec |
| Photoshop CS3-Reshape 26MPixel image | 160sec | 154sec |
| Photoshop CS3-Texture Finishing 26MPixel image | 44sec | 36sec |
Green shows the faster performer and red marks where the speed of a system is 2 or more times slower than the other. Two things are obvious from these benchmarks. Windows XP on a 2.26GHz Dual Core PC with 3GB of RAM plus a 250GB hard drive consistently outperforms Windows 7 on 2.10GHz Dual Core PC with 4GB of RAM plus a 500GB SATA hard drive.
Well of course Windows XP does, it has a faster CPU. Well, not exactly.
I looked up the Passmark ratings of the two CPUs and the Windows XP Dual Core is rated at 1046 while the Windows 7 is rated at 1251. So Windows 7’s CPU should have a 19.6% speed advantage over the Windows XP PC - but XP consistently outperforms Windows 7 even with 1GB less of memory available. But what is of real concern is the very poor performance of Windows 7 doing Web development tasks. PHP and MySQL ran significantly slower in Windows 7 using a XAMPP provided Apache-MySQL–PHP server combo. So I had to drag out PHPed and Xdebug to profile and debug the problems. To date I do not have any solutions or insights.
I was going to do some AutoCAD, Corel Draw, Java, JavaScript, Oracle and PostgreSQL benchmarking but the the Windows 7 problems with PHP and MySQL forced a delay in game penalty. If any readers have insights into the Windows 7 problems with PHP and/or MySQL please post a comment.
Summary of Benchmarks
On one hand Windows 7 on a Gateway with a 17.3 inch screen at $650 Canadian [before taxes] is a pleasure to work with. Users will have plenty of room to work in. And the new Windows 7 taskbar makes moving among apps a lot easier. The system had close to XP-like Boot up, Shutdown and Start-up or Hibernate restore times. And copy operations [ notably faster than XP] and open file popups are not painfully slow like in Vista. Also I have yet to run into the annoying Vista roller-coaster performance lags [bright and peppy for 20 minutes then slow as molasses in January for the next 25, then back to .....]. But I have to admit, it is galling to have to endure a 5-20% performance hit in photo and video editing when the new Windows 7 machine should be delivering the very opposite in performance.
3 Comments
The only way to run a fair benchmark would be to wipe a machine, install XP, run the benchmark, then wipe the same machine, install W7 and run the benchmark again. Too many variables come into play when you use two disparate hardware platforms… chipsets, video adapters, memory and bus speeds… photo and video editing speed is much more dependent on your video hardware and chipset than the OS itself.
Unless you use the same machine, this means absolutely nothing.
The hard drives themselves will make more of a difference than the OS, as will RAM latancy etc.
And those MySQL results look veeery strange, I can’t believe it’s 23x slower at anything!
I sympathize with the observations that identically the same machine testing would be the best; however no one is yet willing to pay for that … so I stick with the next best approximation. But my other benchmarks for clients between these two machines plus other XP and Win 7 machines show a consistent speed advantage to Windows XP in the 20-50% range doing Java and C compiles plus various defragging, backup and utility operations.
I also have found the MySQL results curious - so if I get a contract that I am bidding on then I will download the latest MySQL and do some additional benchmarking.