Whidbey Web Forms Editor team has beaten Whidbey Beta 1 performance goals!

Test Machine: PIII600/128MB, W2K Pro FAT32

Scenario (Unit) 40512 Beta1 Goal Diff Beta1 Bound By
Add a new web form (cold) (s) 11.01 18.00 39% I/O 
Cold F5 (PortalVBSDK) (s) 256MB 35.62 60.00 41% I/O 
Open DeskTopDefault.aspx (cold) (s) 10.67 20.00 47% I/O 
Load Web Site (Cold) (s) 4.71 10.00 53% I/O 
Warm F5 (s) 256MB 2.29 5.00 54% CPU 
Devenv ShoppingCart.aspx - cold (s) 15.22 41.00 63% I/O 
Devenv ShoppingCart.aspx - warm (s) 5.98 16.00 63% I/O 

Many people contributed to the result, especially our performance guru Ibrahim (who does not have a blog - yet). We hope you (our customers) will like the product performancel. We are glad that we are now “officially“ beating expectations since according to the official VS Performance team measurements we have been slow all along (see the following chart, X axis is build number)

We could not explain why our local team measurements were much better than the official VS results despite using machine with the same CPU type and speed, same amount of RAM, disk drive with the same RPM, etc. Our team implemented tons of performance improvements and yet could not get official number to move! We finally figured out that we have been testing on NTFS while VS performance team had been using FAT32! On top of that we had a bug in the file time comparison that only manifested itself up on FAT32. The bug caused excessive regeneration of validation schemas which was killing our startup performance numbers.

Now on to Beta 2 goals... ;-) New scenarios, new challenges...