VB2008 Outperforms VB2005! (Lisa Feigenbaum)

VB2008 Outperforms VB2005! (Lisa Feigenbaum)

  • Comments 50

Looking for another reason to upgrade to 2008? How about performance! In addition to making tasks easier for you from release to release, another way we can save you time and make you more productive is by making the product faster! This is a goal we are very commited to on the VB team, and for which we have dedicated an entire virtual team for the Orcas release. Take a look at the stats below to see some interesting improvements.

[**DISCLAIMER: The data posted here is what I have observed from comparing 2 runs in our internal labs. These are my observations using two large customer solutions on a specific machine. Some people will have faster systems and some will have slower. Also, the work required for each operation will differ based on your solution. Thus, your results may be better or worse. The interesting thing is to look at the improvement between versions on identical hardware. I am posting this data to try and give you a general idea of the changes - but these numbers are by no means meant to be used as an industry standard.

For those interested, here's some data on the solutions and machine that I used... One solution contains 29 projects and 2772 files, with xml comments on all methods and types. The other solution used contains 783 files. The machine has the following specs: Dual Core Pentium-D - 3.0GHz, 1GB RAM, 10K RPM HD. OS=XP. These are one of many machine specifications that we use for our internal performance testing.]

Scenario VB2005 Time (ms) VB2008 Time (ms) VB2008 is x times faster than VB2005 (2005 time/ 2008 time) VB2008 takes x% as much time as VB2005 (100% * 2008 time/ 2005 time)
Build a large project (using background compilation) 222206.25 1352.88 164.25 0.61%
Build a large multi-project solution (explicit build operation) 1618604.75 57542.75 28.13 3.56%
Build a large multi-project solution (using background compilation) 222925.50 19861.88 11.22 8.91%
Responsiveness after adding a member to a class 327.00 36.50 8.96 11.16%
Responsiveness after opening a project 255551.25 38769.38 6.59 15.17%
Invoke Intellisense to see a list of types (first time) 1192.50 530.5 2.25 44.49%
Edit-and-Continue in a solution with xml comments (first time) 441.25 210.5 2.10 47.71%
Responsiveness after changing a method statement 390.25 236.38 1.65 60.57%
10 Steps in the debugger (subsequent times) 1850.75 1167.13 1.59 63.06%
Invoke Intellisense to see a list of types (subsequent times) 79.25 51.5 1.54 64.98%
F5 when the solution is already built (subsequent times) 385.20 278.7 1.38 72.35%
Item gets added to the Error List after making an error 531.25 394.5 1.35 74.26%
10 Steps in the debugger (first time) 1336.50 1150 1.16 86.05%
Responsiveness while background compiling on an open solution 4803.00 4284.75 1.12 89.21%
Load a large solution (subsequent times) 13667.5 12407.25 1.10 90.78%
Load a large solution (first time)
(Note: This is the improvement on XP. Vista has seen about twice as much improvement for the Load Solution scenario than XP. )
19946.25 18222 1.09 91.36%
** All times are found by performing operations on customer solutions. Don't you wish we were using your solution to optimize our times? Send it in so that we can optimize for you! (vbperf@microsoft.com) We are always interested to get more customer projects for our testing.

Is there another scenario you're interested that you don't see above, where you continue to experience performance lags in your regular usage of the product? Please send in repro steps so we can take a look! (vbperf@microsoft.com)

Other interesting VB2008 Performance content to check out is listed below:

Leave a Comment
  • Please add 6 and 5 and type the answer here:
  • Post
  • The math is correct if you take into consideration that 0 to 99% is the same amount of time as 99% to 100% and that time is relative to processor and download speed in the world of microsoft.

  • Very good!! now let us see the results verified by an independent source other than Micrsoft funded blog.

  • The percentages column works if you define 100% as the greatest possible improvement (i.e. it works instantly), 0% as no improvement, and -100% as the worst possible slowdown (it takes forever)

    200% would mean that it is finished before you ask it to do anything. We can only dream.

  • I would like to see some OLEDB or ODBC processing time, or data control population, etc.  How long does it take to go through a half-million records, or 5 million?  How long to update 1000 records in a database?  Can we see those stats?  

    I am quite sure I'll be upgrading to VS 2008, and will enjoy the upgrades and advances.  But, I'd love to see some data optimization.

  • Nowadays, performance of IDE is not very important as long as it is functioning because of the power of machine (the power is increasing every month). The most important aspects is the new features and ease-of-use  the IDE.

  • I immediately noticed improvements to the responsiveness of VS2008 while using VB.  The intellisense improvements were especially noticeable, which is even more admirable when you consider the additional functionality in this release.  No unexpected bugs or crashes noted so far, although the multi-project solution I'm working  with is still relatively small.  After using the VS2008 IDE, it's hard to go back to VS2005, and I will likely be converting a number of projects over just to take advantage of the improvements.

    Thanks to Microsoft for putting emphasis on performance.

  • I think that is interesting as well but am going to wait and see how many IDE bugs pop out of this iteration of the IDE.     VS2005 had and continues to have at least as many problems with large solutions (150+ projects) as VS2003.     Prove to me that the IDE won't disappear periodically and there are no hotfixes for large solutions and we'll consider taking ours up to 2008.

  • Is there a comparison for VC# 2005 and VC#2008?

  • The % column has confused a number of people, so I've changed it to mean "VB2008 requires x% of the VB2005 time to complete this scenario". I hope that will be more clear. I've also included equations in the column headers. I apologize for any inconvenience.

    Lisa

  • These types of benchmarks should also include information on the test computer. Users could expect dramatic increases on their machine only to find they need to spend another $600 on their machine to get the performance enhancements.

  • Does it let me edit the program while debugging, like VB6, to fix bugs without ever having to stop-and-restart my debug session? That would be a real performance improvement!

  • Hi Pete Austin,

    Yes - VB2005 and VB2008 both have the "Edit and Continue" feature that you're remembering from VB6. Try it out!

  • To the person asking for times for updating a database - that is nothing to do with VB and is down to the .NET framwork ADO.NET classes.  This blog is about the speed of the VB implementation of the VS IDE which has always had speed issues due particularly to the background compilation.

  • I didn't use the vs2k8 yet,  if that apps are much faster than the former apps i use, i think its time to shift to it, so if you have a new license copy, can you send me one?

  • xybr0xym0r0n,

    We cannot give out license copies. (I know this might come as a surprise. :) )

    However, the Express editions are available at the following link:

    http://msdn2.microsoft.com/en-us/express/future/bb421473.aspx

Page 2 of 4 (50 items) 1234