To NGen or Not to NGen?

One of the topics we often get questions on is about when it makes sense to invest the extra effort to pre-compile assemblies via NGen instead of simply relying on the JIT compiler to generate native code on the fly at application runtime. I thought I would try to answer that question in our first "real" post.

The JIT is optimized to balance code generation time against code quality, and it works well for many scenarios. NGen is essentially a performance optimization; so just like for anything else that's performance-related, you'll want to measure performance with vs. without NGen, and then decide whether it really helps your specific application and scenario. Here are a few general guidelines.

When to use NGen:

• Large applications: Applications that run a lot of managed code at start up are likely to see wins in start up time when using NGen. Microsoft Expression Blend for example, uses NGen to minimize start up time. If a large amount of code needs to be JIT-compiled at start up, the time needed to compile the IL might be a substantial portion of the total app launch time (even for cold start up) . Eliminating JIT-compilation from start up can therefore result in warm as well as cold start up wins.
• Frameworks, libraries, and other reusable components: Code produced by our JITs cannot be shared across multiple processes; NGen images, on the other hand, can be. Therefore any code that is likely to be used by multiple applications at the same time is likely to consume less memory when pre-compiled via NGen. Almost the entire Microsoft .NET Framework for instance, uses NGen. Also Microsoft Exchange Server pre-compiles its core DLLs that are shared across various Exchange services.
• Applications running in terminal server environments: Once again NGen helps in such a scenario because the generated code can be shared across the different instances of the application – that in turn increases the number of simultaneous user sessions the server can support.

When not to use NGen:

• Small applications: Small utilities like caspol.exe in the .NET Framework aren't NGen-ed because the time spent JIT-compiling the code is typically a small portion of the overall start up time. As a matter of fact, since NGen images are substantially bigger in size than the corresponding IL assemblies, using NGen might actually result in increased disk I/O and hurt cold start up time.
• Server applications: Server applications that aren't sensitive to long start up times, and don't have shared components are unlikely to benefit significantly from NGen.  In such cases,  the cost of using NGen (more on this below) may not be worth the benefits. SQL-CLR for example, isn't NGen-ed.

If it seems that your application is likely to benefit from NGen, here are a few things you might want to keep in mind:

1. NGen is triggered by issuing commands at install time to the ngen.exe tool in the .NET  Framework redistributable. To find out more about hooking NGen into set up, see here: http://blogs.msdn.com/astebner/archive/2007/03/03/how-to-ngen-files-in-an-msi-based-setup-package-using-wix.aspx.

Note: Currently you need admin privileges to issue commands via ngen.exe.

2. Just running ngen.exe on your assemblies and then re-running your performance scenarios to measure the wins is very likely not going to give you a true indication of the wins. Basically, it takes some effort to get the best performance out of NGen. For detailed guidelines on how to get the best performance via NGen, please take a look at this article in the MSDN magazine: http://msdn.microsoft.com/msdnmag/issues/06/05/CLRInsideOut/default.aspx.
3. Using NGen has implications for servicing your application. Your patch installer will need to issue NGen commands to ensure that the images invalidated by the patch get regenerated. .NET Framework 2.0 supports an "ngen.exe update /queue" command that should work well in most cases (this will regenerate all invalid NGen images at machine idle time). If specific assemblies are perf-critical and need to have up-to-date NGen images immediately after the patch install, you can issue separate NGen commands for those ("ngen.exe install <CriticalDLL>"), followed by the "ngen.exe update /queue" command.
4. Using NGen implies your application's disk footprint will increase (NGen images are fairly big when compared to MSIL assemblies).
5. In a handful of scenarios, cold start up and/or application execution can get slower when you use NGen, so you'll definitely want to measure the performance with vs. without NGen for all scenarios, and validate that it's a win overall for your application to be using NGen.

Finally, another important thing to keep in mind is that NGen isn't a magic solution for application start up time. NGen can speed up application start up by eliminating the time needed for JIT compiling the code executed at start up.

However, regardless of whether you do or don't use NGen, it is important to ensure that your application is architected to be performant. So for instance, if you care about start up time, you should try to minimize the amount of code that needs to get loaded and executed on your application's start up path.

Below are some good resources on measuring and optimizing application start up performance: http://blogs.msdn.com/vancem/archive/tags/Perf/default.aspx

http://msdn.microsoft.com/msdnmag/issues/06/02/CLRInsideOut

http://msdn.microsoft.com/msdnmag/issues/06/03/WindowsFormsPerformance/

And some for NGen:

NGen internals: http://msdn.microsoft.com/msdnmag/issues/05/04/NGen/default.aspx

NGen Service (NGen-ing your assemblies in the background):

http://blogs.msdn.com/davidnotario/ (David isn't on our team any longer)

Thanks for visiting our blog!