Question

Although I do understand the serious implications of playing with this function (or at least that's what I think), I fail to see why it's becoming one of these things that respectable programmers wouldn't ever use, even those who don't even know what it is for.

Let's say I'm developing an application where memory usage varies extremely depending on what the user is doing. The application life cycle can be divided into two main stages: editing and real-time processing. During the editing stage, suppose that billions or even trillions of objects are created; some of them small and some of them not, some may have finalizers and some may not, and suppose their lifetimes vary from a very few milliseconds to long hours. Next, the user decides to switch to the real-time stage. At this point, suppose that performance plays a fundamental role and the slightest alteration in the program's flow could bring catastrophic consequences. Object creation is then reduced to the minimum possible by using object pools and the such but then, the GC chimes in unexpectedly and throws it all away, and someone dies.

The question: In this case, wouldn't it be wise to call GC.Collect() before entering the second stage?

After all, these two stages never overlap in time with each other and all the optimization and statistics the GC could have gathered would be of little use here...

Note: As some of you have pointed out, .NET might not be the best platform for an application like this, but that's beyond the scope of this question. The intent is to clarify whether a GC.Collect() call can improve an application's overall behaviour/performance or not. We all agree that the circumstances under which you would do such a thing are extremely rare but then again, the GC tries to guess and does it perfectly well most of the time, but it's still about guessing.

Thanks.

Was it helpful?

Solution

From Rico's Blog...

Rule #1

Don't.

This is really the most important rule. It's fair to say that most usages of GC.Collect() are a bad idea and I went into that in some detail in the orginal posting so I won't repeat all that here. So let's move on to...

Rule #2

Consider calling GC.Collect() if some non-recurring event has just happened and this event is highly likely to have caused a lot of old objects to die.

A classic example of this is if you're writing a client application and you display a very large and complicated form that has a lot of data associated with it. Your user has just interacted with this form potentially creating some large objects... things like XML documents, or a large DataSet or two. When the form closes these objects are dead and so GC.Collect() will reclaim the memory associated with them...

So it sounds like this situation may fall under Rule #2, you know that there's a moment in time where a lot of old objects have died, and it's non-recurring. However, don't forget Rico's parting words.

Rule #1 should trump Rule #2 without strong evidence.

Measure, measure, measure.

OTHER TIPS

If you call GC.Collect() in production code you are essentially declaring that you know more then the authors of the GC. That may be the case. However it's usually not, and therefore strongly discouraged.

So how about when you are using COM objects like MS Word or MS Excel from .NET? Without calling GC.Collect after releasing the COM objects we have found that the Word or Excel application instances still exist.

In fact the code we use is:

Utils.ReleaseCOMObject(objExcel)

' Call the Garbage Collector twice. The GC needs to be called twice in order to get the
' Finalizers called - the first time in, it simply makes a list of what is to be finalized,
' the second time in, it actually does the finalizing. Only then will the object do its 
' automatic ReleaseComObject. Note: Calling the GC is a time-consuming process, 
' but one that may be necessary when automating Excel because it is the only way to 
' release all the Excel COM objects referenced indirectly.
' Ref: http://www.informit.com/articles/article.aspx?p=1346865&seqNum=5
' Ref: http://support.microsoft.com/default.aspx?scid=KB;EN-US;q317109
GC.Collect()
GC.WaitForPendingFinalizers()
GC.Collect()
GC.WaitForPendingFinalizers()

So would that be an incorrect use of the garbage collector? If so how do we get the Interop objects to die? Also if it isn't meant to be used like this, why is the GC's Collect method even Public?

Well, the GC is one of those things I have a love / hate relationship with. We have broken it in the past through VistaDB and blogged about it. They have fixed it, but it takes a LONG time to get fixes from them on things like this.

The GC is complex, and a one size fits all approach is very, very hard to pull off on something this large. MS has done a fairly good job of it, but it is possible to fool the GC at times.

In general you should not add a Collect unless you know for a fact you just dumped a ton of memory and it will go to a mid life crisis if the GC doesn't get it cleaned up now.

You can screw up the entire machine with a series of bad GC.Collect statements. The need for a collect statement almost always points to a larger underlying error. The memory leak usually has to do with references and a lack of understanding to how they work. Or using of the IDisposable on objects that don't need it and putting a much higher load on the GC.

Watch closely the % of time spent in GC through the system performance counters. If you see your app using 20% or more of its time in the GC you have serious object management issues (or an abnormal usage pattern). You want to always minimize the time the GC spends because it will speed up your entire app.

It is also important to note that the GC is different on servers than workstations. I have seen a number of small difficult to track down problems with people not testing both of them (or not even aware that their are two of them).

And just to be as full in my answer as possible you should also test under Mono if you are targeting that platform as well. Since it is a totally different implementation it may experience totally different problems that the MS implementation.

There are situations where it's useful, but in general it should be avoided. You could compare it to GOTO, or riding a moped: you do it when you need to, but you don't tell your friends about it.

From my experience it has never been advisable to make a call to GC.Collect() in production code. In debugging, yes, it has it's advantages to help clarify potential memory leaks. I guess my fundamental reason is that the GC has been written and optimized by programmers much smarter then I, and if I get to a point that I feel I need to call GC.Collect() it is a clue that I have gone off path somewhere. In your situation it doesn't sound like you actually have memory issues, just that you are concerned what instability the collection will bring to your process. Seeing that it will not clean out objects still in use, and that it adapts very quickly to both rising and lowering demands, I would think you will not have to worry about it.

One of the biggest reasons to call GC.Collect() is when you have just performed a significant event which creates lots of garbage, such as what you describe. Calling GC.Collect() can be a good idea here; otherwise, the GC may not understand that it was a 'one time' event.

Of course, you should profile it, and see for yourself.

Well, obviously you should not write code with real-time requirements in languages with non-real-time garbage collection.

In a case with well-defined stages, there is no problem with triggering the garbage-collector. But this case is extremely rare. The problem is that many developers are going to try to use this to paper-over problems in a cargo-cult style, and adding it indiscriminately will cause performance problems.

Calling GC.Collect() forces the CLR to do a stack walk to see if each object can be truely be released by checking references. This will affect scalability if the number of objects is high, and has also been known to trigger garbage collection too often. Trust the CLR and let the garbage collector run itself when appropriate.

Creating images in a loop - even if you call dispose, the memory is not recovered. Garbage collect every time. I went from 1.7GB memory on my photo processing app to 24MB and performance is excellent.

There are absolutely time that you need to call GC.Collect.

Infact, I don't think it is a very bad practice to call GC.Collect.
There may be cases when we need that. Just for instance, I have a form which runs a thread, which inturn opens differnt tables in a database, extracts the contents in a BLOB field to a temp file, encrypt the file, then read the file into a binarystream and back into a BLOB field in another table.

The whole operation takes quite a lot of memory, and it is not certain about the number of rows and size of file content in the tables.

I used to get OutofMemory Exception often and I thought it would be wise to periodically run GC.Collect based on a counter variable. I increment a counter and when a specified level is reached, GC is called to collect any garbage that may have formed, and to reclaim any memory lost due to unforeseen memory leaks.

After this, I think it is working well, atleast no exception!!!
I call in the following way:

var obj = /* object utilizing the memory, in my case Form itself */
GC.Collect(GC.GetGeneration(obj ,GCCollectionMode.Optimized).

Under .net, the time required to perform a garbage collection is much more strongly related to the amount of stuff that isn't garbage, than to the amount of stuff that is. Indeed, unless an object overrides Finalize (either explicitly, or via C# destructor), is the target of a WeakReference, sits on the Large Object Heap, or is special in some other gc-related way, the only thing identifying the memory in which it sits as being an object is the existence of rooted references to it. Otherwise, the GC's operation is analogous to taking from a building everything of value, and dynamiting the building, building a new one on the site of the old one, and putting all the valuable items in it. The effort required to dynamite the building is totally independent of the amount of garbage within it.

Consequently, calling GC.Collect is apt to increase the overall amount of work the system has to do. It will delay the occurrence of the next collection, but will probably do just as much work immediately as the next collection would have required when it occurred; at the point when the next collection would have occurred, the total amount of time spent collecting will have been about the same as had GC.Collect not been called, but the system will have accumulated some garbage, causing the succeeding collection to be required sooner than had GC.Collect not been called.

The times I can see GC.Collect really being useful are when one needs to either measure the memory usage of some code (since memory usage figures are only really meaningful following a collection), or profile which of several algorithms is better (calling GC.Collect() before running each of several pieces of code can help ensure a consistent baseline state). There are a few other cases where one might know things the GC doesn't, but unless one is writing a single-threaded program, there's no way one can know that a GC.Collect call which would help one thread's data structures avoid "mid-life crisis" wouldn't cause other threads' data to have a "mid-life crises" which would otherwise have been avoided.

We had a similar issue with the garbage collector not collecting garbage and freeing up memory.

In our program, we were processing some modest sized Excel Spreadsheets with OpenXML. The spreadsheets contained anywhere from 5 to 10 "sheets" with about 1000 rows of 14 columns.

The program in a 32 bit environment (x86) would crash with an "out of memory" error. We did get it to run in an x64 environment, but we wanted a better solution.

We found one.

Here are some simplified code fragments of what didn't work and what did work when it comes to explicitly calling the Garbage Collector to free up memory from disposed objects.

Calling the GC from inside the subroutine didn't work. Memory was never reclaimed...

For Each Sheet in Spreadsheets
    ProcessSheet(FileName,sheet)
Next

Private Sub ProcessSheet(ByVal Filename as string, ByVal Sheet as string)
    ' open the spreadsheet 
    Using SLDoc as SLDocument = New SLDocument(Filename, Sheet)
        ' do some work....
        SLDoc.Save
    End Using
    GC.Collect()
    GC.WaitForPendingFinalizers()
    GC.Collect()
    GC.WaitForPendingFinalizers()
End Sub

By Moving the GC call to outside the scope of the subroutine, the garbage was collected and the memory was freed up.

For Each Sheet in Spreadsheets
    ProcessSheet(FileName,sheet)
    GC.Collect()
    GC.WaitForPendingFinalizers()
    GC.Collect()
    GC.WaitForPendingFinalizers()
Next

Private Sub ProcessSheet(ByVal Filename as string, ByVal Sheet as string)
    ' open the spreadsheet 
    Using SLDoc as SLDocument = New SLDocument(Filename, Sheet)
        ' do some work....
        SLDoc.Save
    End Using
End Sub

I hope this helps others that are frustrated with the .NET garbage collection when it appears to ignore the calls to GC.Collect().

Paul Smith

Nothing is wrong with explicitly calling for a collection. Some people just really want to believe that if it is a service provided by the vendor, don't question it. Oh, and all of those random freezes at the wrong moments of your interactive application? The next version will make it better!

Letting a background process deal with memory manipulation means not having to deal with it ourselves, true. But this does not logically mean that it is best for us to not deal with it ourselves under all circumstances. The GC is optimized for most cases. But this does not logically mean that it is optimized in all cases.

Have you ever answered an open question such as 'which is the best sorting algorithm' with a definitive answer? If so, don't touch the GC. For those of you who asked for the conditions, or gave 'in this case' type answers, you may proceed to learn about the GC and when to activate it.

Gotta say, I've had application freezes in Chrome and Firefox that frustrate the hell out of me, and even then for some cases the memory grows unhindered -- If only they'd learn to call the garbage collector -- or given me a button so that as I begin to read the text of a page I can hit it and thus be free of freezes for the next 20 minutes.

I think you are right about the scenario, but I'm not sure about the API.

Microsoft says that in such cases you should add memory pressure as a hint to the GC that it should soon perform a collection.

What's wrong with it? The fact that you're second-guessing the garbage collector and memory allocator, which between them have a much greater idea about your application's actual memory usage at runtime than you do.

The desire to call GC.Collect() usually is trying to cover up for mistakes you made somewhere else!

It would be better if you find where you forgot to dispose stuff you didn't need anymore.

Bottom line, you can profile the application and see how these additional collections affect things. I'd suggest staying away from it though unless you are going to profile. The GC is designed to take care of itself and as the runtime evolves, they may increase efficiency. You don't want a bunch of code hanging around that may muck up the works and not be able to take advantage of these improvements. There is a similar argument for using foreach instead of for, that being, that future improvements under the covers can be added to foreach and your code doesn't have to change to take advantage.

The .NET Framework itself was never designed to run in a realtime environment. If you truly need realtime processing you would either use an embedded realtime language that isn't based on .NET or use the .NET Compact Framework running on a Windows CE device.

The worst it will do is make your program freeze for a bit. So if that's OK with you, do it. Usually it's not needed for thick client or web apps with mostly user interaction.

I have found that sometimes programs with long-running threads, or batch programs, will get OutOfMemory exception even though they are disposing objects properly. One I recall was a line-of-business database transaction processing; the other was an indexing routine on a background thread in a thick client app.

In both cases, the result was simple: No GC.Collect, out of memory, consistently; GC.Collect, flawless performance.

I've tried it to solve memory problems several other times, to no avail. I took it out.

In short, don't put it in unless you're getting errors. If you put it in and it doesn't fix the memory problem, take it back out. Remember to test in Release mode and compare apples to apples.

The only time things can go wrong with this is when you get moralistic about it. It's not a values issue; many programmers have died and gone straight to heaven with many unneccessary GC.Collects in their code, which outlives them.

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top