Search Results

Search found 53 results on 3 pages for 'profilers'.

Page 3/3 | < Previous Page | 1 2 3 

  • Is valgrind crazy or is this is a genuine std map iterator memory leak?

    - by Alberto Toglia
    Well, I'm very new to Valgrind and memory leak profilers in general. And I must say it is a bit scary when you start using them cause you can't stop wondering how many leaks you might have left unsolved before! To the point, as I'm not an experienced in c++ programmer, I would like to check if this is certainly a memory leak or is it that Valgrind is doing a false positive? typedef std::vector<int> Vector; typedef std::vector<Vector> VectorVector; typedef std::map<std::string, Vector*> MapVector; typedef std::pair<std::string, Vector*> PairVector; typedef std::map<std::string, Vector*>::iterator IteratorVector; VectorVector vv; MapVector m1; MapVector m2; vv.push_back(Vector()); m1.insert(PairVector("one", &vv.back())); vv.push_back(Vector()); m2.insert(PairVector("two", &vv.back())); IteratorVector i = m1.find("one"); i->second->push_back(10); m2.insert(PairVector("one", i->second)); m2.clear(); m1.clear(); vv.clear(); Why is that? Shouldn't the clear command call the destructor of every object and every vector? Now after doing some tests I found different solutions to the leak: 1) Deleting the line i-second-push_back(10); 2) adding a delete i-second; after it's been used. 3) Deleting the second vv.push_back(Vector()); and m2.insert(PairVector("two", &vv.back())); statements. Using solution 2) makes Valgring print: 10 allocs, 11 frees Is that OK? As I'm not using new why should I delete? Thanks, for any help!

    Read the article

  • Code excavations, wishful invocations, perimeters and domain specific unit test frameworks

    - by RoyOsherove
    One of the talks I did at QCON London was about a subject that I’ve come across fairly recently , when I was building SilverUnit – a “pure” unit test framework for silverlight objects that depend on the silverlight runtime to run. It is the concept of “cogs in the machine” – when your piece of code needs to run inside a host framework or runtime that you have little or no control over for testability related matters. Examples of such cogs and machines can be: your custom control running inside silverlight runtime in the browser your plug-in running inside an IDE your activity running inside a windows workflow your code running inside a java EE bean your code inheriting from a COM+ (enterprise services) component etc.. Not all of these are necessarily testability problems. The main testability problem usually comes when your code actually inherits form something inside the system. For example. one of the biggest problems with testing objects like silverlight controls is the way they depend on the silverlight runtime – they don’t implement some silverlight interface, they don’t just call external static methods against the framework runtime that surrounds them – they actually inherit parts of the framework: they all inherit (in this case) from the silverlight DependencyObject Wrapping it up? An inheritance dependency is uniquely challenging to bring under test, because “classic” methods such as wrapping the object under test with a framework wrapper will not work, and the only way to do manually is to create parallel testable objects that get delegated with all the possible actions from the dependencies.    In silverlight’s case, that would mean creating your own custom logic class that would be called directly from controls that inherit from silverlight, and would be tested independently of these controls. The pro side is that you get the benefit of understanding the “contract” and the “roles” your system plays against your logic, but unfortunately, more often than not, it can be very tedious to create, and may sometimes feel unnecessary or like code duplication. About perimeters A perimeter is that invisible line that your draw around your pieces of logic during a test, that separate the code under test from any dependencies that it uses. Most of the time, a test perimeter around an object will be the list of seams (dependencies that can be replaced such as interfaces, virtual methods etc.) that are actually replaced for that test or for all the tests. Role based perimeters In the case of creating a wrapper around an object – one really creates a “role based” perimeter around the logic that is being tested – that wrapper takes on roles that are required by the code under test, and also communicates with the host system to implement those roles and provide any inputs to the logic under test. in the image below – we have the code we want to test represented as a star. No perimeter is drawn yet (we haven’t wrapped it up in anything yet). in the image below is what happens when you wrap your logic with a role based wrapper – you get a role based perimeter anywhere your code interacts with the system: There’s another way to bring that code under test – using isolation frameworks like typemock, rhino mocks and MOQ (but if your code inherits from the system, Typemock might be the only way to isolate the code from the system interaction.   Ad-Hoc Isolation perimeters the image below shows what I call ad-hoc perimeter that might be vastly different between different tests: This perimeter’s surface is much smaller, because for that specific test, that is all the “change” that is required to the host system behavior.   The third way of isolating the code from the host system is the main “meat” of this post: Subterranean perimeters Subterranean perimeters are Deep rooted perimeters  - “always on” seams that that can lie very deep in the heart of the host system where they are fully invisible even to the test itself, not just to the code under test. Because they lie deep inside a system you can’t control, the only way I’ve found to control them is with runtime (not compile time) interception of method calls on the system. One way to get such abilities is by using Aspect oriented frameworks – for example, in SilverUnit, I’ve used the CThru AOP framework based on Typemock hooks and CLR profilers to intercept such system level method calls and effectively turn them into seams that lie deep down at the heart of the silverlight runtime. the image below depicts an example of what such a perimeter could look like: As you can see, the actual seams can be very far away form the actual code under test, and as you’ll discover, that’s actually a very good thing. Here is only a partial list of examples of such deep rooted seams : disabling the constructor of a base class five levels below the code under test (this.base.base.base.base) faking static methods of a type that’s being called several levels down the stack: method x() calls y() calls z() calls SomeType.StaticMethod()  Replacing an async mechanism with a synchronous one (replacing all timers with your own timer behavior that always Ticks immediately upon calls to “start()” on the same caller thread for example) Replacing event mechanisms with your own event mechanism (to allow “firing” system events) Changing the way the system saves information with your own saving behavior (in silverunit, I replaced all Dependency Property set and get with calls to an in memory value store instead of using the one built into silverlight which threw exceptions without a browser) several questions could jump in: How do you know what to fake? (how do you discover the perimeter?) How do you fake it? Wouldn’t this be problematic  - to fake something you don’t own? it might change in the future How do you discover the perimeter to fake? To discover a perimeter all you have to do is start with a wishful invocation. a wishful invocation is the act of trying to invoke a method (or even just create an instance ) of an object using “regular” test code. You invoke the thing that you’d like to do in a real unit test, to see what happens: Can I even create an instance of this object without getting an exception? Can I invoke this method on that instance without getting an exception? Can I verify that some call into the system happened? You make the invocation, get an exception (because there is a dependency) and look at the stack trace. choose a location in the stack trace and disable it. Then try the invocation again. if you don’t get an exception the perimeter is good for that invocation, so you can move to trying out other methods on that object. in a future post I will show the process using CThru, and how you end up with something close to a domain specific test framework after you’re done creating the perimeter you need.

    Read the article

  • ANTS CLR and Memory Profiler In Depth Review (Part 1 of 2 &ndash; CLR Profiler)

    - by ToStringTheory
    One of the things that people might not know about me, is my obsession to make my code as efficient as possible.  Many people might not realize how much of a task or undertaking that this might be, but it is surely a task as monumental as climbing Mount Everest, except this time it is a challenge for the mind…  In trying to make code efficient, there are many different factors that play a part – size of project or solution, tiers, language used, experience and training of the programmer, technologies used, maintainability of the code – the list can go on for quite some time. I spend quite a bit of time when developing trying to determine what is the best way to implement a feature to accomplish the efficiency that I look to achieve.  One program that I have recently come to learn about – Red Gate ANTS Performance (CLR) and Memory profiler gives me tools to accomplish that job more efficiently as well.  In this review, I am going to cover some of the features of the ANTS profiler set by compiling some hideous example code to test against. Notice As a member of the Geeks With Blogs Influencers program, one of the perks is the ability to review products, in exchange for a free license to the program.  I have not let this affect my opinions of the product in any way, and Red Gate nor Geeks With Blogs has tried to influence my opinion regarding this product in any way. Introduction The ANTS Profiler pack provided by Red Gate was something that I had not heard of before receiving an email regarding an offer to review it for a license.  Since I look to make my code efficient, it was a no brainer for me to try it out!  One thing that I have to say took me by surprise is that upon downloading the program and installing it you fill out a form for your usual contact information.  Sure enough within 2 hours, I received an email from a sales representative at Red Gate asking if she could help me to achieve the most out of my trial time so it wouldn’t go to waste.  After replying to her and explaining that I was looking to review its feature set, she put me in contact with someone that setup a demo session to give me a quick rundown of its features via an online meeting.  After having dealt with a massive ordeal with one of my utility companies and their complete lack of customer service, Red Gates friendly and helpful representatives were a breath of fresh air, and something I was thankful for. ANTS CLR Profiler The ANTS CLR profiler is the thing I want to focus on the most in this post, so I am going to dive right in now. Install was simple and took no time at all.  It installed both the profiler for the CLR and Memory, but also visual studio extensions to facilitate the usage of the profilers (click any images for full size images): The Visual Studio menu options (under ANTS menu) Starting the CLR Performance Profiler from the start menu yields this window If you follow the instructions after launching the program from the start menu (Click File > New Profiling Session to start a new project), you are given a dialog with plenty of options for profiling: The New Session dialog.  Lots of options.  One thing I noticed is that the buttons in the lower right were half-covered by the panel of the application.  If I had to guess, I would imagine that this is caused by my DPI settings being set to 125%.  This is a problem I have seen in other applications as well that don’t scale well to different dpi scales. The profiler options give you the ability to profile: .NET Executable ASP.NET web application (hosted in IIS) ASP.NET web application (hosted in IIS express) ASP.NET web application (hosted in Cassini Web Development Server) SharePoint web application (hosted in IIS) Silverlight 4+ application Windows Service COM+ server XBAP (local XAML browser application) Attach to an already running .NET 4 process Choosing each option provides a varying set of other variables/options that one can set including options such as application arguments, operating path, record I/O performance performance counters to record (43 counters in all!), etc…  All in all, they give you the ability to profile many different .Net project types, and make it simple to do so.  In most cases of my using this application, I would be using the built in Visual Studio extensions, as they automatically start a new profiling project in ANTS with the options setup, and start your program, however RedGate has made it easy enough to profile outside of Visual Studio as well. On the flip side of this, as someone who lives most of their work life in Visual Studio, one thing I do wish is that instead of opening an entirely separate application/gui to perform profiling after launching, that instead they would provide a Visual Studio panel with the information, and integrate more of the profiling project information into Visual Studio.  So, now that we have an idea of what options that the profiler gives us, its time to test its abilities and features. Horrendous Example Code – Prime Number Generator One of my interests besides development, is Physics and Math – what I went to college for.  I have especially always been interested in prime numbers, as they are something of a mystery…  So, I decided that I would go ahead and to test the abilities of the profiler, I would write a small program, website, and library to generate prime numbers in the quantity that you ask for.  I am going to start off with some terrible code, and show how I would see the profiler being used as a development tool. First off, the IPrimes interface (all code is downloadable at the end of the post): interface IPrimes { IEnumerable<int> GetPrimes(int retrieve); } Simple enough, right?  Anything that implements the interface will (hopefully) provide an IEnumerable of int, with the quantity specified in the parameter argument.  Next, I am going to implement this interface in the most basic way: public class DumbPrimes : IPrimes { public IEnumerable<int> GetPrimes(int retrieve) { //store a list of primes already found var _foundPrimes = new List<int>() { 2, 3 }; //if i ask for 1 or two primes, return what asked for if (retrieve <= _foundPrimes.Count()) return _foundPrimes.Take(retrieve); //the next number to look at int _analyzing = 4; //since I already determined I don't have enough //execute at least once, and until quantity is sufficed do { //assume prime until otherwise determined bool isPrime = true; //start dividing at 2 //divide until number is reached, or determined not prime for (int i = 2; i < _analyzing && isPrime; i++) { //if (i) goes into _analyzing without a remainder, //_analyzing is NOT prime if (_analyzing % i == 0) isPrime = false; } //if it is prime, add to found list if (isPrime) _foundPrimes.Add(_analyzing); //increment number to analyze next _analyzing++; } while (_foundPrimes.Count() < retrieve); return _foundPrimes; } } This is the simplest way to get primes in my opinion.  Checking each number by the straight definition of a prime – is it divisible by anything besides 1 and itself. I have included this code in a base class library for my solution, as I am going to use it to demonstrate a couple of features of ANTS.  This class library is consumed by a simple non-MVVM WPF application, and a simple MVC4 website.  I will not post the WPF code here inline, as it is simply an ObservableCollection<int>, a label, two textbox’s, and a button. Starting a new Profiling Session So, in Visual Studio, I have just completed my first stint developing the GUI and DumbPrimes IPrimes class, so now I want to check my codes efficiency by profiling it.  All I have to do is build the solution (surprised initiating a profiling session doesn’t do this, but I suppose I can understand it), and then click the ANTS menu, followed by Profile Performance.  I am then greeted by the profiler starting up and already monitoring my program live: You are provided with a realtime graph at the top, and a pane at the bottom giving you information on how to proceed.  I am going to start by asking my program to show me the first 15000 primes: After the program finally began responding again (I did all the work on the main UI thread – how bad!), I stopped the profiler, which did kill the process of my program too.  One important thing to note, is that the profiler by default wants to give you a lot of detail about the operation – line hit counts, time per line, percent time per line, etc…  The important thing to remember is that this itself takes a lot of time.  When running my program without the profiler attached, it can generate the 15000 primes in 5.18 seconds, compared to 74.5 seconds – almost a 1500 percent increase.  While this may seem like a lot, remember that there is a trade off.  It may be WAY more inefficient, however, I am able to drill down and make improvements to specific problem areas, and then decrease execution time all around. Analyzing the Profiling Session After clicking ‘Stop Profiling’, the process running my application stopped, and the entire execution time was automatically selected by ANTS, and the results shown below: Now there are a number of interesting things going on here, I am going to cover each in a section of its own: Real Time Performance Counter Bar (top of screen) At the top of the screen, is the real time performance bar.  As your application is running, this will constantly update with the currently selected performance counters status.  A couple of cool things to note are the fact that you can drag a selection around specific time periods to drill down the detail views in the lower 2 panels to information pertaining to only that period. After selecting a time period, you can bookmark a section and name it, so that it is easy to find later, or after reloaded at a later time.  You can also zoom in, out, or fit the graph to the space provided – useful for drilling down. It may be hard to see, but at the top of the processor time graph below the time ticks, but above the red usage graph, there is a green bar. This bar shows at what times a method that is selected in the ‘Call tree’ panel is called. Very cool to be able to click on a method and see at what times it made an impact. As I said before, ANTS provides 43 different performance counters you can hook into.  Click the arrow next to the Performance tab at the top will allow you to change between different counters if you have them selected: Method Call Tree, ADO.Net Database Calls, File IO – Detail Panel Red Gate really hit the mark here I think. When you select a section of the run with the graph, the call tree populates to fill a hierarchical tree of method calls, with information regarding each of the methods.   By default, methods are hidden where the source is not provided (framework type code), however, Red Gate has integrated Reflector into ANTS, so even if you don’t have source for something, you can select a method and get the source if you want.  Methods are also hidden where the impact is seen as insignificant – methods that are only executed for 1% of the time of the overall calling methods time; in other words, working on making them better is not where your efforts should be focused. – Smart! Source Panel – Detail Panel The source panel is where you can see line level information on your code, showing the code for the currently selected method from the Method Call Tree.  If the code is not available, Reflector takes care of it and shows the code anyways! As you can notice, there does seem to be a problem with how ANTS determines what line is the actual line that a call is completed on.  I have suspicions that this may be due to some of the inline code optimizations that the CLR applies upon compilation of the assembly.  In a method with comments, the problem is much more severe: As you can see here, apparently the most offending code in my base library was a comment – *gasp*!  Removing the comments does help quite a bit, however I hope that Red Gate works on their counter algorithm soon to improve the logic on positioning for statistics: I did a small test just to demonstrate the lines are correct without comments. For me, it isn’t a deal breaker, as I can usually determine the correct placements by looking at the application code in the region and determining what makes sense, but it is something that would probably build up some irritation with time. Feature – Suggest Method for Optimization A neat feature to really help those in need of a pointer, is the menu option under tools to automatically suggest methods to optimize/improve: Nice feature – clicking it filters the call tree and stars methods that it thinks are good candidates for optimization.  I do wish that they would have made it more visible for those of use who aren’t great on sight: Process Integration I do think that this could have a place in my process.  After experimenting with the profiler, I do think it would be a great benefit to do some development, testing, and then after all the bugs are worked out, use the profiler to check on things to make sure nothing seems like it is hogging more than its fair share.  For example, with this program, I would have developed it, ran it, tested it – it works, but slowly. After looking at the profiler, and seeing the massive amount of time spent in 1 method, I might go ahead and try to re-implement IPrimes (I actually would probably rewrite the offending code, but so that I can distribute both sets of code easily, I’m just going to make another implementation of IPrimes).  Using two pieces of knowledge about prime numbers can make this method MUCH more efficient – prime numbers fall into two buckets 6k+/-1 , and a number is prime if it is not divisible by any other primes before it: public class SmartPrimes : IPrimes { public IEnumerable<int> GetPrimes(int retrieve) { //store a list of primes already found var _foundPrimes = new List<int>() { 2, 3 }; //if i ask for 1 or two primes, return what asked for if (retrieve <= _foundPrimes.Count()) return _foundPrimes.Take(retrieve); //the next number to look at int _k = 1; //since I already determined I don't have enough //execute at least once, and until quantity is sufficed do { //assume prime until otherwise determined bool isPrime = true; int potentialPrime; //analyze 6k-1 //assign the value to potential potentialPrime = 6 * _k - 1; //if there are any primes that divise this, it is NOT a prime number //using PLINQ for quick boost isPrime = !_foundPrimes.AsParallel() .Any(prime => potentialPrime % prime == 0); //if it is prime, add to found list if (isPrime) _foundPrimes.Add(potentialPrime); if (_foundPrimes.Count() == retrieve) break; //analyze 6k+1 //assign the value to potential potentialPrime = 6 * _k + 1; //if there are any primes that divise this, it is NOT a prime number //using PLINQ for quick boost isPrime = !_foundPrimes.AsParallel() .Any(prime => potentialPrime % prime == 0); //if it is prime, add to found list if (isPrime) _foundPrimes.Add(potentialPrime); //increment k to analyze next _k++; } while (_foundPrimes.Count() < retrieve); return _foundPrimes; } } Now there are definitely more things I can do to help make this more efficient, but for the scope of this example, I think this is fine (but still hideous)! Profiling this now yields a happy surprise 27 seconds to generate the 15000 primes with the profiler attached, and only 1.43 seconds without.  One important thing I wanted to call out though was the performance graph now: Notice anything odd?  The %Processor time is above 100%.  This is because there is now more than 1 core in the operation.  A better label for the chart in my mind would have been %Core time, but to each their own. Another odd thing I noticed was that the profiler seemed to be spot on this time in my DumbPrimes class with line details in source, even with comments..  Odd. Profiling Web Applications The last thing that I wanted to cover, that means a lot to me as a web developer, is the great amount of work that Red Gate put into the profiler when profiling web applications.  In my solution, I have a simple MVC4 application setup with 1 page, a single input form, that will output prime values as my WPF app did.  Launching the profiler from Visual Studio as before, nothing is really different in the profiler window, however I did receive a UAC prompt for a Red Gate helper app to integrate with the web server without notification. After requesting 500, 1000, 2000, and 5000 primes, and looking at the profiler session, things are slightly different from before: As you can see, there are 4 spikes of activity in the processor time graph, but there is also something new in the call tree: That’s right – ANTS will actually group method calls by get/post operations, so it is easier to find out what action/page is giving the largest problems…  Pretty cool in my mind! Overview Overall, I think that Red Gate ANTS CLR Profiler has a lot to offer, however I think it also has a long ways to go.  3 Biggest Pros: Ability to easily drill down from time graph, to method calls, to source code Wide variety of counters to choose from when profiling your application Excellent integration/grouping of methods being called from web applications by request – BRILLIANT! 3 Biggest Cons: Issue regarding line details in source view Nit pick – Processor time vs. Core time Nit pick – Lack of full integration with Visual Studio Ratings Ease of Use (7/10) – I marked down here because of the problems with the line level details and the extra work that that entails, and the lack of better integration with Visual Studio. Effectiveness (10/10) – I believe that the profiler does EXACTLY what it purports to do.  Especially with its large variety of performance counters, a definite plus! Features (9/10) – Besides the real time performance monitoring, and the drill downs that I’ve shown here, ANTS also has great integration with ADO.Net, with the ability to show database queries run by your application in the profiler.  This, with the line level details, the web request grouping, reflector integration, and various options to customize your profiling session I think create a great set of features! Customer Service (10/10) – My entire experience with Red Gate personnel has been nothing but good.  their people are friendly, helpful, and happy! UI / UX (8/10) – The interface is very easy to get around, and all of the options are easy to find.  With a little bit of poking around, you’ll be optimizing Hello World in no time flat! Overall (8/10) – Overall, I am happy with the Performance Profiler and its features, as well as with the service I received when working with the Red Gate personnel.  I WOULD recommend you trying the application and seeing if it would fit into your process, BUT, remember there are still some kinks in it to hopefully be worked out. My next post will definitely be shorter (hopefully), but thank you for reading up to here, or skipping ahead!  Please, if you do try the product, drop me a message and let me know what you think!  I would love to hear any opinions you may have on the product. Code Feel free to download the code I used above – download via DropBox

    Read the article

< Previous Page | 1 2 3