C++ performance vs. Java/C#

Question

My understanding is that C/C++ produces native code to run on a particular machine architecture. Conversely, languages like Java and C# run on top of a virtual machine which abstracts away the native architecture. Logically it would seem impossible for Java or C# to match the speed of C++ because of this intermediate step, however I've been told that the latest compilers ("hot spot") can attain this speed or even exceed it.

Perhaps this is more of a compiler question than a language question, but can anyone explain in plain English how it is possible for one of these virtual machine languages to perform better than a native language?

Solution 1

Generally, C# and Java can be just as fast or faster because the JIT compiler -- a compiler that compiles your IL the first time it's executed -- can make optimizations that a C++ compiled program cannot because it can query the machine. It can determine if the machine is Intel or AMD; Pentium 4, Core Solo, or Core Duo; or if supports SSE4, etc.

A C++ program has to be compiled beforehand usually with mixed optimizations so that it runs decently well on all machines, but is not optimized as much as it could be for a single configuration (i.e. processor, instruction set, other hardware).

Additionally certain language features allow the compiler in C# and Java to make assumptions about your code that allows it to optimize certain parts away that just aren't safe for the C/C++ compiler to do. When you have access to pointers there's a lot of optimizations that just aren't safe.

Also Java and C# can do heap allocations more efficiently than C++ because the layer of abstraction between the garbage collector and your code allows it to do all of its heap compression at once (a fairly expensive operation).

Now I can't speak for Java on this next point, but I know that C# for example will actually remove methods and method calls when it knows the body of the method is empty. And it will use this kind of logic throughout your code.

So as you can see, there are lots of reasons why certain C# or Java implementations will be faster.

Now this all said, specific optimizations can be made in C++ that will blow away anything that you could do with C#, especially in the graphics realm and anytime you're close to the hardware. Pointers do wonders here.

So depending on what you're writing I would go with one or the other. But if you're writing something that isn't hardware dependent (driver, video game, etc), I wouldn't worry about the performance of C# (again can't speak about Java). It'll do just fine.

One the Java side, @Swati points out a good article:

https://www.ibm.com/developerworks/library/j-jtp09275

Solution 2

JIT vs. Static Compiler

As already said in the previous posts, JIT can compile IL/bytecode into native code at runtime. The cost of that was mentionned, but not to its conclusion:

JIT has one massive problem is that it can't compile everything: JIT compiling takes time, so the JIT will compile only some parts of the code, whereas a static compiler will produce a full native binary: For some kind of programs, the static compiler will simply easily outperform the JIT.

Of course, C# (or Java, or VB) is usually faster to produce viable and robust solution than is C++ (if only because C++ has complex semantics, and C++ standard library, while interesting and powerful, is quite poor when compared with the full scope of the standard library from .NET or Java), so usually, the difference between C++ and .NET or Java JIT won't be visible to most users, and for those binaries that are critical, well, you can still call C++ processing from C# or Java (even if this kind of native calls can be quite costly in themselves)...

C++ metaprograming

Note that usually, you are comparing C++ runtime code with its equivalent in C# or Java. But C++ has one feature that can outperform Java/C# out of the box, that is template metaprograming: The code processing will be done at compilation time (thus, increasing vastly compilation time), resulting into zero (or almost zero) runtime.

I have yet so see a real life effect on this (I played only with concepts, but by then, the difference was seconds of execution for JIT, and zero for C++), but this is worth mentioning, alongside the fact template metaprograming is not trivial...

Edit 2011-06-10: In C++, playing with types is done at compile time, meaning producing generic code which calls non-generic code (e.g. a generic parser from string to type T, calling standard library API for types T it recognizes, and making the parser easily extensible by its user) is very easy and very efficient, whereas the equivalent in Java or C# is painful at best to write, and will always be slower and resolved at runtime even when the types are known at compile time, meaning your only hope is for the JIT to inline the whole thing.

...

Edit 2011-09-20: The team behind Blitz++ (Homepage, Wikipedia) went that way, and apparently, their goal is to reach FORTRAN's performance on scientific calculations by moving as much as possible from runtime execution to compilation time, via C++ template metaprogramming. So the "I have yet so see a real life effect on this" part I wrote above apparently does exist in real life.

Native C++ Memory Usage

C++ has a memory usage different from Java/C#, and thus, has different advantages/flaws.

No matter the JIT optimization, nothing will go has fast as direct pointer access to memory (let's ignore for a moment processor caches, etc.). So, if you have contiguous data in memory, accessing it through C++ pointers (i.e. C pointers... Let's give Caesar its due) will goes times faster than in Java/C#. And C++ has RAII, which makes a lot of processing a lot easier than in C# or even in Java. C++ does not need using to scope the existence of its objects. And C++ does not have a finally clause. This is not an error.

:-)

And despite C# primitive-like structs, C++ "on the stack" objects will cost nothing at allocation and destruction, and will need no GC to work in an independent thread to do the cleaning.

As for memory fragmentation, memory allocators in 2008 are not the old memory allocators from 1980 that are usually compared with a GC: C++ allocation can't be moved in memory, true, but then, like on a Linux filesystem: Who needs hard disk defragmenting when fragmentation does not happen? Using the right allocator for the right task should be part of the C++ developer toolkit. Now, writing allocators is not easy, and then, most of us have better things to do, and for the most of use, RAII or GC is more than good enough.

Edit 2011-10-04: For examples about efficient allocators: On Windows platforms, since Vista, the Low Fragmentation Heap is enabled by default. For previous versions, the LFH can be activated by calling the WinAPI function HeapSetInformation). On other OSes, alternative allocators are provided (see https://secure.wikimedia.org/wikipedia/en/wiki/Malloc for a list)

Now, the memory model is somewhat becoming more complicated with the rise of multicore and multithreading technology. In this field, I guess .NET has the advantage, and Java, I was told, held the upper ground. It's easy for some "on the bare metal" hacker to praise his "near the machine" code. But now, it is quite more difficult to produce better assembly by hand than letting the compiler to its job. For C++, the compiler became usually better than the hacker since a decade. For C# and Java, this is even easier.

Still, the new standard C++0x will impose a simple memory model to C++ compilers, which will standardize (and thus simplify) effective multiprocessing/parallel/threading code in C++, and make optimizations easier and safer for compilers. But then, we'll see in some couple of years if its promises are held true.

C++/CLI vs. C#/VB.NET

Note: In this section, I am talking about C++/CLI, that is, the C++ hosted by .NET, not the native C++.

Last week, I had a training on .NET optimization, and discovered that the static compiler is very important anyway. As important than JIT.

The very same code compiled in C++/CLI (or its ancestor, Managed C++) could be times faster than the same code produced in C# (or VB.NET, whose compiler produces the same IL than C#).

Because the C++ static compiler was a lot better to produce already optimized code than C#'s.

For example, function inlining in .NET is limited to functions whose bytecode is less or equal than 32 bytes in length. So, some code in C# will produce a 40 bytes accessor, which won't be ever inlined by the JIT. The same code in C++/CLI will produce a 20 bytes accessor, which will be inlined by the JIT.

Another example is temporary variables, that are simply compiled away by the C++ compiler while still being mentioned in the IL produced by the C# compiler. C++ static compilation optimization will result in less code, thus authorizes a more aggressive JIT optimization, again.

The reason for this was speculated to be the fact C++/CLI compiler profited from the vast optimization techniques from C++ native compiler.

Conclusion

I love C++.

But as far as I see it, C# or Java are all in all a better bet. Not because they are faster than C++, but because when you add up their qualities, they end up being more productive, needing less training, and having more complete standard libraries than C++. And as for most of programs, their speed differences (in one way or another) will be negligible...

Edit (2011-06-06)

My experience on C#/.NET

I have now 5 months of almost exclusive professional C# coding (which adds up to my CV already full of C++ and Java, and a touch of C++/CLI).

I played with WinForms (Ahem...) and WCF (cool!), and WPF (Cool!!!! Both through XAML and raw C#. WPF is so easy I believe Swing just cannot compare to it), and C# 4.0.

The conclusion is that while it's easier/faster to produce a code that works in C#/Java than in C++, it's a lot harder to produce a strong, safe and robust code in C# (and even harder in Java) than in C++. Reasons abound, but it can be summarized by:

1. Generics are not as powerful as templates (try to write an efficient generic Parse method (from string to T), or an efficient equivalent of boost::lexical_cast in C# to understand the problem)
2. RAII remains unmatched (GC still can leak (yes, I had to handle that problem) and will only handle memory. Even C#'s using is not as easy and powerful because writing a correct Dispose implementations is difficult)
3. C# readonly and Java final are nowhere as useful as C++'s const (There's no way you can expose readonly complex data (a Tree of Nodes, for example) in C# without tremendous work, while it's a built-in feature of C++. Immutable data is an interesting solution, but not everything can be made immutable, so it's not even enough, by far).

So, C# remains an pleasant language as long as you want something that works, but a frustrating language the moment you want something that always and safely works.

Java is even more frustrating, as it has the same problems than C#, and more: Lacking the equivalent of C#'s using keyword, a very skilled colleague of mine spent too much time making sure its resources where correctly freed, whereas the equivalent in C++ would have been easy (using destructors and smart pointers).

So I guess C#/Java's productivity gain is visible for most code... until the day you need the code to be as perfect as possible. That day, you'll know pain. (you won't believe what's asked from our server and GUI apps...).

About Server-side Java and C++

I kept contact with the server teams (I worked 2 years among them, before getting back to the GUI team), at the other side of the building, and I learned something interesting.

Last years, the trend was to have the Java server apps be destined to replace the old C++ server apps, as Java has a lot of frameworks/tools, and is easy to maintain, deploy, etc. etc..

...Until the problem of low-latency reared its ugly head the last months. Then, the Java server apps, no matter the optimization attempted by our skilled Java team, simply and cleanly lost the race against the old, not really optimized C++ server.

Currently, the decision is to keep the Java servers for common use where performance while still important, is not concerned by the low-latency target, and aggressively optimize the already faster C++ server applications for low-latency and ultra-low-latency needs.

Conclusion

Nothing is as simple as expected.

Java, and even more C#, are cool languages, with extensive standard libraries and frameworks, where you can code fast, and have result very soon.

But when you need raw power, powerful and systematic optimizations, strong compiler support, powerful language features and absolute safety, Java and C# make it difficult to win the last missing but critical percents of quality you need to remain above the competition.

It's as if you needed less time and less experienced developers in C#/Java than in C++ to produce average quality code, but in the other hand, the moment you needed excellent to perfect quality code, it was suddenly easier and faster to get the results right in C++.

Of course, this is my own perception, perhaps limited to our specific needs.

But still, it is what happens today, both in the GUI teams and the server-side teams.

Of course, I'll update this post if something new happens.

Edit (2011-06-22)

"We find that in regards to performance, C++ wins out by a large margin. However, it also required the most extensive tuning efforts, many of which were done at a level of sophistication that would not be available to the average programmer.

[...] The Java version was probably the simplest to implement, but the hardest to analyze for performance. Specifically the effects around garbage collection were complicated and very hard to tune."

Sources:

• https://days2011.scala-lang.org/sites/days2011/files/ws3-1-Hundt.pdf
• http://www.computing.co.uk/ctg/news/2076322/-winner-google-language-tests

Edit (2011-09-20)

"The going word at Facebook is that 'reasonably written C++ code just runs fast,' which underscores the enormous effort spent at optimizing PHP and Java code. Paradoxically, C++ code is more difficult to write than in other languages, but efficient code is a lot easier [to write in C++ than in other languages]."

– Herb Sutter at //build/, quoting Andrei Alexandrescu

Sources:

• http://channel9.msdn.com/Events/BUILD/BUILD2011/TOOL-835T
• http://video.ch9.ms/build/2011/slides/TOOL-835T_Sutter.pptx

Solution 3

Whenever I talk managed vs. unmanaged performance, I like to point to the series Rico (and Raymond) did comparing C++ and C# versions of a Chinese/English dictionary. This google search will let you read for yourself, but I like Rico's summary.

So am I ashamed by my crushing defeat? Hardly. The managed code got a very good result for hardly any effort. To defeat the managed Raymond had to:

• Write his own file I/O stuff
• Write his own string class
• Write his own allocator
• Write his own international mapping

Of course he used available lower level libraries to do this, but that's still a lot of work. Can you call what's left an STL program? I don't think so, I think he kept the std::vector class which ultimately was never a problem and he kept the find function. Pretty much everything else is gone.

So, yup, you can definately beat the CLR. Raymond can make his program go even faster I think.

Interestingly, the time to parse the file as reported by both programs internal timers is about the same -- 30ms for each. The difference is in the overhead.

For me the bottom line is that it took 6 revisions for the unmanaged version to beat the managed version that was a simple port of the original unmanaged code. If you need every last bit of performance (and have the time and expertise to get it), you'll have to go unmanaged, but for me, I'll take the order of magnitude advantage I have on the first versions over the 33% I gain if I try 6 times.

Solution 4

The compile for specific CPU optimizations are usually overrated. Just take a program in C++ and compile with optimization for pentium PRO and run on a pentium 4. Then recompile with optimize for pentium 4. I passed long afternoons doing it with several programs. General results?? Usually less than 2-3% performance increase. So the theoretical JIT advantages are almost none. Most differences of performance can only be observed when using scalar data processing features, something that will eventually need manual fine tunning to achieve maximum performance anyway. Optimizations of that sort are slow and costly to perform making them sometimes unsuitable for JIT anyway.

On real world and real application C++ is still usually faster than java, mainly because of lighter memory footprint that result in better cache performance.

But to use all of C++ capability you, the developer must work hard. You can achieve superior results, but you must use your brain for that. C++ is a language that decided to present you with more tools, charging the price that you must learn them to be able to use the language well.

Solution 5

JIT (Just In Time Compiling) can be incredibly fast because it optimizes for the target platform.

This means that it can take advantage of any compiler trick your CPU can support, regardless of what CPU the developer wrote the code on.

The basic concept of the .NET JIT works like this (heavily simplified):

Calling a method for the first time:

• Your program code calls a method Foo()
• The CLR looks at the type that implements Foo() and gets the metadata associated with it
• From the metadata, the CLR knows what memory address the IL (Intermediate byte code) is stored in.
• The CLR allocates a block of memory, and calls the JIT.
• The JIT compiles the IL into native code, places it into the allocated memory, and then changes the function pointer in Foo()'s type metadata to point to this native code.
• The native code is ran.

Calling a method for the second time:

• Your program code calls a method Foo()
• The CLR looks at the type that implements Foo() and finds the function pointer in the metadata.
• The native code at this memory location is ran.

As you can see, the 2nd time around, its virtually the same process as C++, except with the advantage of real time optimizations.

That said, there are still other overhead issues that slow down a managed language, but the JIT helps a lot.

View more solutions

Comments

[Tony BenBrahim]

The problem is that for long running processes, such as a web server, your memory over time will become so fragmented (in a C++ written program) that you will have to implement something that resembles garbage collection (or restart every so often, see IIS).

[Brian R. Bondy]

By the way Jonathan, I think someone is still downvoting your things. When I voted you up you had a -1 on this post.

[Brian R. Bondy]

I also believe that there is a certain language weight as well, when working at a lower level, with less abstraction, you will be developing a program that is faster. This is unrelated the points about the bytecode execution itself.

[paercebal]

Or if you're using a better C++ allocator and/or pool of objects. This is far from magic, from a C++ view point, and it can boils down to have "heap allocation" become as fast a stack allocation.

[cero]

Re item 2, "2/ Vast portions of the Java libraries are compiled so that, when you call a library function, you're executing compiled code, not interpreted": Do you have a citation for that? If it were really as you describe, I'd expect to run into native code from my debugger a lot, but I don't.

[Guvante]

Re: cero Debuggers often utilize less efficient but more expressive paths, and are therefore not a good marker for anything performance related.

[Bill K]

It's not so much that you are compiling for CPU optimization, but you are compiling for runtime path optimization. If you find that a method is very often called with a specific parameter, you could pre-compile that routine with that parameter as a constant which could (in the case of a boolean that controls flow) factor out gigantic chunks of work. C++ cannot come close to doing that kind of optimization.

[David Thornley]

So how do JITs do at recompiling routines to take advantage of observed runpaths, and how much difference does that make?

[David Thornley]

I haven't observed that on the big Unix programs that are meant to run forever. They tend to be written in C, which is even worse for memory management than C++.

[David Thornley]

I suspect that the saying "You can write FORTRAN in any language" predates Jeff's career.

[Liran Orevi]

There is another huge performance gain to This HUGH library - library code is probably better written than what many programmers will write on their own (given a limited time, and lack of specialty knowledge) and on Java, because of many reasons, programmers often use the library.

[kyoryu]

Of course, the question is whether we're comparing an implementation of a program in managed vs. unmanaged code, or the theoretical top performance of the language. Clearly, unmanaged code can always be at least as fast as managed, as in the worst case you could just write an unmanaged program that did exactly the same thing as the managed code! But most performance issues are algorithmic, not micro. Also, you don't optimize managed and unmanaged code the same way, so "C++ in C#" is usually going to not work well.

[Frunsi]

Well, I think he has a good idea of its drawbacks, he also said: "C makes it easy to shoot yourself in the foot; C++ makes it harder, but when you do it blows your whole leg off" ;)

[Frunsi]

In C/C++ you can allocate short lived objects on the stack, and you do when its appropriate. In managed code you cannot, you have no choice. Also, in C/C++ you can allocate lists of objects in contigous areas (new Foo[100]), in managed code you cannot. So, your comparison is not valid. Well, this power of choices places a burden on the developers, but this way they learn to know the world they live in (memory......).

[Frunsi]

If you'd always allocate everything on the heap, then .NET and Java may even perform better than C/C++. But you just will not do this in C/C++.

[Hardy]

@Bill I may be mixing two things... but isn't branch prediction done at run time in the instruction pipeline achieve similar goals independent of the language?

[Bill K]

@Hardy yes, the CPU can do branch prediction regardless of language, but it cannot factor out an entire loop by observing that the loop has no effect on anything. It will also not observe that mult(0) is hard-wired to return 0 and just replace the entire method call with if(param == 0) result=0; and avoid the entire function/method call. C could do these things if the compiler had a comprehensive overview of what was happening, but generally it doesn't have enough info at compile time.

[toc777]

"Do they require a huge library installed" Java is addressing this problem with project jigsaw I believe.

[gjvdkamp]

link is dead, found mentioned article here: blogs.msdn.com/b/ricom/archive/2005/05/10/416151.aspx

[paercebal]

Could you expand on the assertion Since hand-tuning Assembly is almost always faster than C++ compiled code? What do you mean by "hand-tuning Assembly" and "C++ compiled code"?

[jciconsult]

Well, it's based on the idea that a compiler's optimizer follows rules, and coders don't. So there's always going to be code that the optimizer finds it can't optimize perfectly, whereas a human could, either by looking at a bigger picture or knowing more about what the code really is doing. I'll add that this is a 3 year old comment, and I know more about HotSpot than I used to, and I can easily see dynamic optimization being a VERY nice way to get code running faster.

[paercebal]

1. Optimizations from Hotspot or any other JIT are still compiler optimizations. JIT have the advantage over a static compiler of being able to inline some results (code frequently called), or even to make optimizations based on the executing processor, but it is still a compiler optimization. . . 2. I guess you are talking about algorithm optimization, not "assembly fine tuning". "manual assembly fine tuning by a human coder" failed to produce better results than compiler optimizations since more than a decade. In fact, a human playing with assembly usually screw away any optimization...

[jciconsult]

Ok, I get that I'm using the wrong terminology, "compiler optimiziation" rather than "static optimization". I would point out that, at least in the gaming industry, as recently as for the PS2 we were still using hand-coded assembly in places to "optimize" for the specific chips we knew were on the console; cross-compilers for these new chips not being yet as sophisticated as those for x86 architectures. Back to the original question above: the JIT has the benefit of being able to measure prior to optimizing, which is a Good Thing(TM)

[Totti]

Note that most production GCs also use hand-written assembler because C/C++ don't cut it.

[Totti]

Note that free and delete are not deterministic either and GC can be made deterministic by not allocating.

[Totti]

"In C++ you get what you pay for, not more, not less". Counter example: I benchmarked an RB tree implementation in OCaml and C++ (GNU GCC) that used an exception to long jump out from the recursion if an element being added was already present in order to reuse the existing set. OCaml was up to 6x faster than C++ because it does not pay for checking for destructors as the stack is unwound.

[Totti]

@frunsi: "in C/C++ you can allocate lists of objects in contigous areas (new Foo[100]), in managed code you cannot". That is incorrect. Local value types are stack allocated and you can even stack allocate arrays of them in C#. There are even production systems written in C# that are completely allocationless in the steady state.

[Frunsi]

@Jon: but at some (later?) point in time it has to destruct the objects anyway (at least it has to release its memory). And also note, that exceptions are for exceptional cases, at least in C++ that rule should be respected. C++ exceptions may be heavy when exceptions occur, that is a trade-off.

[Frunsi]

@Jon: Ok, true for value types and arrays of value types. But I wrote about objects (in C# speech we talk about all kinds of reference types). In an OOP world they are where the fun starts, and C# heap allocates all those objects.

[Frunsi]

@Jon: maybe try to repeat your benchmark with times on a shell. So that it checks the whole program, not just a single aspect. Are the results similar then?

[Женя Борісевіч]

Your reasoning is bogus - C++ programs get built for their target architecture, they don't need to switch at runtime.

[Voo]

@Justicle The best your c++ compiler will offer for different architectures is usually x86, x64, ARM and whatnot. Now you can tell it to use specific features (say SSE2) and if you're lucky it'll even generate some backup code if that feature isn't available, but that's about as fine-grained as one can get. Certainly no specialization depending on cache sizes and whatnot.

[Женя Борісевіч]

See shootout.alioth.debian.org/u32/… for examples of this theory not happening.

[Sebastian Mach]

You edit after 5 months of C# describes exactly my own experience (templates better, const better, RAII). +1. Those three remain my personal killer features for C++ (or D, which I hadn't the time for, yet).

[Sebastian Mach]

To be honest, this is one of the worst answers. It is so unfounded, I could just invert it. Too much generalisation, too much unknowledge (optimizing away empty functions is really just the tip of the iceberg). One luxury C++ compilers have: Time. Another luxury: No checking is enforced. But find more in stackoverflow.com/questions/145110/c-performance-vs-java-c/… .

[Totti]

"The code processing will be done at compilation time". Hence template metaprogramming only works in the program is available at compile time which is often not the case, e.g. it is impossible to write a competitively performant regular expression library in vanilla C++ because it is incapable of run-time code generation (an important aspect of metaprogramming).

[Totti]

"playing with types is done at compile time...the equivalent in Java or C# is painful at best to write, and will always be slower and resolved at runtime even when the types are known at compile time". In C#, that is only true of reference types and is not true for value types.

[Totti]

"No matter the JIT optimization, nothing will go has fast as direct pointer access to memory...if you have contiguous data in memory, accessing it through C++ pointers (i.e. C pointers... Let's give Caesar its due) will goes times faster than in Java/C#". People have observed Java beating C++ on the SOR test from the SciMark2 benchmark precisely because pointers impede aliasing-related optimizations. blogs.oracle.com/dagastine/entry/sun_java_is_faster_than

[Totti]

Also worth noting that .NET does type specialization of generics across dynamically-linked libraries after linking whereas C++ cannot because templates must be resolved before linking. And obviously the big advantage generics have over templates is comprehensible error messages.

[Johnydep]

you are defending C++ because you know everyone is switching to JAVA. All academic universities, research institutes have java as their preferred simulation/development environment, and if you go to production level development, again C# is most widely used specially for embedded systems. Servers are shifting to Java EE, C++ would probably remain the choice for desktop application developers only targeting naive end users.

[Orion Adrian]

@Justicle I went there and looked at the C# Mono code which looked like a C++ programmer went and wrong C# code in C++ style. Of course it's not going to behave or even be as maintainable as C# code. But I can write a lot of really complicated stuff that's relatively efficient in only a few lines of code in C# because it offers those high-level features. Besides most of the C# code out there isn't going to need to do very fast math processing or manual register manipulation. All these comparisons feel weighted to what C++ does well and what the others languages do poorly. The tests are biased.

[Orion Adrian]

There seems to be a misconception or lack of knowledge of when the C# compiler does a lot of its work. The C# compiler does most of the hard lifting at build time. IL is incredibly easy to work with and optimize which is why the supposed benefit of infinite time for the C++ compilers just isn't borne out as much as some would claim. In the end you do loose some speed for protection features (e.g. index bounds checking), but the really powerful structures you gain out of it along with the simple nature of the language usually pays off in increased productivity and targeted optimization.

[Женя Борісевіч]

SO @OrionAdrian - 1. The C# code in the tests is wrong, or 2. C# doesn't need to be fast. Which is it?

[Orion Adrian]

@Justicle Pure raw performance isn't the only consideration, otherwise we'd all be hand-coding perfect assembly. The question for me has always been, given X time to develop and Y-sized task, what will give me the best performance Z. And depending on what I'm doing, different languages will have the best Z. C# doesn't try to directly compete with C++ for being the fastest at heavy math processing -- why would it, that's what C++ is for. C# has a decent core, but adds fast, safe metadata programming, reflection, database access, ORM, MVC and easy to use Windows controls. It does those fast.

[Женя Борісевіч]

@OrionAdrian The idea that it's easier or more appropriate to use other languages is irrelevant to the question.

[Orion Adrian]

@Justicle The question asks how can code run faster with the abstraction layer of the VM. For one, key, basic algorithms can be simpler in the C# world because you get to make assumptions about a lot of things, like memory (de-)allocation. In general, the more assumptions you can make about something the faster you can make it. Algorithm choice is going to have a larger effect on performance than anything and the benefit to having so many core parts of the experience done for you is that those can be super-optimized. The C# allocator is really fast along with strings and file I/O and others.

[Женя Борісевіч]

@OrionAdrian ok we're full circle now ... See shootout.alioth.debian.org/u32/… for examples of this theory not happening. In other words, show us that your theory can be proven correct before making vague speculative statements.

[Tavison]

RAII and Generic programming are the critical missing pieces in Java and C#. Your answer matches my experience. Multi pass full code optimization is also very hard to beat.

[stinky472]

We're all speaking very conceptually though. In practice, one would be hard-pressed to find something like an efficient raytracer implemented in managed languages like Java or C#, and there's a reason mobile devices are turning back to C/C++ for native applications (Android switching more to C/C++ instead of Java, MS from C# to C/C++, etc). Embedded systems are very much in the same vein. That said, these are all valid points are JIT performance, but we also have to keep in mind the language-level overhead of managed languages. For example, in C/C++, my user-defined types can be allocated...

[stinky472]

... on the stack (and often are except for variable-sized aggregates or objects that need to persist outside of some immediate scope). Regardless of the fact that Java and C# have that additional level of indirection where the GC is allowed to compress the heap, we're in turn generally using the heap a lot more in those languages and a heap allocation is still generally hundreds of cycles compared to a few for the stack. We're also paying a heavier cost for the indirection and super performance-critical apps generally need everything to fit in the cache, e.g. - not something easy in Java.

[stinky472]

First of all, if we look at Raymond Chen's code, he clearly does not understand C++ or data structures very well. His code almost reaches straight for low-level C code even in cases where the C code has no performance benefits (it just seems to be a sort of distrust and maybe a lack of knowledge of how to use profilers). He also failed to understand the most algorithmically sound way of implementing a dictionary (he used std::find for Christ's sake). If there's something good about Java, Python, C#, etc. - they all provide very efficient dictionaries...

[stinky472]

Tries or even std::map would fare much more favorably towards C++ or even a hash table. Finally, a dictionary is exactly the type of program that benefits most from high-level libraries and frameworks. It doesn't demonstrate differences in the language so much as the libraries involved (of which, I would happily say that C# is far more complete and provides much more tools suited for the task). Show a program that manipulates large blocks of memory in comparison, like a large-scale matrix/vector code. That'll settle this quite quickly even if, as in this case, the coders don't know what...

[stinky472]

We rarely see people dealing with managed languages talking about things like the cache line or memory access patterns because everything is all over the heap. Efficient code of that sort is generally impractical for anything but arrays of plain old data. We can't do it with, say, a matrix library or hand-coded raster operations, pixel filters, raytracers, physics simulators, particle engines, realtime motion tracking, video compositing, audio processing, etc. On the other hand, if it's business applications we're talking about, e.g., or database middleware, I wouldn't be surprised to find...

[stinky472]

... the average Java or C# or even Python application outperforming the average C++ equivalent since, 1) most of these are actually implemented in C/C++ anyway (the JNI libraries, native implementations of .NET, or Python C modules involved), and 2) the reasons cited above about the benefits of JIT compilation and 3) these languages optimize productivity above all else, and the programmer who gets things done faster just has more time to spend optimizing and fixing bugs.

[Orion Adrian]

@stinky472 DotNet uses the stack quite effectively actually. Registers too. Any custom type that uses a struct will also be placed on the stack (though that's an implementation issue). DotNet does keep track of cachelines and the JIT compiler is smart about these kinds of resources.

[stinky472]

@OrionAdrian Efficient cache line usage comes down to contiguous memory. If you have a ArrayList of particles, a fundamental aspect of whether that can be efficiently traversed is whether it is, indeed, contiguous. By definition, if each particle is created with operator new, it is on the managed heap - efficiency goes out the door. Only types inheriting from System.ValueType have this special CLR characteristic of not being allocated on the managed heap. That's, by far, not the majority of types, and I've seen too many people trying to create something like particle instances on the heap.

[stinky472]

@OrionAdrian That's not to disagree entirely about conceptual differences and implementation-defined issues. But the fact that operator new creates data on the managed heap (not the hardware stack) is not implementation-defined, that's how new is defined in .NET. And this is a fundamental point of where managed languages tend to perform poorly in comparison to something like C or FORTRAN. Those latter languages excel at working with large, contiguous buffers of data in performance-critical loops: areas like video processing.