I have some big mysql databases with data for calculations and some parts where I need to get data from external websites.
I used python to do the whole thing until now, but what shall I say: its not a speedster.
Now I'm thinking about mixing Python with C++ using Boost::Python and Python C API.
The question I've got now is: what is the better way to get some speed.
Shall I extend python with some c++ code or shall I embedd python code into a c++ programm?
I will get fore sure some speed increment using c++ code for the calculating parts and I think that calling the Python interpreter inside of an C-application will not be better, because the python interpreter will run the whole time. And I must wrap things python-libraries like mysqldb or urllib3 to have a nice way to work inside c++.
So what whould you suggest is the better way to go: extending or embedding?
( I love the python language, but I'm also familiar with c++ and respect it for speed )
Update:
So I switched some parts from python to c++ and used multi threading (real one) in my c modules and my programm now needs instead of 7 hours 30 minutes :))))
In principle, I agree with the first two answers. Anything coming from disk or across a network connection is likely to be a bigger bottleneck than the application.
All the research of the last 50 years indicates that people often have inaccurate intuition about system performance issues. So IMHO, you really need to gather some evidence, by measuring what is actually happening, then chose a solution based on that evidence.
To try to confirm what is causing the slow performance, measure the system and user time of your application (e.g time python prog.py), and measure the load on the machine.
It the application is maxing-out the CPU, and most of that time is spent in the application (user time), then there may be a case for using a more effective technology for the application.
But if the CPU is not maxed, or the application spends most of its time in the system (system time), and not in the application (user time), then it is unlikely that changing the application programming technology will help significantly. (This is an example of Amdahl's Law http://en.wikipedia.org/wiki/Amdahl%27s_law)
You may also need to measure the performance of your database server, and maybe network connection, to identify the source of the bottle neck, but start with the easiest part.
In my opinion, in your case it makes no sense to embed Python in C++, while the reverse could be beneficial.
In most of programs, the performance problems are very localized, which means that you should rewrite the problematic code in C++ only where it makes sense, leaving Python for the rest.
This gives you the best of both world: the speed of C++ where you need it, the ease of use and flexibility of Python everywhere else. What is also great is that you can do this process step by step, replacing the slow code paths by the by, leaving you always with the whole application in an usable (and testable!) state.
The reverse wouldn't make sense: you'd have to rewrite almost all the code, sacrificing the flexibility of the Python structure.
Still, as always when talking about performance, before acting measure: if your bottleneck is not CPU/memory bound switching to C++ isn't likely to produce much advantages.
Related
I know how to program in Python but I am also interested in learning C++. I have heard that it is much faster than python and for the programs I am writing currently, I would prefer them to run as quickly and efficiently as possible. I know that a lot of that comes from just writing good code but I was also wondering if using another language, such as C++, would help.
While I was pondering this, I realized that since most of my programs will be mainly using the internet (as in implementing Google APIs and using the information from them to submit data to other websites) then maybe the speed of the language doesn't matter if the speed of my internet connection is always going to be relatively the same. I have two ways I am connecting to the internet: Selenium (or some kind of automated browser) for things that require a browser, and just HTTP requests.
How much difference would I see between python and a different language even though the major focus of my programs is on the internet?
Thanks.
Scenarios
The main benefit you would get from using language that is compiled to machine code is that you can do lots of byte and bit-magic. Lets say, modifying image data, transforming audio, analysing indices of a genomic sequence database.
Typical tasks
Serving web-pages you typically have problems if a completely different sort: You will be loading a resource from hard disk, serve them directly if its an image or audio, or you will be executing different transformation steps on a text resource until it becomes the final HTML document. The latter will be using template engines, database queries, and so on.
If you look at that you can see that most of the things, say 90-99% are pretty high-level stuff -- in Python you will use an API that is optimized by many, many users for optimal performance (meaning: time and space). "Open a file" will be almost as fast in C as it is in Python, so is reading from it and serving it to some Socket. Transforming text data could be a bit faster in C++ then it is in Python, but... how fast does it have to be? A use is very likely willing to wait 200ms, isnt't he? And that is a lot of time for a nice high-level template engine to transform a bit of text.
What C++ and Python can do for you
A typical Python web-service is much faster to write and a easier to deploy then a server written in C++. If you would do it in C++ you firstly need to handle sockets and connections -- and for those would either use an existing library or write your own handling. If you use an existing library (which I strongly recommend) you are basically not doing anything differently then Python does. If you write your own handling, you have many, many low-level things you can do wrong that will burn the performance you wish for. No, that is not an option.
If you need speed, and Python and the server and template framework is not enough you should re-think your architectural approach. Then take a look at the c10k-problem and write tiny pieces in C. (Look at this c10k very hot topic, too) But I can not see many reasons not to use a high-level language like Python, if you are only looking for performance in a medium-complex web-serving application.
Summary: The difference
If you are just serving files from the hard-drive I guess your Python program would even be faster then your hand-crafted C++-server. If you use a framework written in C or C++ and just drop in your static pages, I guess you get a boost like 2-5fold against Python. Then again, if your web-application is a bit more complex then serving static content, I estimate that the difference will diminish very quickly and you will get 1-2fold speed gain at most.
It's not all about speed...
One note about another difference between C++ and Python one should not forget: Since C++ is really compiled and not as dynamic as Python you would gain a lot of static error analysis by using Python. Writing correct code is always difficult, but can be done in C++ and Python with good tests and static analysis -- the latter is simpler in C++ (my opinion). If that is an issue for you, you may think again, but you asked about speed.
I have to write a server program that implements some fuzzy logic and I choose to write it in Node.js to take advantage of its event orientation.
I have to work with difficult mathematic computational problem, and I don't know what's the best way to obtain performance:
Write all in Node.js and use the power of the V8 engine for the mathematical task.
Write a module in C++ that implements all the mathematical function and call it from Node.
Anyone that have experience in these type of computation on both platform?
Since you need the Node.js part anyway, go ahead, implement everything in Node.js. If it is fast enough, this is easy enough to maintain. It's very hard to predict the power of a virtual machine / JIT compiler.
If it is not fast enough, first think about algorithmic improvements. If this doesn't help and if profiling shows that the computation is the problem, go ahead, re-implement it in C++. But be aware that writing performant C++ code is not trivial. Make sure that you have a good profiler at hand and measure often.
In general I'd say C++ code is faster if written correctly. The tricky part is to write it correctly. Please check this article Google Paper on C++, Java, Scala, Go for more information. The gist is - managed languages make it a lot easier to write & maintain the code but if you need raw performance, C++ is the best. But it comes at the price of needing a lot of expertise and that the code is harder to maintain.
denshade, your C implementation goes only to 2e5 not 2e6, like you've done for js (linking to today's revs on Github):
primes.c
primes.js
Piping to /dev/null, and changing js also to 2e5, I get about 6.5 seconds for C and about 8.5 seconds for js (using some version of node) on my current computer.
Since your algorithm is O(n^2), I would expect 2e6 to take some 15 minutes, not 15 hours, but I haven't tried it. Maybe it does fall apart that bad for some reason.
(Note that I couldn't comment directly, since I'm brand new on SO and have no rep.)
it's pretty much impossible to answer this kind of question. The answer as always for these things is it depends on your skills and how much time and effort you are willing to put into it.
C++ always has the potential to be faster and more efficient as you have much closer control over all the things that matter. The downside it that you have to do all the things that matter and the generic implementations in the other language are probably done by someone who knows what they are doing and could well be better than a naive or quick implementation in c++
Plus often you'll find that the bottleneck isn't what you think it will be anyway, for example if reading in your data turns out to take 20 times as long as the calculations which isn't impossible then it hardly matters how fast the calculations are. And intuition about where the bottlenecks lie is often badly wrong even for experienced developers.
http://benchmarksgame.alioth.debian.org/u64q/compare.php?lang=node&lang2=gpp
The above link is dead and now in the wayback--
https://web.archive.org/web/20180324192118/http://benchmarksgame.alioth.debian.org/u64q/compare.php?lang=node&lang2=gpp
C++ uses the CPU and performs up to 10X faster than Node.js doing mathematical operations.
That site moved over here
https://benchmarksgame-team.pages.debian.net/benchmarksgame/which-programs-are-fastest.html
One thing to consider with going the C++ route for complex mathematical computations is that you might be able to leverage an existing high performance library, such as BLAS, LAPACK, ARMA, etc.. where other developers have already put significant time and effort into providing highly optimized functionality. I doubt you'll find a similar level of high performance library for JavaScript. Certainly if you have an identified bottleneck around matrix calculations or linear algebra, one of these C++ libraries is the way to go.
I've runned #denshade codes removing prints and the timing on 100000 numbers are exceptional:
3 sec. for nodejs!
6 sec. for gcc/clang compiled c
6 sec. for hhvm ( php )
14 sec for php7 w/ opcache
15 sec for php7 w/o opcache
Nodejs is so fast because it's compiled and optimized over-time.
so, maybe you just need to test by your self which is the best language that fits your needs in this case.
If your calculations aren't trivial I'd like to issue a warning. JavaScript is very bad when you are going for heavy calculations. My story involves a simple prime program which you can find here: https://github.com/denshade/speedFun
Long story short. I created a simple, be it inefficient prime check function implemented in C & JavaScript. Both are implemented in the same way. The first 2000 000 primes are verified in 5 seconds in C. The same function in javascript lasted for more than 16 hours when run in node.js.
Following are the areas where Node.js is proving itself as a perfect technology
partner.
● I/O bound Applications
● Data Streaming Applications
● Data Intensive Real-time Applications (DIRT)
● JSON APIs based Applications
● Single Page Applications
It is not advisable to use Node.js for CPU intensive applications.
here is API comparisons :
https://www.linkedin.com/pulse/nodejs-vs-java-which-faster-apis-owen-rubel
I'm developing a C++ application that is extended/ scriptable with Python. Of course C++ is much faster than Python, in general, but does that necessarily mean that you should prefer to execute C++ code over Python code as often as possible?
I'm asking this because I'm not sure, is there any performance cost of switching control between code written in C++ and code written in Python? Should I use code written in C++ on every occasion, or should I avoid calling back to C++ for simple tasks because any speed gain you might have from executing C++ code is outmatched by the cost of switching between languages?
Edit: I should make this clear, I'm not asking this to actually solve a problem. I'm asking purely out of curiosity and it's something worth keeping in mind for the future. So I'm not interested in alternative solutions, I just want to know the answer, from a technical standpoint. :)
I don't know there is a concrete rule for this, but a general rule that many follow is to:
Prototype in python. This is quicker to write, and may be easier to read/reason about.
Once you have a prototype, you can now identify the slow portions that should be written in c++ (through profiling).
Depending on the domain of your code, the slow bits are usually isolated to the 'inner loop' types of code, so the number of switches between python an this code should be relatively small.
If your program is sufficiently fast, you've successfully avoided prematurely optimizing your code by writing too much in c++.
Keep it simple and tune performance as needed. The primary reason for embedding an interpreter in a C++ app is to allow run-time configuration/data to specify some processing - i.e. you can modify the script without recompiling the C++ program - that's your guide for when to call into the interpreter. Once in some interpreter call, the primary reasons to call back into C++ are:
to access or update some data that can't reasonably be exposed as a parameter to the call (or via some other registration process the interpreter supports)
to get better performance during some critical part of the processing
For the latter, try the script first (assuming it's as easy to develop there), then if it's slow identify where and how some C++ code might help. If/where performance does prove a problem - as a general guideline when calling from C++ to the interpreter or vice versa: try to line up as much work as possible then make the call into the other system. If you get stuck, come back to stackoverflow with a specific problem and actual code.
The cost is present but negligible. That's because you probably do a fair bit of work converting python's high level datatypes to C++-compatible representations. Of course this is similar to the cost of calling one C++ function from another, there's some overhead. The rules for when it's a good idea to switch from python to C++ are:
A function with few arguments
A function which does a large amount of processing on a small amount of data
A function which is called as rarely as possible - consolidate function calls if possible
The best metric should be something that wieghs up for you....
Makes development, debugging and testing easier (lowers dev cost)
Lowers the cost of maintenance
meets the performance requirement (provides solution)
I'm asking the question after reading this article
http://stevehanov.ca/blog/index.php?id=95
Also isn't it a penalty to use cgi instead of fastcgi ?
Update: why some people do pretend like in answer "that you get 20-30% performance improvement" ? Is it pure guess or is this number coming from solid benchmark ? I have looked at HipHop performance is more in the scale of 10 times.
I've done webdev in a few languages and frameworks, including python, php, and perl. I host them myself and my biggest sites get around 20k hits a day.
Any language and framework that has reasonable speed can be scaled up to take 20k hits a day just by throwing resources at it. Some take more resources than others. (Plone, Joomla. I'm looking at you).
My Witty sites (none in production yet) take a lot more (from memory around 5000% more) pounding (using seige) than for example my python sites. Ie. When I hit them as hard as I can with seige, the witty sites serve a lot more pages per second.
I know it's not a true general test though.
Other speed advantages that witty gives you:
Multi threading
If you deploy with the built in websrever (behind ha-proxy for example) and have your app be multi-threaded .. it'll load a lot less memory than say a perl or php app.
Generally with php and perl apps, you'll have Apache fire up a process for each incoming connection, and each process loads the whole php interpreter, all the code and variables and objects and what not. With heavy frameworks like Joomla and Wordpress (depending on the number of plugins), each process can get pretyy humungous on memory consumption.
With the Wt app, each session loads a WApplication instance (a C++ object) and it's whole tree of widgets and stuff. But the memory the code uses stays the same, no matter how many connections.
The inbuilt Web2.0 ness
Generally with traditional apps, they're still built around the old 'http request comes in' .. 'we serve a page' .. 'done' style of things. I know they are adding more and more AJAXy kind of thigns all the time.
With Wt, it defaults to using WebSockets where possible, to only update the part of the page that needs updating. It falls back to standard AJAX, then if that's not supported http requests. With the AJAX and WebSockets enabled clients, the same WApplication C++ object is continually used .. so no speed is lost in setting up a new session and all that.
In response to the 'C++ is too hard for webdev'
C++ does have a bit of a learning curve. In the mid nineties we did websites in Java j2ee. That was considered commercially viable back then, and was a super duper pain to develop in, but it did have a good advantage of encouraging good documentation and coding practices.
With scripting websites, it's easy to take shortcuts and not realize they're there. For example one 8 year old perl site I worked on had some code duplicated and nobody noticed. Each time it showed a list of products, it was running the same SQL query twice.
With a C++ site, I think it'd have less chance because, in the perl site, there wasn't that much programming structure (like functions), it was just perl and embedded html. In C++ you'd likely have methods with names and end up with a name clash.
Types
One time, there was a method that took an int identifier, later on we changed it to a uuid string. The Python code was great, we didn't think we needed to change it; it ran fine. However there was little line buried deep down that had a different effect when you passed it a string. Very hard to track down bug, corrupted the database. (Luckily only on dev and test machines).
C++ would have certainly complained a lot, and forced us to re-write the functions involved and not be lazy buggers.
With C++ and Java, the compiler errors and warns a lot of those sorts of mistakes for you.
I find unit testing is generally not as completely necessary with C++ apps (don't shoot me), compared to scripting language apps. This is due to the language enforcing a lot of stuff that you'd normally put in a unit test for say a python app.
Summary
From my experience so far .. Wt does take longer to develop stuff in than existing frameworks .. mainly because the existing frameworks have a lot more out of the box stuff there. However it is easier to make extremely customized apps in Wt than say Wordpress imho.
From people I've spoken with who've moved from PHP to Wt (a C++ web framework) reported significant improvements. From the small applications I've created using Wt to learn it, I've seen it run faster than the same PHP type applications I created. Take the information for what you will, but I'm sold.
This reminds me how 20-30 years ago people were putting Assembly vs C, and then 10-20 years ago C vs C++. Of course C++ will be faster than PHP/Rails but it'll take 5x more effort to build maintainable and scalable application.
The point is that you get 20-30% performance improvement while sacrificing your development resources. Would you rather have you app work 30% faster or have 1/2 of the features implemented?
Most web applications are network-bound instead of processor-bound. Writing your application in C++ instead of a higher-level language doesn't make much sense unless you're doing really heavy computation. Also, writing correct C++ programs is difficult. It will take longer to write the application and it is more likely that the program will fail in spectacular ways due to misused pointers, memory errors, undefined behavior, etc. In general, I would say it is not worth it.
Whenever you eliminate a layer of interpretive or OS abstraction, you are bound to get some performance gain. That being said, the language or technology itself does not automatically mean all your problems are solved. I've fixed C++ code that took many hours to process a relatively simple set of records. The problem was in the implementation, and the fix was not related to the language's features or limitations.
Assuming things are all implemented correctly, you're sure to get better performance. The problem will be in finding the bugs. One of the problems with C++ is that many developers are currently "trained" or accustomed to having a lot of details related to memory management behind objects. This eliminates the need to consider things like, "What can happen if I pass this pointer around to several threads?" Sometimes it works well, but not always. You still have some subtleties of the language that you need to consider regardless of how the objects hide the nasty details.
In my experience, you'll need several seasoned C++ developers watching over the code to be able to keep the bugs and memory leaks from getting out of hand.
I'm certainly not sold on this. If you want a performance gain over PHP why not use a Java (or better yet Scala) framework? These are much better for web development, have nice, relatively easy to use frameworks and avoid a lot of the headaches of C++. I've always seen one of the main pluses of web-development (and most modern non-scientific/high performance applications) as being able to avoid the headaches that come along with C/C++ development.
I work in a group that does a large mix of research development and full shipping code.
Half the time I develop processes that run on our real time system ( somewhere between soft real-time & hard real-time, medium real-time? )
The other half I write or optimize processes for our researchers who don't necessarily care about the code at all.
Currently I'm working on a process which I have to fork into two different branches.
There is a research version for one group, and a production version that will need to occasionally be merged with the research code to get the latest and greatest into production.
To test these processes you need to setup a semi complicated testing environment that will send the data we analyze to the process at the correct time (real time system).
I was thinking about how I could make the:
Idea
Implement
Test
GOTO #1
Cycle as easy, fast and pain free as possible for my colleagues.
One Idea I had was to embed a scripting language inside these long running processes.
So as the process run's they could tweak the actual algorithm & it's parameters.
Off the bat I looked at embedding:
Lua (useful guide)
Python (useful guide)
These both seem doable and might actually fully solve the given problem.
Any other bright idea's out there?
Recompiling after a 1-2 line change, redeploying to the test environment and restarting just sucks.
The system is fairly complicated and hopefully I explained it half decently.
If you can change enough of the program through a script to be useful, without a full recompile, maybe you should think about breaking the system up into smaller parts. You could have a "server" that handles data loading etc and then the client code that does the actual processing. Each time the system loads new data, it could check and see if the client code has been re-compiled and then use it if that's the case.
I think there would be a couple of advantages here, the largest of which would be that the whole system would be much less complex. Now you're working in one language instead of two. There is less of a chance that people can mess things up when moving from python or lua mode to c++ mode in their heads. By embedding some other language in the system you also run the risk of becoming dependent on it. If you use python or lua to tweek the program, those languages either become a dependency when it becomes time to deploy, or you need to back things out to C++. If you choose to port things to C++ theres another chance for bugs to crop up during the switch.
Embedding Lua is much easier than embedding Python.
Lua was designed from the start to be embedded; Python's embeddability was grafted on after the fact.
Lua is about 20x smaller and simpler than Python.
You don't say much about your build process, but building and testing can be simplified significantly by using a really powerful version of make. I use Andrew Hume's mk, but you would probably be even better off investing the time to master Glenn Fowler's nmake, which can add dependencies on the fly and eliminate the need for a separate configuration step. I don't ordinarily recommend nmake because it is somewhat complicated, but it is very clear that Fowler and his group have built into nmake solutions for lots of scaling and portability problems. For your particular situation, it may be worth the effort to master it.
Not sure I understand your system, but if the build and deployment is too complicated, maybe you could automate it? If deployment is completely automatic, would that solve the problem?
I don't understand how a scripting language would solve the problem? If you change your algorithm, you still need to restart calculation from the beginning, don't you?
It kind of sounds like what you need is CruiseControl or something similar; every time hyou touch the baseline code, it rebuilds and reruns tests.