Search Results

Search found 4944 results on 198 pages for 'multi threaded'.

Page 85/198 | < Previous Page | 81 82 83 84 85 86 87 88 89 90 91 92  | Next Page >

  • Logic / Render phases with a single thread

    - by DevilWithin
    The question I have may generate different opinions from different developers, but I'd still like to have an answer on this. Its all about the updating and rendering steps of the game loop, and their use under multi and single threaded environments. Currently, there is one thread running, which takes care of sequentially executing events , logic and rendering. Sometimes, the logic part may wish to change the game state to something else, and in between do some loading of files. The result is that the game hangs completely while loading, and then proceeds to normal rendering of the new state. To go around this, i could make another thread, do the loading there while the main thread renders a smooth loading animation, and then proceed normally. The real question is about if i don't create another thread. I could refresh the screen from the logic thread, and provide some basic loading screen, which could be not so smoothly updated while the files load. In fact, this approach is not loved by a lot of developers, as it scrambles render code in the logic step, which may cause problems of different sorts.. Hope its clear!

    Read the article

  • CPU Affinity on ARM processors

    - by dsljanus
    I am using some RaspberryPI boards for a data acquisition system. They are nice boards, with plenty of community support around them, but they are really slow. I am thinking of gradually replacing them with ODROID multicore boards, with the Samsung Exynos processors. I have some experience using taskset to set CPU affinity on my servers because I am always running Node.js applications that are by definition single threaded. Now, is it possible to do this on an ARM board? I do not see why it would not in theory, but I have doubts over how well it is going to work. Does anyone have experience with this kind of hack? Also, I would appreciate any comments about ARM CPUs and how they differ from x86.

    Read the article

  • 12-24 rack, 10-32 server thumbscrew. How to mount?

    - by JJ.
    We have just purchased a APC rack (model AR204A) with 12-24 threaded holes. We couldn't get a "square hole" model in time for our setup deadline. Unfortunately our rack servers (Lenovo RD240) appear to have 10-32 thumbscrews for securing the server to the rack. We've successfully mounted the server rails to the rack using 12-24 screws however the 10-32 thumbscrews in the server front won't "grab" the 12-24 holes in the rack, thus there is nothing to stop the server from sliding right off the rack if pushed from the back. The thumb screws on the server don't seem to be removable, so we can't simply use 12-24 screws instead. Any suggestions on how to work around this problem? Is there any way to "convert" a 12-24 hole to a 10-32 thread (or similar approach)? Thanks in advance.

    Read the article

  • E-Business Suite : Role of CHUNK_SIZE in Oracle Payroll

    - by Giri Mandalika
    Different batch processes in Oracle Payroll flow have the ability to spawn multiple child processes (or threads) to complete the work in hand. The number of child processes to fork is controlled by the THREADS parameter in APPS.PAY_ACTION_PARAMETERS view. THREADS parameter The default value for THREADS parameter is 1, which is fine for a single-processor system but not optimal for the modern multi-core multi-processor systems. Setting the THREADS parameter to a value equal to or less than the total number of [virtual] processors available on the system may improve the performance of payroll processing. However on the down side, since multiple child processes operate against the same set of payroll tables in HR schema, database may experience undesired consequences such as buffer busy waits and index contention, which results in giving up some of the gains achieved by using multiple child processes/threads to process the work. Couple of other action parameters, CHUNK_SIZE and CHUNK_SHUFFLE, help alleviate the database contention. eg., Set a value for THREADS parameter as shown below. CONNECT APPS/APPS_PASSWORD UPDATE PAY_ACTION_PARAMETERS SET PARAMETER_VALUE = DESIRED_VALUE WHERE PARAMETER_NAME = 'THREADS'; COMMIT; (I am not aware of any maximum value for THREADS parameter) CHUNK_SIZE parameter The size of each commit unit for the batch process is controlled by the CHUNK_SIZE action parameter. In other words, chunking is the act of splitting the assignment actions into commit groups of desired size represented by the CHUNK_SIZE parameter. The default value is 20, and each thread processes one chunk at a time -- which means each child process inserts or processes 20 assignment actions at any time. When multiple threads are configured, each thread picks up a chunk to process, completes the assignment actions and then picks up another chunk. This is repeated until all the chunks are exhausted. It is possible to use different chunk sizes in different batch processes. During the initial phase of processing, CHUNK_SIZE number of assignment actions are inserted into relevant table(s). When multiple child processes are inserting data at the same time into the same set of tables, as explained earlier, database may experience contention. The default value of 20 is mostly optimal in such a case. Experiment with different values for the initial phase by +/-10 for CHUNK_SIZE parameter and observe the performance impact. A larger value may make sense during the main processing phase. Again experimentation is the key in finding the suitable value for your environment. Start with a large value such as 2000 for the chunk size, then increment or decrement the size by 500 at a time until an optimal value is found. eg., Set a value for CHUNK_SIZE parameter as shown below. CONNECT APPS/APPS_PASSWORD UPDATE PAY_ACTION_PARAMETERS SET PARAMETER_VALUE = DESIRED_VALUE WHERE PARAMETER_NAME = 'CHUNK_SIZE'; COMMIT; CHUNK_SIZE action parameter accepts a value that is as low as 1 or as high as 16000. CHUNK SHUFFLE parameter By default, chunks of assignment actions are processed sequentially by all threads - which may not be a good thing especially given that all child processes/threads performing similar actions against the same set of tables almost at the same time. By saying not a good thing, I mean to say that the default behavior leads to contention in the database (in data blocks, for example). It is possible to relieve some of that database contention by randomizing the processing order of chunks of assignment actions. This behavior is controlled by the CHUNK SHUFFLE action parameter. Chunk processing is not randomized unless explicitly configured. eg., Set chunk shuffling as shown below. CONNECT APPS/APPS_PASSWORD UPDATE PAY_ACTION_PARAMETERS SET PARAMETER_VALUE = 'Y' WHERE PARAMETER_NAME = 'CHUNK SHUFFLE'; COMMIT; Finally I recommend checking the following document out for additional details and additional pay action tunable parameters that may speed up the processing of Oracle Payroll.     My Oracle Support Doc ID: 226987.1 Oracle 11i & R12 Human Resources (HRMS) & Benefits (BEN) Tuning & System Health Checks Also experiment with different combinations of parameters and values until the right set of action parameters and values are found for your deployment.

    Read the article

  • BusEnum2 and a Minor Bug Fix

    - by Kate Moss' Open Space
    The default root bus driver, BusEnum, enumerate and active drivers one by one in synchronized manner. It is not only slowing the boot time but in the even if any of driver's init function (XXX_init) get hanged, the whole system won't boot at all. There is a sample of enhanced root bus driver, BusEnum2, on the http://msdn.microsoft.com/en-us/library/dd187254.aspx The page provides the sample code and the detail explanation of the design concept. With multi-threaded BusEnum2 on CE7 with SMP enabled system, the scalability is even more significant. Since you have more than one processor and it can load drivers in parallel! Everything looks good so far, except to there is a small bug in the sample code. Fortunately, it is easy to fix. But hard to trace if you ever enc outer it! The BUSENUM2 flag only defined in BUSENUM2\BUSDEF\sources but not in BUSENUM2\BUSENUM\sources. The DeviceFolder is implemented in BUSENUM2\BUSDEF but the instance is created in BUSENUM2\BUSENUM\busenum.cpp, so the result is it allocates less memory than actual need.   Add   CDEFINES=$(CDEFINES) -DBUSENUM2   into BUSENUM2\BUSENUM\sources and the problem fixed!

    Read the article

  • How to keep background requests in sequence

    - by Jason Lewis
    I'm faced with implementing interfaces for some rather archaic systems, for handling online deposits to stored value accounts (think campus card accounts for students). Here's my dilemma: stage 1 of the process involves passing the user off to a thrid-party site for the credit card transaction, like old-school PayPal. Step two involves using a proprietary protocol for communicating with a legacy system for conducting the actual deposit. Step two requires that each transaction have a unique sequence number, and that the requests' seqnums are in order. Since we're logging each transaction in Postgres, my first thought was to take a number from a sequence in the DB, guaranteeing uniqueness. But since we're dealing with web requests that might come in near-simultaneously, and since latency with the return from the off-ste payment processor is beyond our control, there's always the chance for a race condition in the order of requests passed back to the proprietary system, and if the seqnums are out of order, the request fails silently (brilliant, right?). I thought about enqueuing the requests in Redis and using Resque workers to process them (single worker, single process, so they are processed in order), but we need to be able to give the user feedback as to whether the transaction was processed successfully, so this seems less feasible to me. I've tried to make this application handle concurrency well (as much as possible for a Ruby on Rails app), but now we're in a situation where we have to interact with a system that is designed to be single process, single threaded, and sequential. If it at least gave an "out of order" error, I could just increment (or take the next value off the sequence), but it's designed to fail silently in the event of ANY error. We are handling timeouts in a way that blocks on I/O, but since the application uses multiple workers (Unicorn), that's no guarantee. Any ideas/suggestions would be appreciated.

    Read the article

  • What determines which Javascript functions are blocking vs non-blocking?

    - by Sean
    I have been doing web-based Javascript (vanilla JS, jQuery, Backbone, etc.) for a few years now, and recently I've been doing some work with Node.js. It took me a while to get the hang of "non-blocking" programming, but I've now gotten used to using callbacks for IO operations and whatnot. I understand that Javascript is single-threaded by nature. I understand the concept of the Node "event queue". What I DON'T understand is what determines whether an individual javascript operation is "blocking" vs. "non-blocking". How do I know which operations I can depend on to produce an output synchronously for me to use in later code, and which ones I'll need to pass callbacks to so I can process the output after the initial operation has completed? Is there a list of Javascript functions somewhere that are asynchronous/non-blocking, and a list of ones that are synchronous/blocking? What is preventing my Javascript app from being one giant race condition? I know that operations that take a long time, like IO operations in Node and AJAX operations on the web, require them to be asynchronous and therefore use callbacks - but who is determining what qualifies as "a long time"? Is there some sort of trigger within these operations that removes them from the normal "event queue"? If not, what makes them different from simple operations like assigning values to variables or looping through arrays, which it seems we can depend on to finish in a synchronous manner? Perhaps I'm not even thinking of this correctly - hoping someone can set me straight. Thanks!

    Read the article

  • information about /proc/pid/sched

    - by redeye
    Not sure this is the right place for this question, but here goes: I'm trying to make some sense of the /proc/pid/sched and /proc/pid/task/tid/sched files for a highly threaded server process, however I was not able to find a good explanation of how to interpret this file ( just a few bits here: http://knol.google.com/k/linux-performance-tuning-and-measurement# ) . I assume this entry in procfs is related to newer versions of the kernel that run with the CFS scheduler? CentOS distro running on a 2.6.24.7-149.el5rt kernel version with preempt rt patch. Any thoughts?

    Read the article

  • Concurrency pattern of logger in multithreaded application

    - by Dipan Mehta
    The context: We are working on a multi-threaded (Linux-C) application that follows a pipeline model. Each module has a private thread and encapsulated objects which do processing of data; and each stage has a standard form of exchanging data with next unit. The application is free from memory leak and is threadsafe using locks at the point where they exchange data. Total number of threads is about 15- and each thread can have from 1 to 4 objects. Making about 25 - 30 odd objects which all have some critical logging to do. Most discussion I have seen about different levels as in Log4J and it's other translations. The real big questions is about how the overall logging should really happen? One approach is all local logging does fprintf to stderr. The stderr is redirected to some file. This approach is very bad when logs become too big. If all object instantiate their individual loggers - (about 30-40 of them) there will be too many files. And unlike above, one won't have the idea of true order of events. Timestamping is one possibility - but it is still a mess to collate. If there is a single global logger (singleton) pattern - it indirectly blocks so many threads while one is busy putting up logs. This is unacceptable when processing of the threads are heavy. So what should be the ideal way to structure the logging objects? What are some of the best practices in actual large scale applications? I would also love to learn from some of the real designs of large scale applications to get inspirations from!

    Read the article

  • Should I manage authentication on my own if the alternative is very low in usability and I am already managing roles?

    - by rumtscho
    As a small in-house dev department, we only have experience with developing applications for our intranet. We use the existing Active Directory for user account management. It contains the accounts of all company employees and many (but not all) of the business partners we have a cooperation with. Now, the top management wants a technology exchange application, and I am the lead dev on the new project. Basically, it is a database containing our know-how, with a web frontend. Our employees, our cooperating business partners, and people who wish to become our cooperating business partners should have access to it and see what technologies we have, so they can trade for them with the department which owns them. The technologies are not patented, but very valuable to competitors, so the department bosses are paranoid about somebody unauthorized gaining access to their technology description. This constraint necessitates a nightmarishly complicated multi-dimensional RBAC-hybrid model. As the Active Directory doesn't even contain all the information needed to infer the roles I use, I will have to manage roles plus per-technology per-user granted access exceptions within my system. The current plan is to use Active Directory for authentication. This will result in a multi-hour registration process for our business partners where the database owner has to manually create logins in our Active Directory and send them credentials. If I manage the logins in my own system, we could improve the usability a lot, for example by letting people have an active (but unprivileged) account as soon as they register. It seems to me that, after I am having a users table in the DB anyway (and managing ugly details like storing historical user IDs so that recycled user IDs within the Active Directory don't unexpectedly get rights to view someone's technologies), the additional complexity from implementing authentication functionality will be minimal. Therefore, I am starting to lean towards doing my own user login management and forgetting the AD altogether. On the other hand, I see some reasons to stay with Active Directory. First, the conventional wisdom I have heard from experienced programmers is to not do your own user management if you can avoid it. Second, we have code I can reuse for connection to the active directory, while I would have to code the authentication if done in-system (and my boss has clearly stated that getting the project delivered on time has much higher priority than delivering a system with high usability). Third, I am not a very experienced developer (this is my first lead position) and have never done user management before, so I am afraid that I am overlooking some important reasons to use the AD, or that I am underestimating the amount of work left to do my own authentication. I would like to know if there are more reasons to go with the AD authentication mechanism. Specifically, if I want to do my own authentication, what would I have to implement besides a secure connection for the login screen (which I would need anyway even if I am only transporting the pw to the AD), lookup of a password hash and a mechanism for password recovery (which will probably include manual identity verification, so no need for complex mTAN-like solutions)? And, if you have experience with such security-critical systems, which one would you use and why?

    Read the article

  • How exactly to implement multiple threads in a game

    - by xerwin
    So I recently started learning Java, and having a interest in playing games as well as developing them, naturally I want to create game in Java. I have experience with games in C# and C++ but all of them were single-threaded simple games. But now, I learned how easy it is to make threads in Java, I want to take things to the next level. I started thinking about how would I actually implement threading in a game. I read couple of articles that say the same thing "Usually you have thread for rendering, for updating game logic, for AI, ..." but I haven't (or didn't look hard enough) found example of implementation. My idea how to make implementation is something like this (example for AI) public class AIThread implements Runnable{ private List<AI> ai; private Player player; /*...*/ public void run() { for (int i = 0; i < ai.size(); i++){ ai.get(i).update(player); } Thread.sleep(/* sleep until the next game "tick" */); } } I think this could work. If I also had a rendering and updating thread list of AI in both those threads, since I need to draw the AI and I need to calculate the logic between player and AI(But that could be moved to AIThread, but as an example) . Coming from C++ I'm used to do thing elegantly and efficiently, and this seems like neither of those. So what would be the correct way to handle this? Should I just keep multiple copies of resources in each thread or should I have the resources on one spot, declared with synchronized keyword? I'm afraid that could cause deadlocks, but I'm not yet qualified enough to know when a code will produce deadlock.

    Read the article

  • Why do certain replied emails missing threading when replied back in Mutt?

    - by yarun can
    I use Mutt for emails. I have threads enabled and I can see that most of the emails are threaded in Mutt. So that is all good. But sometimes I reply to an email and the answer(from other person) to my replied email wont be part of any threads. The thing is that when I reply in Mutt (which I use Vim to edit them), the subject parts keeps getting longer and longer with many "Re"s. That is the case with those emails with missing threads. I have: set strict_threads="yes" set sort="threads" set edit_headers=yes I am wondering if this has anything to do with Mutt or the person I am communicating over email. Could this one be the culprit? set metoo=yes Any suggestions?

    Read the article

  • Increase the compression performance of VPN

    - by Martin
    I am currently switching from a system with HPN-SSH tunnels and enabled compression to something VPN based. I have tried tinc and n2n so far, hamachi requires a library I do not have. In my primitive benchmarks I am not satisfied with the achievable bandwidth compared to the SSH tunnels. In tinc the low LZO setting performed best, but compression is only available in UDP mode. Ideally I would like to have a TCP-based VPN with a multi-threaded compression. Can you suggest me some ideas how to increase the performance? Would it be possible to somehow put a compression filter in front of the tun interface? Or are there any VPN implementations that might be better suited for my needs (fast compression, TCP-based, switch mode, does not have to be super-secure)? I would consider tunnelling Ethernet over SSH, but according to some articles it is not advisable.

    Read the article

  • No MAU required on a T4

    - by jsavit
    Cryptic background One of the powerful features of the T-series servers is its hardware crypto acceleration, which dramatically speeds up the compute intensive algorithms needed to encrypt and decrypt data. Previously, administrators setting up logical domains on older T-series servers had to explicitly assign crypto resources (called "MAU" for historical reasons from the T1 chip that had "modular arithmetic units") to domains that had a significant crypto workload (say, an SSL based web server). This could be an administrative burden, as you had to choose which domains got the crypto units, and issue the appropriate ldm set-mau N mydomain commands. The T4 changes things The T4 is fast. Really fast. Its clock rate and out-of-order (OOO) execution that provides the single-thread performance that T-series machines previously did not have. If you have any preconceptions about T-series performance, or SPARC in general, based on the older servers (which, it must be said, were absolutely outstanding for multi-threaded applications), those assumptions are now obsolete. The T4 provides outstanding. performance for all kinds of workload, as illustrated at https://blogs.oracle.com/bestperf. While we all focused on this (did I mention the T4 is fast?), another feature of the T4 went largely unnoticed: The T4 servers have crypto acceleration "just built in" so administrators no longer have to assign crypto accelerator units to domains - it "just happens". This is way way better since you have crypto everywhere by default without having to manage it like a discrete and limited resource. It's a feature of the processor, like doing an integer add. With T4, there is no management necessary, you just have HW crypto everywhere all the time seamlessly. This change hasn't been widely advertised, and some administrators have wondered why there were unable to assign a MAU to a domain as they did with T2 and T3 machines. The answer is that there is no longer any separate MAU, so you don't have to take any action at all - just leave the default of 0. Summary Besides being much faster than its predecessors, the T4 also integrates hardware crypto acceleration so its seamlessly available to applications, whether domains are being used or not. Administrators no longer have to control how they are allocated - it "just happens"

    Read the article

  • Thinking Local, Regional and Global

    - by Apeksha Singh-Oracle
    The FIFA World Cup tournament is the biggest single-sport competition: it’s watched by about 1 billion people around the world. Every four years each national team’s manager is challenged to pull together a group players who ply their trade across the globe. For example, of the 23 members of Brazil’s national team, only four actually play for Brazilian teams, and the rest play in England, France, Germany, Spain, Italy and Ukraine. Each country’s national league, each team and each coach has a unique style. Getting all these “localized” players to work together successfully as one unit is no easy feat. In addition to $35 million in prize money, much is at stake – not least national pride and global bragging rights until the next World Cup in four years time. Achieving economic integration in the ASEAN region by 2015 is a bit like trying to create the next World Cup champion by 2018. The team comprises Brunei Darussalam, Cambodia, Indonesia, Lao PDR, Malaysia, Myanmar, Philippines, Singapore, Thailand and Vietnam. All have different languages, currencies, cultures and customs, rules and regulations. But if they can pull together as one unit, the opportunity is not only great for business and the economy, but it’s also a source of regional pride. BCG expects by 2020 the number of firms headquartered in Asia with revenue exceeding $1 billion will double to more than 5,000. Their trade in the region and with the world is forecast to increase to 37% of an estimated $37 trillion of global commerce by 2020 from 30% in 2010. Banks offering transactional banking services to the emerging market place need to prepare to repond to customer needs across the spectrum – MSMEs, SMEs, corporates and multi national corporations. Customers want innovative, differentiated, value added products and services that provide: • Pan regional operational independence while enabling single source of truth at a regional level • Regional connectivity and Cash & Liquidity  optimization • Enabling Consistent experience for their customers  by offering standardized products & services across all ASEAN countries • Multi-channel & self service capabilities / access to real-time information on liquidity and cash flows • Convergence of cash management with supply chain and trade finance While enabling the above to meet customer demands, the need for a comprehensive and robust credit management solution for effective regional banking operations is a must to manage risk. According to BCG, Asia-Pacific wholesale transaction-banking revenues are expected to triple to $139 billion by 2022 from $46 billion in 2012. To take advantage of the trend, banks will have to manage and maximize their own growth opportunities, compete on a broader scale, manage the complexity within the region and increase efficiency. They’ll also have to choose the right operating model and regional IT platform to offer: • Account Services • Cash & Liquidity Management • Trade Services & Supply Chain Financing • Payments • Securities services • Credit and Lending • Treasury services The core platform should be able to balance global needs and local nuances. Certain functions need to be performed at a regional level, while others need to be performed on a country level. Financial reporting and regulatory compliance are a case in point. The ASEAN Economic Community is in the final lap of its preparations for the ultimate challenge: becoming a formidable team in the global league. Meanwhile, transaction banks are designing their own hat trick: implementing a world-class IT platform, positioning themselves to repond to customer needs and establishing a foundation for revenue generation for years to come. Anand Ramachandran Senior Director, Global Banking Solutions Practice Oracle Financial Services Global Business Unit

    Read the article

  • Best Practice: Migrating Email Boxes (maildir format)

    - by GruffTech
    So here's the situation. I've got about 20,000 maildir email accounts chewing up a several hundred GB of space on our email server. Maildir by nature keeps thousands of tiny a** little files, instead of one .mbox file or the like... So i need to migrate all of these several millions of files from one server to the other, for both space and life-cycle reasons. the conventional methods i would use all work just fine. rsync is the option that comes immediately to mind, however i wanted to see if there are any other "better" options out there. Rsync not handling multi-threaded transfers in this situation sucks because it never actually gets up to speed and saturates my network connection, because of this the transfer from one server to another will take hours beyond hours, when it shouldn't really take more then one or two. I know this is highly opinionated and subjective and will therefore be marked community wiki.

    Read the article

  • MVC pattern synchronisation

    - by Hariprasad
    I am facing a problem in synchronizing my model and view threads I have a view which is table. In it, user can select a few rows. I update the view as soon as the user clicks on any row since I don't want the UI to be slow. This updating is done by a logic which runs in the controller thread below. At the same time, the controller will update the model data too, which takes place in a different thread. i.e., controller puts the query in a queue, which is then executed by the model thread - which is a single-threaded interface. As soon as the query executes, controller will get a signal. Now, In order to keep the view and model synchronized, I will update the view again based on the return value of the query (the data returned by model) - even though I updated the view already for that user action. But, I am facing issues because, its taking a lot of time for the model to return the result, by that time user would have performed multiple clicks. So, as a result of updating the view again based on the information from model, the view sometimes goes back to the state in which the previous clicks were made (Suppose user clicks thrice on different rows. I update the view as soon as the click happens. Also, I update the view when I get data back from the model - which is supposed to be same as the already updated state of the view. Now, when the user clicks third time, I get data for the first click from model. As a result, view goes back to a state which is generated by the first click) Is there any way to handle such a synchronization issue?

    Read the article

  • Is using a dedicated thread just for sending gpu commands a good idea?

    - by tigrou
    The most basic game loop is like this : while(1) { update(); draw(); swapbuffers(); } This is very simple but have a problem : some drawing commands can be blocking and cpu will wait while he could do other things (like processing next update() call). Another possible solution i have in mind would be to use two threads : one for updating and preparing commands to be sent to gpu, and one for sending these commands to the gpu : //first thread while(1) { update(); render(); // use gamestate to generate all needed triangles and commands for gpu // put them in a buffer, no command is send to gpu // two buffers will be used, see below pulse(); //signal the other thread data is ready } //second thread while(1) { wait(); // wait for second thread for data to come send_data_togpu(); // send prepared commands from buffer to graphic card swapbuffers(); } also : two buffers would be used, so one buffer could be filled with gpu commands while the other would be processed by gpu. Do you thing such a solution would be effective ? What would be advantages and disadvantages of such a solution (especially against a simpler solution (eg : single threaded with triple buffering enabled) ?

    Read the article

  • Impact of Server Failure on Coherence Request Processing

    - by jpurdy
    Requests against a given cache server may be temporarily blocked for several seconds following the failure of other cluster members. This may cause issues for applications that can not tolerate multi-second response times even during failover processing (ignoring for the moment that in practice there are a variety of issues that make such absolute guarantees challenging even when there are no server failures). In general, Coherence is designed around the principle that failures in one member should not affect the rest of the cluster if at all possible. However, it's obvious that if that failed member was managing a piece of state that another member depends on, the second member will need to wait until a new member assumes responsibility for managing that state. This transfer of responsibility is (as of Coherence 3.7) performed by the primary service thread for each cache service. The finest possible granularity for transferring responsibility is a single partition. So the question becomes how to minimize the time spent processing each partition. Here are some optimizations that may reduce this period: Reduce the size of each partition (by increasing the partition count) Increase the number of JVMs across the cluster (increasing the total number of primary service threads) Increase the number of CPUs across the cluster (making sure that each JVM has a CPU core when needed) Re-evaluate the set of configured indexes (as these will need to be rebuilt when a partition moves) Make sure that the backing map is as fast as possible (in most cases this means running on-heap) Make sure that the cluster is running on hardware with fast CPU cores (since the partition processing is single-threaded) As always, proper testing is required to make sure that configuration changes have the desired effect (and also to quantify that effect).

    Read the article

  • Using ConcurrentQueue for thread-safe Performance Bookkeeping.

    - by Strenium
    Just a small tidbit that's sprung up today. I had to book-keep and emit diagnostics for the average thread performance in a highly-threaded code over a period of last X number of calls and no more. Need of the day: a thread-safe, self-managing stats container. Since .NET 4.0 introduced new thread-safe 'Collections.Concurrent' objects and I've been using them frequently - the one in particular seemed like a good fit for storing each threads' performance data - ConcurrentQueue. But I wanted to store only the most recent X# of calls and since the ConcurrentQueue currently does not support size constraint I had to come up with my own generic version which attempts to restrict usage to numeric types only: unfortunately there is no IArithmetic-like interface which constrains to only numeric types – so the constraints here here aren't as elegant as they could be. (Note the use of the Average() method, of course you can use others as well as make your own).   FIFO FixedSizedConcurrentQueue using System;using System.Collections.Concurrent;using System.Linq; namespace xxxxx.Data.Infrastructure{    [Serializable]    public class FixedSizedConcurrentQueue<T> where T : struct, IConvertible, IComparable<T>    {        private FixedSizedConcurrentQueue() { }         public FixedSizedConcurrentQueue(ConcurrentQueue<T> queue)        {            _queue = queue;        }         ConcurrentQueue<T> _queue = new ConcurrentQueue<T>();         public int Size { get { return _queue.Count; } }        public double Average { get { return _queue.Average(arg => Convert.ToInt32(arg)); } }         public int Limit { get; set; }        public void Enqueue(T obj)        {            _queue.Enqueue(obj);            lock (this)            {                T @out;                while (_queue.Count > Limit) _queue.TryDequeue(out @out);            }        }    } }   The usage case is straight-forward, in this case I’m using a FIFO queue of maximum size of 200 to store doubles to which I simply Enqueue() the calculated rates: Usage var RateQueue = new FixedSizedConcurrentQueue<double>(new ConcurrentQueue<double>()) { Limit = 200 }; /* greater size == longer history */   That’s about it. Happy coding!

    Read the article

  • How to verify the code that could take a substantial time to compile? [on hold]

    - by user18404
    As a follow up to my prev question: What is the best aproach for coding in a slow compilation environment To recap: I am stuck with a large software system with which a TDD ideology of "test often" does not work. And to make it even worse the features like pre-compiled headers/multi-threaded compilation/incremental linking, etc is not available to me - hence I think that the best way out would be to add the extensive logging into the system and to start "coding in large chunks", which I understand as code for a two-three hours first (as opposed to 15-20 mins in TDD) - thoroughly eyeball the code for a 15 minutes and only after all that do the compilation and run the tests. As I have been doing TDD for a quite a while, my code eyeballing / code verification skills got rusty (you don't really need this that much if you can quickly verify what you've done in 5 seconds by running a test or two) - so I am after a recommendations on how to learn these source code verification/error spotting skills again. I know I was able to do that easily some 5-10 years ago when I din't have much support from the compiler/unit testing tools I had until recently, thus there should be a way to get back to the basics.

    Read the article

  • Creating a steady rhythm for music-based game in XNA

    - by A-Type
    I'm looking to develop a game for Windows Phone to explore an idea I had which involves the user building notes into a sequencer while playing a puzzle game. The issue I'm running into is that, while my implementation is very close to being on-beat, there is the occasional pause between beats which makes the whole thing sound sloppy. I'm just not sure how to get around this inside XNA's infrastructure. Currently I'm running this code in the Update method of my GameBoard: public void Update(GameTime gameTime) { onBeat = IsOnBeat(gameTime); [...] if (onBeat) BeatUpdate(); } private bool IsOnBeat(GameTime gameTime) { beatTime += gameTime.ElapsedGameTime.TotalSeconds; if (Math.Abs(beatTime - beatLength) < 0.0166666) { beatTime -= beatLength; return true; } return false; } private void BeatUpdate() { cursor.BeatUpdate(); board.CursorPass((int)cursor.CursorPosition % Board.GRID_WIDTH); } Update checks to see if the time is on beat, and if it is, it calls the BeatUpdate method which moves the cursor over the board (sequencer). The cursor reports its X position to the board, which then plays any notes which are in that position on the sequencer. Notes are SoundEffectInstances, preloaded and ready to play. Oh, and TargetElapsedTime is set to 166666, or 60FPS target. Obviously totaling up the time and then subtracting isn't the most accurate way to go but I can't figure out a way to work within XNA's system in order to overcome this issue. This current system is just horribly unstable. Beats lag and fire too early and it's obvious. I thought about perhaps some sort of threaded solution but I'm not familiar enough with multithreading to figure out how that would work. Any ideas?

    Read the article

  • What are the pros & cons of these MySQL engines for OLTP -- XtraDB, PBXT, or TokuDB?

    - by Continuation
    I'm working on a social website with an approximate read/write split of 90/10. Trying to decide on a MySQL engine. The ones I'm interested in are: XtraDB PBXT TokuDB What are the pros and cons of them for my use case? A few specific questions: PBXT uses log-based structure that avoids double-writes. It sounds very elegant, but the benchmark I've seen doesn't show any/much advantages over XtraDB. Do you have any experience with PBXT/XtraDB you can share? TokuDB sounds VERY interesting. But all the benchmarks I've seen are about single-threaded bulk inserts - inserting 100M rows for example. that's not very relevant for OLTP. What about its performance with large number of concurrent threads writing and reading at the same time? Anyone has tried that?

    Read the article

  • Separate update and render

    - by NSAddict
    I'm programming a simple Snake in Java. I'm a complete newbie when it comes to Java and Game Developing, so please bear with me ;) Until now, I have been using a UI thread, as well as a update-thread. The update thread just set the position, set the GameObjects, and so on. I didn't think much of concurrency, but now I've come to a problem. I wanted to modify the ArrayList<GameObject>, but it throws a java.util.ConcurrentModificationException. With a little research I found out that this happens because the two threads are trying to access the variables at the same time. But I didn't really find a way to prevent this. I thought about copying the array and swapping them out when the rendering is finished, but I would have to deep-copy them, which isn't really the best solution in my opinion. It probably eats up more CPU resources than a single-threaded game. Are there any other ways to prevent this? Thanks a lot for your help!

    Read the article

  • [SOLVED]Can't enable mysql and mysqli extension in PHP [closed]

    - by Sydcul
    I used to had my website hosted at a hosting company. But i decided to start my own webserver and in order to get phpBB, MediaWiki, etc. working i need PHP and MySQL. So after a bit of screwing around i could get those working but the MySQL and MySQLi extensions do not seem to work. When i use phpMyAdmin, phpBB, whatever it would say it is not installed correctly. I uncommented it in php.ini, i put my PHP folder in path, still not working. Please note i am not a PHP developer at all. Using: Windows Server 2003 Small Buisiness (too lazy to install Linux) Apache2 (not sure what version) PHP 5.2 (threaded, manually installed) MySQL 5.5.28 Thanks in advance, -Sydcul EDIT: Solved. Don't know how, just used the installer and it worked.

    Read the article

< Previous Page | 81 82 83 84 85 86 87 88 89 90 91 92  | Next Page >