Search Results

Search found 8429 results on 338 pages for 'batch processing'.

Page 168/338 | < Previous Page | 164 165 166 167 168 169 170 171 172 173 174 175  | Next Page >

  • Grid based collision - How many cells?

    - by Fibericon
    The game I'm creating is a bullet hell game, so there can be quite a few objects on the screen at any given time. It probably maxes out at about 40 enemies and 200 or so bullets. That being said, I'm splitting up the playing field into a grid for my collision checking. Right now, it's only 8 cells. How many would be optimal? I'm worried that if I use too many, I'll be wasting CPU power. My main concern is processing power, to make the game run smoothly. RAM is not a big concern for me.

    Read the article

  • Gigaom Article on Oracle, Freescale, and the push for Java on Internet of Things (IoT)

    - by hinkmond
    Here's an interesting article that came out during JavaOne which talks about the Oracle and Freescale partnership, where we are putting Java technology onto the Freescale i.MX6 based "one box" gateway. See: Oracle and Prosyst team up Here's a quote: When it comes to connected devices, there’s still plenty of debate over the right operating system, the correct protocols for sending data and even the basics of where processing will take place — on premise or in the cloud. This might seem esoteric, but if you’re waiting for your phone to unlock your front door, that round trip to the cloud or a fat OS isn’t going to win accolades if you’re waiting in the rain. With all of this in mind, Oracle and Freescale have teamed up to offer an appliance and a Java-based software stack for the internet of things. The first version of the "one box" will work in the connected smart home, but soon after that, Oracle and Freescale will develop later boxes for other industries ranging from healthcare, smart grid to manufacturing. Hinkmond

    Read the article

  • NeoAxis 3D disponible en version 1.3, le moteur de jeu ajoute le SSAO et une meilleure gestion du terrain

    NeoAxis 3D et maintenant disponible en version 1.3 La nouvelle version du moteur de jeu apporte le SSAO, des améliorations sur le terrain et plein d'autres surprises C'est en cette période de Noël que NeoAxis 3D, le moteur de jeu multi-plateforme revient avec une hotte fournit en cadeau pour nous, les développeurs. Tout d'abord, cette version estampillée 1.3 apporte le Screen Space Ambient Occlusion (SSAO). Une technique de rendu temps réel qui est grandement utilisée dans l'industrie pour permettre une approximation efficace de l'ambient occlusion (des ombres sont rajoutés selon le buffer de profondeur).. Celui-ci est implémenté tel un effet après rendu (post processing) afin de garder une bonne synergi...

    Read the article

  • Webcast Replay: Extreme Performance for Consolidated Workloads with Oracle Exadata

    - by kimberly.billings
    If you missed our live webcast Extreme Performance for Consolidated Workloads with Oracle Exadata last week, the replay is now available. Watch the free on-demand webcast in which Tim Shetler, Vice President of Oracle Database Product Management, and Richard Exley, Consulting Member of Technical Staff, discuss how Oracle Exadata can help you can significantly improve application performance and reduce infrastructure costs by consolidating transaction processing, data warehousing, or mixed workloads on Oracle Exadata. Note: (1) Turn off pop-up blockers if the slides do not advance automatically. (2) Slides are available for download. var gaJsHost = (("https:" == document.location.protocol) ? "https://ssl." : "http://www."); document.write(unescape("%3Cscript src='" + gaJsHost + "google-analytics.com/ga.js' type='text/javascript'%3E%3C/script%3E")); try { var pageTracker = _gat._getTracker("UA-13185312-1"); pageTracker._trackPageview(); } catch(err) {}

    Read the article

  • Map, Set use cases in a general web app

    - by user2541902
    I am currently working on my own Java web app (to be shown in interview to get a Java job). So I've not worked on Java in professional environment, so no guidance. I have database, entity classes, JPA relationships. Use cases are like, user has albums, album has pics, user has locations, location has co-ordinates etc. I used List (ArrayList) everywhere. I can do anything with List and DB, get some entry, find etc. For example, I will keep the list of users in List, then use queries to get some entry (why would I keep them in Map with id/email as key?). I know very well the working and features, implementing classes of Map, Set. I can use them for solving some algorithm, processing some data etc. In interviews, I get asked have you worked with these, where have you used them etc. So, Please tell me cases where they should be used (DB or any popular real use case).

    Read the article

  • How to get audio spectrum analysis?

    - by Mrwolfy
    I need to find or create a tool that analyzes the audio spectrum of a sound file (like a .wav or .mp3). I need to output the "volume" or power of x number of frequency bands and output the data as text. This will be used to produce a visualization, a graphic equalizer like you'd see on a stereo. I am currently looking at python to do it. My question is are there some tools out there that would do this (signal processing), like math works or others? I don't have any experience with them so any advice would be appreciated.

    Read the article

  • Looking for an old classic book about Unix command-line tools

    - by Little Bobby Tables
    I am looking for a book about the Unix command-line toolkit (sh, grep, sed, awk, cut, etc.) that I read some time ago. It was an excellent book, but I totally forgot its name. The great thing about this specific book was the running example. It showed how to implement a university bookkeeping system using only text-processing tools. You would find a student by name with grep, update grades with sed, calculate average grades with awk, attach grades to IDs with cut, and so on. If my memory serve, this book had a black cover, and was published circa 1980. Does anyone remember this book? I would appreciate any help in finding it.

    Read the article

  • Is this safe? <a href=http://javascript:...>

    - by KajMagnus
    I wonder if href and src attributes on <a> and <img> tags are always safe w.r.t. XSS attacks, if they start with http:// or https://. For example, is it possible to append javascript: ... to the href and src attribute in some manner, to execute code? Disregarding whether or not the destination page is e.g. a pishing site, or the <img src=...> triggers a terribly troublesome HTTP GET request. Background: I'm processing text with markdown, and then I sanitize the resulting HTML (using Google Caja's JsHtmlSanitizer). Some sample code in Google Caja assumes all hrefs and srcs that start with http:// or https:// are safe -- I wonder if it's safe to use that sample code. Kind regards, Kaj-Magnus

    Read the article

  • What are some good, simple examples for queues?

    - by Michael Ekstrand
    I'm teaching CS2 (Java and data structures), and am having some difficulty coming up with good examples to use when teaching queues. The two major applications I use them for are multithreaded message passing (but MT programming is out of scope for the course), and BFS-style algorithms (and I won't be covering graphs until later in the term). I also want to avoid contrived examples. Most things that I think of, if I were actually going to solve them in a single-threaded fashion I would just use a list rather than a queue. I tend to only use queues when processing and discovery are interleaved (e.g. search), or in other special cases like length-limited buffers (e.g. maintaining last N items). To the extent practical, I am trying to teach my students good ways to actually do things in real programs, not just toys to show off a feature. Any suggestions of good, simple algorithms or applications of queues that I can use as examples but that require a minimum of other prior knowledge?

    Read the article

  • New Exadata and Exalogic public references

    - by Javier Puerta
    The following are new public references for Exadata and Exalogic: Allegis Accelerates HR Processing for 130,000 Contractors  Oracle customer, Allegis, describes how Oracle Exadata and Oracle Exalogic helped consolidate and optimize critical processes running in Oracle's PeopleSoft.  Hyundai Motor Company Document Cuts Repository Management and Access Times Approximately 85%, Saves More Than US$1 Million in Yearly Printing and Paper Costs The company implemented Oracle Exalogic Elastic Cloud, Oracle Exadata Database Machine, Oracle WebLogic, and Oracle WebCenter Content 11g to ensure high performance and stability for its new document-centralization system  University of Minnesota Reduces Data Center Footprint while Enhancing Performance and Manageability with Oracle Exadata Database Machine   Leading Research Institution Consolidates More Than 200 Databases to Approximately 20 while Maximizing Availability for Thousands of Users SThree Prepares to Triple in Size with a Cloud-Based Architecture and a Consolidated, Stable, and Scalable Global Platform  By consolidating 68 databases into a single Oracle Exadata Database Machine, SThree achieved the stability and scalability it needed to support its growth targets. Further enhancements to the organization’s core systems include a planned upgrade for Siebel Contact Center and improved integration with Oracle Fusion Middleware.

    Read the article

  • WP: Oracle Multitenant on SuperCluster T5-8: Study of Database Consolidation Efficiency

    - by uwes
    Consolidation in the data center is the driving factor in reducing capital and operational expense in IT today. This is particularly relevant as customers invest more in cloud infrastructure and associated service delivery. Database consolidation is a strategic component in this effort. Oracle Database 12 c introduces Oracle Multitenant , a new database consolidation model in which multiple Pluggable Databases (PDBs) are consolidated within a Container Database (CDB). While keeping many of the isolation aspects of single databases, it allows PDBs to share the system global area (SGA) and background processes of a common CDB . The white paper recently published on OTN: Oracle Multitenant on SuperCluster T5-8: Study of Database Consolidation Efficiency analyzes and quantifies savings in compute resources, efficiencies in transaction processing, and consolidation density of Oracle Multitenant compared to consolidated single instance databases (SIDBs) running in a bare-metal environment.

    Read the article

  • Using normals in DirectX 10

    - by Dave
    I've got a working OBJ loader that loads vertices, indices, texture coordinates, and normals. As of right now it doesn't process texture coordinates or normals but it stores them in arrays and creates a valid mesh with the vertices and indices. Now I am trying to figure out how can I make the shader use the correct normal in the array for the current vertex if I can't setnormals() to my mesh. If I were to just use an index in my array of normals corresponding to the index in the vertices, how would I retrieve the current index the shader is processing? BTW: I am trying to write a blinn-phong shader technique. Also when I create the input layout and I've added the semantic NORMAL to it, how would I list the multiple semantics in that single parameter? Would I just separate it with a space? PS: If you need to see any code, just let me know.

    Read the article

  • When designing a job queue, what should determine the scope of a job?

    - by Stuart Pegg
    We've got a job queue system that'll cheerfully process any kind of job given to it. We intend to use it to process jobs that each contain 2 tasks: Job (Pass information from one server to another) Fetch task (get the data, slowly) Send task (send the data, comparatively quickly) The difficulty we're having is that we don't know whether to break the tasks into separate jobs, or process the job in one go. Are there any best practices or useful references on this subject? Is there some obvious benefit to a method that we're missing? So far we can see these benefits for each method: Split Job lease length reflects job length: Rather than total of two Finer granularity on recovery: If we lose outgoing connectivity we can tell them all to retry The starting state of the second task is saved to job history: Helps with debugging (although similar logging could be added in single task method) Single Single job to be scheduled: Less processing overhead Data not stale on recovery: If the outgoing downtime is quite long, the pending Send jobs could be outdated

    Read the article

  • DRY, string, and unit testing

    - by Rodrigue
    I have a recurring question when writing unit tests for code that involves constant string values. Let's take an example of a method/function that does some processing and returns a string containing a pre-defined constant. In python, that would be something like: STRING_TEMPLATE = "/some/constant/string/with/%s/that/needs/interpolation/" def process(some_param): # We do some meaningful work that gives us a value result = _some_meaningful_action() return STRING_TEMPLATE % result If I want to unit test process, one of my tests will check the return value. This is where I wonder what the best solution is. In my unit test, I can: apply DRY and use the already defined constant repeat myself and rewrite the entire string def test_foo_should_return_correct_url(): string_result = process() # Applying DRY and using the already defined constant assert STRING_TEMPLATE % "1234" == string_result # Repeating myself, repeating myself assert "/some/constant/string/with/1234/that/needs/interpolation/" == url The advantage I see in the former is that my test will break if I put the wrong string value in my constant. The inconvenient is that I may be rewriting the same string over and over again across different unit tests.

    Read the article

  • Problem Implementing Texture on Libgdx Mesh of Randomized Terrain

    - by BrotherJack
    I'm having problems understanding how to apply a texture to a non-rectangular object. The following code creates textures such as this: from the debug renderer I think I've got the physical shape of the "earth" correct. However, I don't know how to apply a texture to it. I have a 50x50 pixel image (in the environment constructor as "dirt.png"), that I want to apply to the hills. I have a vague idea that this seems to involve the mesh class and possibly a ShapeRenderer, but the little i'm finding online is just confusing me. Bellow is code from the class that makes and regulates the terrain and the code in a separate file that is supposed to render it (but crashes on the mesh.render() call). Any pointers would be appreciated. public class Environment extends Actor{ Pixmap sky; public Texture groundTexture; Texture skyTexture; double tankypos; //TODO delete, temp public Tank etank; //TODO delete, temp int destructionRes; // how wide is a static pixel private final float viewWidth; private final float viewHeight; private ChainShape terrain; public Texture dirtTexture; private World world; public Mesh terrainMesh; private static final String LOG = Environment.class.getSimpleName(); // Constructor public Environment(Tank tank, FileHandle sfileHandle, float w, float h, int destructionRes) { world = new World(new Vector2(0, -10), true); this.destructionRes = destructionRes; sky = new Pixmap(sfileHandle); viewWidth = w; viewHeight = h; skyTexture = new Texture(sky); terrain = new ChainShape(); genTerrain((int)w, (int)h, 6); Texture tankSprite = new Texture(Gdx.files.internal("TankSpriteBase.png")); Texture turretSprite = new Texture(Gdx.files.internal("TankSpriteTurret.png")); tank = new Tank(0, true, tankSprite, turretSprite); Rectangle tankrect = new Rectangle(300, (int)tankypos, 44, 45); tank.setRect(tankrect); BodyDef terrainDef = new BodyDef(); terrainDef.type = BodyType.StaticBody; terrainDef.position.set(0, 0); Body terrainBody = world.createBody(terrainDef); FixtureDef fixtureDef = new FixtureDef(); fixtureDef.shape = terrain; terrainBody.createFixture(fixtureDef); BodyDef tankDef = new BodyDef(); Rectangle rect = tank.getRect(); tankDef.type = BodyType.DynamicBody; tankDef.position.set(0,0); tankDef.position.x = rect.x; tankDef.position.y = rect.y; Body tankBody = world.createBody(tankDef); FixtureDef tankFixture = new FixtureDef(); PolygonShape shape = new PolygonShape(); shape.setAsBox(rect.width*WORLD_TO_BOX, rect.height*WORLD_TO_BOX); fixtureDef.shape = shape; dirtTexture = new Texture(Gdx.files.internal("dirt.png")); etank = tank; } private void genTerrain(int w, int h, int hillnessFactor){ int width = w; int height = h; Random rand = new Random(); //min and max bracket the freq's of the sin/cos series //The higher the max the hillier the environment int min = 1; //allocating horizon for screen width Vector2[] horizon = new Vector2[width+2]; horizon[0] = new Vector2(0,0); double[] skyline = new double[width]; //TODO skyline necessary as an array? //ratio of amplitude of screen height to landscape variation double r = (int) 2.0/5.0; //number of terms to be used in sine/cosine series int n = 4; int[] f = new int[n*2]; //calculating omegas for sine series for(int i = 0; i < n*2 ; i ++){ f[i] = rand.nextInt(hillnessFactor - min + 1) + min; } //amp is the amplitude of the series int amp = (int) (r*height); double lastPoint = 0.0; for(int i = 0 ; i < width; i ++){ skyline[i] = 0; for(int j = 0; j < n; j++){ skyline[i] += ( Math.sin( (f[j]*Math.PI*i/height) ) + Math.cos(f[j+n]*Math.PI*i/height) ); } skyline[i] *= amp/(n*2); skyline[i] += (height/2); skyline[i] = (int)skyline[i]; //TODO Possible un-necessary float to int to float conversions tankypos = skyline[i]; horizon[i+1] = new Vector2((float)i, (float)skyline[i]); if(i == width) lastPoint = skyline[i]; } horizon[width+1] = new Vector2(800, (float)lastPoint); terrain.createChain(horizon); terrain.createLoop(horizon); //I have no idea if the following does anything useful :( terrainMesh = new Mesh(true, (width+2)*2, (width+2)*2, new VertexAttribute(Usage.Position, (width+2)*2, "a_position")); float[] vertices = new float[(width+2)*2]; short[] indices = new short[(width+2)*2]; for(int i=0; i < (width+2); i+=2){ vertices[i] = horizon[i].x; vertices[i+1] = horizon[i].y; indices[i] = (short)i; indices[i+1] = (short)(i+1); } terrainMesh.setVertices(vertices); terrainMesh.setIndices(indices); } Here is the code that is (supposed to) render the terrain. @Override public void render(float delta) { Gdx.gl.glClearColor(1, 1, 1, 1); Gdx.gl.glClear(GL10.GL_COLOR_BUFFER_BIT); // tell the camera to update its matrices. camera.update(); // tell the SpriteBatch to render in the // coordinate system specified by the camera. backgroundStage.draw(); backgroundStage.act(delta); uistage.draw(); uistage.act(delta); batch.begin(); debugRenderer.render(this.ground.getWorld(), camera.combined); batch.end(); //Gdx.graphics.getGL10().glEnable(GL10.GL_TEXTURE_2D); ground.dirtTexture.bind(); ground.terrainMesh.render(GL10.GL_TRIANGLE_FAN); //I'm particularly lost on this ground.step(); }

    Read the article

  • How to apply a filter to the screen of a running program?

    - by Shahbaz
    The idea is to take old games without modifying them, but have the graphics card apply a series of filters to their output before sending them to the monitor. A very crude example would be to take a game that has a resolution of 640x480 and do: Increase the resolution to 1280x960 Apply a blur (low pass filter) Apply a sharpen (1 + high pass filter) These steps may not necessarily be the best to improve the visuals of an old game, but there are a lot of techniques that are well-known in image processing for this purpose. The question is, do the (NVidia) graphics cards give the ability to load a program that modifies the screen before sending it to the monitor? If so, how are they called and what terminology should I use to search? I would be comfortable with doing the programming myself if this ability is part of a library. Also, would the solution be different between Windows and Linux? If so, either is fine, since most of the games are probably runnable by wine.

    Read the article

  • New customer references for Exadata-based projects

    - by Javier Puerta
    Milletech Systems, Inc. shows a large state university how to improve query response times 15-fold using its grant management solution built on Oracle’s Extreme Performance Infrastructure. Read More. Ação Informática helped Valdecard realize a 15-Fold Improvement in Fleet-Management and Benefit Card Data Processing using Oracle Exadata Machine. Read More. Neusoft deployed Benxi Municipal Human Resources and Social Security Bureau’s cloud-based database platform to process social insurance payments 50-times faster using Oracle Exadata Database Machine. Read More.

    Read the article

  • BPM Parallel Multi Instance sub processes by Niall Commiskey

    - by JuergenKress
    Here is a very simple scenario: An order with lines is processed. The OrderProcess accepts in an order with its attendant lines. The Fulfillment process is called for each order line. We do not have many order lines, and the processing is simple, so we run this in parallel. Let's look at the definition of the Multi Instance sub-process - Read the full article here. SOA & BPM Partner Community For regular information on Oracle SOA Suite become a member in the SOA & BPM Partner Community for registration please visit www.oracle.com/goto/emea/soa (OPN account required) If you need support with your account please contact the Oracle Partner Business Center. Blog Twitter LinkedIn Facebook Wiki Mix Forum Technorati Tags: BPM,Niall Commiskey,SOA Community,Oracle SOA,Oracle BPM,Community,OPN,Jürgen Kress

    Read the article

  • NPOT texture and video memory usage

    - by Eonil
    I read in this QA that NPOT will take memory as much as next POT sized texture. It means it doesn't give any benefit than POT texture with proper management. (maybe even worse because NPOT should be slower!) Is this true? Does NPOT texture take and waste same memory like POT texture? I am considering NPOT texture for post-processing, so if it doesn't give memory space benefit, using of NPOT texture is meaningless to me. Maybe answer can be different for each platforms. I am targeting mobile devices. Such as iPhone or Androids. Does NPOT texture takes same amount of memory on mobile GPUs?

    Read the article

  • it started with a broken package

    - by Inteloidl.com
    all started when i tried to updated to the latest lts, but the problen is that the update interumpted and some broken packages apeared, i used synaptics to repair, but it wont repair, iv tried some console from some forum topic ... and in the end when i try to open synaptic or update managre this is what apears any ideas : E: Dynamic MMap ran out of room. Please increase the size of APT::Cache-Limit. Current value: 25165824. (man 5 apt.conf) E: Error occurred while processing mantis (NewVersion1) E: Problem with MergeList /var/lib/apt/lists/c.archive.ubuntu.com_ubuntu_dists_lucid_universe_binary-i386_Packages E: The package lists or status file could not be parsed or opened. E: _cache-open() failed, please report. W: Unable to munmap

    Read the article

  • Is content slowing down your business?

    - by Lance Shaw
    We are living in a digital world, however paper is everywhere and expensive, right? We all agree content is an important part of our organization and contribute to its decision making. However many of us see dealing with this as a challenge and the growth of content is impacting our ability to scale and respond quickly to our customers. Business always has been content intensive. For JD Edwards customers, this is an important consideration.  After all, the processes being run in JD Edwards are usually very critical to the success of your business and if they are not running as smoothly as they should due to manual process steps involving paper or searching for content, you should look into improving them.  To that end, we hope you will join this webinar and learn how Oracle and KPIT | SYSTIME have partnered to help a JD Edwards customer content-enable its enterprise with Oracle WebCenter Content and Oracle WebCenter Imaging 11g and integrate them back with JD Edwards to significantly improve processing speed and operational costs.

    Read the article

  • Coherence Query Performance in Large Clusters

    - by jpurdy
    Large clusters (measured in terms of the number of storage-enabled members participating in the largest cache services) may introduce challenges when issuing queries. There is no particular cluster size threshold for this, rather a gradually increasing tendency for issues to arise. The most obvious challenges are that a client's perceived query latency will be determined by the slowest responder (more likely to be a factor in larger clusters) as well as the fact that adding additional cache servers will not increase query throughput if the query processing is not compute-bound (which would generally be the case for most indexed queries). If the data set can take advantage of the partition affinity features of Coherence, then the application can use a PartitionedFilter to target a query to a single server (using partition affinity to ensure that all data is in a single partition). If this can not be done, then avoiding an excessive number of cache server JVMs will help, as will ensuring that each cache server has sufficient CPU resources available and is also properly configured to minimize GC pauses (the most common cause of a slow-responding cache server).

    Read the article

  • Broken package ubuntuone-control-panel-qt

    - by baale7
    An interrupted installation of updates resulted in ubuntuone-control-panel-qt breaking and now I cannot install or update anything. There seems to be an error in python2.7 - ImportError: No module named _sysconfigdata_nd dpkg: error processing archive ubuntuone-control-panel-qt_13.09-0ubuntu1_all.deb (--install): When I try to reinstall the package via .deb a dpkg --audit yields: The following packages are in a mess due to serious problems during installation. They must be reinstalled for them (and any packages that depend on them) to function properly: ?python-pexpect Python module for automating interactive applications ?ubuntuone-control-panel-qt Ubuntu One Control Panel - Qt frontend ?hplip-data HP Linux Printing and Imaging - data files ?python-libxml2 Python bindings for the GNOME XML library ?apport automatically generate crash reports for debugging The following packages are only half configured, probably due to problems configuring them the first time. The configuration should be retried using dpkg --configure or the configure menu option in dselect: ?python-ubuntuone-client Ubuntu One client Python libraries ?python-ubuntuone-control-panel Ubuntu One Control Panel - Python Libraries --configure or --configure -a doesn't work Any help?

    Read the article

  • Lightweight Ubuntu

    - by Nick Berardi
    Just to start off, I know of lubuntu but it really doesn't meet what I am looking for. Basically what I am looking for is the standard Desktop Ubuntu install, but with out all the word processing, multimedia, and games installed. I have seen posts out about how to get the desktop environment running on Ubuntu server, but they seem complicated, and never seem to equal the standard Desktop install. So my question is, is there anyway to tell the standard Desktop install not to install all the applications? Or is there a distro available that leaves all the applications out, and just has the standard desktop look and feel? What I really want this for is, is for development purposes to run on a VM to do Mono development.

    Read the article

  • Configuring MySQL Cluster Data Nodes

    - by Mat Keep
    0 0 1 692 3948 Homework 32 9 4631 14.0 Normal 0 false false false EN-US JA X-NONE /* Style Definitions */ table.MsoNormalTable {mso-style-name:"Table Normal"; mso-tstyle-rowband-size:0; mso-tstyle-colband-size:0; mso-style-noshow:yes; mso-style-priority:99; mso-style-parent:""; mso-padding-alt:0cm 5.4pt 0cm 5.4pt; mso-para-margin:0cm; mso-para-margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:12.0pt; font-family:Cambria; mso-ascii-font-family:Cambria; mso-ascii-theme-font:minor-latin; mso-hansi-font-family:Cambria; mso-hansi-theme-font:minor-latin; mso-ansi-language:EN-US;} In my previous blog post, I discussed the enhanced performance and scalability delivered by extensions to the multi-threaded data nodes in MySQL Cluster 7.2. In this post, I’ll share best practices on the configuration of data nodes to achieve optimum performance on the latest generations of multi-core, multi-thread CPU designs. Configuring the Data Nodes The configuration of data node threads can be managed in two ways via the config.ini file: - Simply set MaxNoOfExecutionThreads to the appropriate number of threads to be run in the data node, based on the number of threads presented by the processors used in the host or VM. - Use the new ThreadConfig variable that enables users to configure both the number of each thread type to use and also which CPUs to bind them too. The flexible configuration afforded by the multi-threaded data node enhancements means that it is possible to optimise data nodes to use anything from a single CPU/thread up to a 48 CPU/thread server. Co-locating the MySQL Server with a single data node can fully utilize servers with 64 – 80 CPU/threads. It is also possible to co-locate multiple data nodes per server, but this is now only required for very large servers with 4+ CPU sockets dense multi-core processors. 24 Threads and Beyond! An example of how to make best use of a 24 CPU/thread server box is to configure the following: - 8 ldm threads - 4 tc threads - 3 recv threads - 3 send threads - 1 rep thread for asynchronous replication. Each of those threads should be bound to a CPU. It is possible to bind the main thread (schema management domain) and the IO threads to the same CPU in most installations. In the configuration above, we have bound threads to 20 different CPUs. We should also protect these 20 CPUs from interrupts by using the IRQBALANCE_BANNED_CPUS configuration variable in /etc/sysconfig/irqbalance and setting it to 0x0FFFFF. The reason for doing this is that MySQL Cluster generates a lot of interrupt and OS kernel processing, and so it is recommended to separate activity across CPUs to ensure conflicts with the MySQL Cluster threads are eliminated. When booting a Linux kernel it is also possible to provide an option isolcpus=0-19 in grub.conf. The result is that the Linux scheduler won't use these CPUs for any task. Only by using CPU affinity syscalls can a process be made to run on those CPUs. By using this approach, together with binding MySQL Cluster threads to specific CPUs and banning CPUs IRQ processing on these tasks, a very stable performance environment is created for a MySQL Cluster data node. On a 32 CPU/Thread server: - Increase the number of ldm threads to 12 - Increase tc threads to 6 - Provide 2 more CPUs for the OS and interrupts. - The number of send and receive threads should, in most cases, still be sufficient. On a 40 CPU/Thread server, increase ldm threads to 16, tc threads to 8 and increment send and receive threads to 4. On a 48 CPU/Thread server it is possible to optimize further by using: - 12 tc threads - 2 more CPUs for the OS and interrupts - Avoid using IO threads and main thread on same CPU - Add 1 more receive thread. Summary As both this and the previous post seek to demonstrate, the multi-threaded data node extensions not only serve to increase performance of MySQL Cluster, they also enable users to achieve significantly improved levels of utilization from current and future generations of massively multi-core, multi-thread processor designs. A big thanks to Mikael Ronstrom, Senior MySQL Architect at Oracle, for his work in developing these enhancements and best practices. You can download MySQL Cluster 7.2 today and try out all of these enhancements. The Getting Started guides are an invaluable aid to quickly building a Proof of Concept Don’t forget to check out the MySQL Cluster 7.2 New Features whitepaper to discover everything that is new in the latest GA release

    Read the article

< Previous Page | 164 165 166 167 168 169 170 171 172 173 174 175  | Next Page >