Search Results

Search found 5564 results on 223 pages for 'multi gpu'.

Page 179/223 | < Previous Page | 175 176 177 178 179 180 181 182 183 184 185 186  | Next Page >

  • Macports install of ack doesn't create correct executable

    - by user1664196
    I am trying to install p5-app-ack port from Mac Ports, but it seems it doesn't create a /opt/local/bin/ack binary at the end: $ sudo port search *app-ack Password: p5-app-ack @1.960.0 (perl) A grep replacement that ignores .svn/CVS/blib directories p5.8-app-ack @1.960.0 (perl) A grep replacement that ignores .svn/CVS/blib directories p5.10-app-ack @1.960.0 (perl) A grep replacement that ignores .svn/CVS/blib directories p5.12-app-ack @1.960.0 (perl) A grep replacement that ignores .svn/CVS/blib directories p5.14-app-ack @1.960.0 (perl) A grep replacement that ignores .svn/CVS/blib directories p5.16-app-ack @1.960.0 (perl) A grep replacement that ignores .svn/CVS/blib directories Found 6 ports. $ perl --version This is perl 5, version 12, subversion 4 (v5.12.4) built for darwin-thread-multi-2level Copyright 1987-2010, Larry Wall Perl may be copied only under the terms of either the Artistic License or the GNU General Public License, which may be found in the Perl 5 source kit. Complete documentation for Perl, including FAQ lists, should be found on this system using "man perl" or "perldoc perl". If you have access to the Internet, point your browser at http://www.perl.org/, the Perl Home Page. $ sudo port install p5-app-ack ---> Computing dependencies for p5-app-ack ---> Cleaning p5-app-ack ---> Updating database of binaries: 100.0% ---> Scanning binaries for linking errors: 35.0% ---> No broken files found. $ $ ls /opt/local/bin/ac* /opt/local/bin/ack-5.12 /opt/local/bin/aclocal /opt/local/bin/aclocal-1.12 /opt/local/bin/activation-client /opt/local/bin/acyclic $ which ack $ ack -bash: ack: command not found Update If I then try to install p5.12-app-ack afterwards, I get $ sudo port install p5.12-app-ack Password: ---> Computing dependencies for p5.12-app-ack ---> Cleaning p5.12-app-ack ---> Scanning binaries for linking errors: 100.0% ---> No broken files found. $

    Read the article

  • when to upgrade server to include more cores, versus more processors, versus additional server?

    - by gkdsp
    The server hosting market is separated into single, double, qual, etc., processors, where each processor has several cores, or CPUs. My company will offer a Linux-based web application that relies on an Apache web server and a middle tier for business logic. The middle tier is used to crunch math, and return result to a client. Many clients may access the application simultaneously. The company will start with one processor having 4 cores. I'm trying to understand how the app uses the cores and then how to scale the application as business grows, in terms of servers/processors/cores. For example, I'd assume initially one core would be used for Apache, and the other 3 used to process client's requests for math crunching... Question 1: does that mean, with the 3 cores available, I can handle 3 separate client requests simultaneously (e.g. 1 for each of 3 cores)? I mean, except for the shared RAM, is this effectively like having 3 individual machines (from pt of view or processing client requests simulaneously)? Or, only one client's request may be processed at any one time, but that client's request is divided up into up to 3 cores depending on the type of process running that does the math crunching and whether or not it can take advantage of multi threading (so the # of cores impacts how fast any one client request completes)? I'm confused about what the cores mean to the application here. Question 2: As the business grows and more client requests need to be processed, should the server be upgraded to (A) a new machine with more cores, (B) a new machine with two processors, 4 cores each, or (C) keep the original server and add another server with a single processor? Which route provides the most efficient way to scale the application, in terms of processing more client requests per time interval? Is the choice, for example, limited by RAM (when you need more RAM than box can handle it's time to add another server), or something else? Question 3: Is the total number of client requests processed simultaneously equal to the number of cores times the number of servers (minus the one core for Apache)?

    Read the article

  • What mobile phones are viable for a "nerdy" person? [closed]

    - by Blixt
    This community wiki is for determining a list of good mobile phone choices if you are "nerdy" (definition follows.) As a point of reference, the almost two years old N95 8GB will be used. Mostly because that's what I can most relate to since I've had it since it came out. Today, iPhone and other modern mobile phones really out-shine it in usability and interface. However, it still does everything I want it to do (and here's the definition of "nerdy"; modify as needed): Syncs contacts, calendar, tasks and mail in the background Can run multiple installable applications in the background (Google Maps with Latitude, for example) Good amount of space for music etc. Lets you develop your own little toy applications (Python; not to mention it can also run an Apache server with a public URL!) Tethering! Supports Flash (maybe not very important, but it has its uses) Has a manufacturer that believes in the nerdiness! (The people at Nokia Labs make lots of cool stuff and share early versions with the community and are generally open) A high resolution screen (at least 320x240) Special hardware features such as an accelerometer and GPS What's missing from the N95 8GB but still qualifies as good, "nerdy", qualities: 7.2 Mbit/s (or faster) internet through HSDPA or HSPA+. A good web browser that can do most of the stuff a desktop browser can (especially render sites properly and run JavaScript correctly) Touch (especially multi-touch) More special hardware features such as a compass Intuitive and fluent user interface (Shiny stuff) Ability to configure it to trust root certificates of my own choice A processor fast enough to run Quake 3! =)

    Read the article

  • Extend Linux Desktop to another X Windows Display

    - by unknown (google)
    Hello, I am a long time Linux user of the Xinerama and other technologies for extending a desktop to multiple monitors. However when I travel with my laptop I miss the multi-monitor support I enjoy at home. Recently I acquired a second laptop for a low price. Both laptops are running Fedora (versions 10 and 11 respectively). I use Gnome as my primary desktop environment. I know about synergy. I use synergy all the time to control the screen of other Windows / Linux systems I use. I would like to know, can I sit both my primary and secondary laptops together and achieve a Xinerama-like extended desktop environment? Ideally I would like to start a GNOME session on my primary laptop. And then start a X-Windows Desktop on my secondary laptop and extend my primary laptop's desktop onto it. I would like to be able to move Windows from the primary desktop to the secondary laptop desktop. Would I need to use synergy to do this with some other bit of X-Windows technology? Or is there X-Windows technology that will do all this for me? I am familiar with X Windows ability to display applications remotely. I am also familiar with Nomachine's NoX.

    Read the article

  • Separate Certificate by Subdomain (With multiple IPs)

    - by Brian
    Note: Yes, I realize this problem is easier to solve by just using 1 multi-domain or wildcard certificate. I wish to have an ASP.NET site running on IIS with 2 SSL domains sharing 1 web application but using separate certificates. Assuming I have 2 certificates, this can be solved on IIS7 as follows: Web Application1: Binding 1: http, 80, IP Address *, Host Name * Binding 2: https, 443, IPADDRESS1, using CERTDOMAIN1 (DOMAIN1 resolves to IPADDRESS1) Binding 3: https, 443, IPADDRESS2, using CERTDOMAIN2 (DOMAIN2 resolves to IPADDRESS2) That is to say, 2 certificates and 2 ip addresses, but both mapped to the same web application. In IIS6, the closest I have been able to come to this configuration is: Web Application1: Binding 1: http, 80, IPADDRESS1 Binding 2: https, 443, IPADDRESS1, using CERTDOMAIN1 (DOMAIN1 resolves to IPADDRESS1) Web Application2: Binding 1: http, 80, IPADDRESS2 Binding 2: https, 443, IPADDRESS2, using CERTDOMAIN2 (DOMAIN2 resolves to IPADDRESS2) That is to say, 2 certificates and 2 IP addresses, 2 web applications, both mapped to the same file location. The IIS6 solution is not optimal. Even if sharing an application pool, there are still costs associated with running the same site as two applications. Is upgrading from IIS6 to IIS7 a legitimate way to resolve this problem? Is there an IIS6 way to map 2 IP addresses within the same web application to different certificates?

    Read the article

  • Recommendation for a simple no-frills Windows PDF printer driver?

    - by Scott Bussinger
    I'm looking for an extremely simple Windows PDF printer driver that I can recommend to clients. Ideally it would have these characteristics: When you print something, it should just create it as a temporary file and then display it in their default PDF viewer with no prompting. If they want to save it, they can save it manually from inside the viewer. This workflow should be with no special post-install configuration. Installation should be very simple. A double click the installation program and click "Finish" sort of thing. No complicated multi-step installation, no asking questions your grandmother wouldn't know the answer to (preferably no questions at all), no extra crap being installed. An option for a completely silent installation would be nice, but not necessary. Ideally it would be free to simplify their installing on a small network, but low cost is an option. I've tried a quite a few but none really fit the bill. Some can achieve the first goal but only after careful configuration, some try to install extra toolbars, some have other installation complexities that would make it hard for extremely novice users to succeed. Any suggestions? Thanks!

    Read the article

  • rsync per-site configuration file?

    - by Scott
    I know how to configure a per-site entry for ssh, but is there any kind of a client configuration for rsync that allows per-site configuration options and aliases or similar shortcuts like the .ssh/config? I'm curious because I have a minimal ssh server installed on my android phone and I also have a minimal rsync tool on it as well. I'm getting tired of having to root login onto the phone and sym-link both tools to standard places the android OS looks for executables as the ssh server is bare bones and has a typical *bear multi-link binary for the basic unix commands (that does not include rsync) I end up having to include --rsync-path=/path/to/rsync/android/files/rsync every time I want to do any rsyncing of the files on my phone, but this path is always the same. I've gotten around it in the meantime with a glob approach in a shell script wrapper, but this sometimes limits the customization I can do with the rsync call. I'm just wondering if there is anything similar to the .ssh/config file where I can create an alias for my phone (e.g. 'android') where specifying rsync android:/mnt/sdcard will automatically assume --rsync-path=/blah/blah/blah --no-g --no-p --no-t etc. Tre`

    Read the article

  • What is the typical maximum number of database connections for Oracle running on Windows server ?

    - by Sake
    We are maintaining a database server that serve a large number of clients. Each client typically running serveral client-applications. The total number of connections to the database server (Oracle 9i) is reaching 800 connections on peak load. The windows 2003 server is starting to run out of memory. We are now planning to move to 64bit Windows in order to gain higher memory capability. As a developer I suggest moving to multi-tier architecture with conneciton pooling, which I believe is a natural solution to this problem. However, in order to support my idea, I want the information on: what exactly is the typical number of connections allowed for Oracle database ? What is the problem when the number connections is too high ? Too much memory comsumption ? or too many sockets opened ? or too many context switching between threads ? To be a little bit specific, how could Oracle Forms application scale to thousand of users without facing this problem ? Shall Oracle RAC applied to this case ? I'm sure the answer to this question should depend on quite a number of factors, like the exact spec of the hardware being used. I'm expecting a rough estimation or some experience from the real world.

    Read the article

  • How to resume XMPP groupchat window in Irssi (using bitlbee)?

    - by mcnesium
    I use Bitlbee to chat in XMPP-networks within my IRC-client Irssi. This works great so far, and recently I started using XMPP Multi User Chats as an alternative to IRC-channels. I set up a channel using chat add <account> <[email protected]> in the &bitlbee control window, set chan <room> set autojoin true and entered /join #room in the &bitlbee window to join that groupchat. It then appears as a unique Irssi window in the status bar. This seems to work ok too, but with one exception: Since I idle in the channels 24/7 my irssi has to cope with the every-night-24h-DSL-disconnection by the ISP. After it automatically reconnects, it does kind of rejoin that XMPP-groupchat, but the traffic of the groupchat does not go back to the unique irssi window, but keeps flooding &bitlbee with messages from root telling me about a Groupchat Message from unknown JID <jid>: <message> - which is the traffic of the groupchat. The unique groupchat window is gone after the reconnect, and I will again have to go /join #room in &bitlbee to get it back. Even worse, the window number is unused before I rejoin the groupchat, and if I get a query from any network, the window nests in that unused window spot, so I will first have to remove that query from the spot, and then move the rejoined groupchat to that window number. I want my groupchat window to resume after the reconnect just like every other IRC channel too. How can I get this done? Any ideas?

    Read the article

  • What is the risk of introducing non standard image machines to a corporate environment

    - by Troy Hunt
    I’m after some feedback from those in the managed desktop or network security space on the risks of introducing machines that are not built on a standard desktop image into a large corporate environment. This particular context relates to the standard corporate image (32 bit Win XP) in a large multi-national not being suitable for a particular segment of users. In short, I’m looking at what hurdles we might come across by proposing the introduction of machines which are built and maintained by a handful of software developers and not based on the corporate desktop image (proposing 64 bit Win 7). I suspect the barriers are primarily around virus definition updates, the rollout of service packs and patches and the compatibility of existing applications with the newer OS. In terms of viruses and software updates, if machines were using common virus protection software with automated updates and using Windows Update for service packs and patches, is there still a viable risk to the corporate environment? For that matter, are large corporate environments normally vulnerable to the introduction of a machine not based on a standard image? I’m trying to get my head around how real the risk of infection and other adverse events are from machines being plugged into the network. There are multiple scenarios outside of just the example above where this might happen (i.e. a vendor plugging in a machine for internet access during a presentation). Would a large corporate network normally be sufficiently hardened against such innocuous activity? I appreciate the theory as to why policies such as standard desktop images exist, I’m just interested in the actual, practical risk and how much a network should be protected by means other than what is managed on individual PCs.

    Read the article

  • Vim: auto-comment in new line

    - by padde
    Vim automatically inserts a comment when I start a new line from a commented out line, because I have set formatoptions=tcroql. For example (cursor is *): // this is a comment* and after hitting <Enter> (insert mode) or o (normal mode) i am left with: // this is a comment // * This feature is very handy when writing long multi-line comments, but often I just want a single line comment. Now if I want to end the comment series I have several options: hit <Esc>S hit <BS> three times Both of these afford three keystrokes, taken together with the <Enter> this means four keystrokes for a new line, which I think is too much. Ideally, I would like to just hit <Enter> a second time to be left with: // this is a comment * It is important that the solution will also work with different indentation levels, i.e. int main(void) { // this is a comment* } hit <Enter> int main(void) { // this is a comment // * } hit <Enter> int main(void) { // this is a comment * } I think I have seen this feature in some text editor a few years ago but I do not recall which one it was. Is anyone aware of a solution that will do this for me in Vim? Pointers in the right direction on how to roll my own solution are also very welcome.

    Read the article

  • Does Mac OS X throttle the RATE of socket creation?

    - by pbhogan
    This may seem programming related, but this is an OS question. I'm writing a small high performance daemon that takes thousands of connections per second. It's working fine on Linux (specifically Ubuntu 9.10 on EC2). On Mac OS X if I throw a few thousand connections at it (roughly about 16350) in a benchmark that simply opens a connection, does it's thing and closes the connection, then the benchmark program hangs for several seconds waiting for a socket to become available before continuing (or timing out in the process). I used both Apache Bench as well as Siege (to make sure it wasn't the benchmark application). So why/how is Mac OS X limiting the RATE at which sockets can be used, and can I stop it from doing this? Or is there something else going on? I know there is a file descriptor limit, but I'm not hitting that. There is no error on accepting a socket, it's simply hangs for a while after the first (roughly) 16000, waiting -- I assume -- for the OS to release a socket. This shouldn't happen since all prior the sockets are closed at that point. They're supposed to come available at the rate they're closed, and do on Ubuntu, but there seems to be some kind of multi (5-10?) second delay on Mac OS X. I tried tweaking with ulimit every-which-way. Nada.

    Read the article

  • Does Mac OS X throttle the RATE of socket creation?

    - by pbhogan
    This may seem programming related, but this is an OS question. I'm writing a small high performance daemon that takes thousands of connections per second. It's working fine on Linux (specifically Ubuntu 9.10 on EC2). On Mac OS X if I throw a few thousand connections at it (roughly about 16350) in a benchmark that simply opens a connection, does it's thing and closes the connection, then the benchmark program hangs for several seconds waiting for a socket to become available before continuing (or timing out in the process). I used both Apache Bench as well as Siege (to make sure it wasn't the benchmark application). So why/how is Mac OS X limiting the RATE at which sockets can be used, and can I stop it from doing this? Or is there something else going on? I know there is a file descriptor limit, but I'm not hitting that. There is no error on accepting a socket, it's simply hangs for a while after the first (roughly) 16000, waiting -- I assume -- for the OS to release a socket. This shouldn't happen since all prior the sockets are closed at that point. They're supposed to come available at the rate they're closed, and do on Ubuntu, but there seems to be some kind of multi (5-10?) second delay on Mac OS X. I tried tweaking with ulimit every-which-way. Nada.

    Read the article

  • How to decouple development server from Internet?

    - by intoxicated.roamer
    I am working in a small set-up where there are 4 developers (might grow to 6 or 8 in cuople of years). I want to set-up an environment in which developers get an internet access but can not share any data from the company on internet. I have thought of the following plan: Set-up a centralized git server (Debian). The server will have an internet access. A developer will only have git account on that server, and won't have any other account on it. Do not give internet access to developer's individual machine (Windows XP/Windows 7). Run a virtual machine (any multi-user OS) on the centralized server (the same one on which git is hosted). Developer will have an account on this virtual machine. He/she can access internet via this virtual machine. Any data-movement between this virtual machine and underlying server, as well as any of the developer's machine, is prohibited. All developers require USB port on their local machine, so that they can burn their code into a microcontroller. This port will be made available only to associated software that dumps the code in a microcontroller (MPLAB in current case). All other softwares will be prohibited from accessing the port. As more developers get added, providing internet support for them will become difficult with this plan as it will slow down the virtual machine running on the server. Can anyone suggest an alternative ? Are there any obvious flaws in the above plan ? Some key details of the server are as below: 1) OS:Debian 2) RAM: 8GB 3) CPU: Intel Xeon E3-1220v2 4C/4T

    Read the article

  • VMware ESX Linux Guest Customization

    - by andyh_ky
    Hello, I am interested in deploying several RHEL 4 Update 8 virtual machines for creation of a test environment. Here are the steps I am taking: In off hours, P2V/V2V the production machines and convert them to templates Deploy the virtual machines with a customization specification that changes hostname, IP address I am interested in how these processes are done and if there are any options for further customization. Are the machines brought on the network when they are powered on, before they are reconfigured? Is there a potential IP address conflict? Is there an option to run additional scripts which reside on the guest as a part of the reconfiguration? For example, restoring an Oracle Database. This is an option with Windows guests and sysprep, but I have been unable to locate anything showing a RHEL equivalent. I am dealing with a multi tier application. The main issue I am attempting to mitigate is that the application servers reference database servers by hostname and in tnsnames files. I am interested in scripting the reconfiguration of the application in the deployment so that the app/db servers are pointing to the test environment. I am OK with placing the 'cleanup' script on the source and executing it after the machine has been brought up. I am interested in the automation of the script's execution post clone/boot, as well as if there could be an IP address conflict. (cross posted to VMTN's ESX 4 community)

    Read the article

  • Methods and practices for managing a network that has no internet connection

    - by FaultyJuggler
    Originally asked in Super User but realized this belongs here. Long story short, I am setting up a network with 32 servers of varying specs that will be used for testing and development. We will be using RedHat Linux, we also do not have a router as of yet and were looking into making one of the servers act as our router/DHCP etc. The small cluster will be on an isolated network with no internet. I can use external harddrives and discs to transfer anything from external sources into machines on the network, so this isn't a locked down secure network, it just won't have a direct connection to the outside world. I've worked on such setups before, but always long after they were setup. So I'm reaching out to see what everyone knows as far as how groups have handled initial setup and maintenance of such a situation. What is the best way to get them all configured and up to date? What are the best ways to automate updates, network wide installs, etc. With the only given that I have large multi-terabyte external hard drives that would be used to drop whatever files are needed onto a central server, how do i then distribute those files and install their contents? I've done perl scripting, some teammates have played with puppet, so we aren't completely in the dark, I just wanted to avoid reinventing the wheel since this is a common challenge.

    Read the article

  • [openVPN] server & client on same machine . And multiple VPN servers

    - by HiWorld
    Hello everyone, im stucked configuring openvpn to build a multi vpn connection. like this: CLIENT - VPN1 - VPN2 - INTERNET Well, i already have and know how to done a normal sigle vpn but want to use a chain of vpns, so i explain what i have done and how i did it. ON VPN1. i have 1 openvpn instance running as server( where client connect to) and another as client connecting to VPN2 running as server. { Here comes the problem } when i connect VPN1 as client of VPN2 i cant connect to VPN1 from CLIENT, my question is HOW TO procced with this... Also have another third instance working as server to use VPN1 without chains. ON VPN2. 1 openvpn instance as server where VPN1 will connect and then forward to the NET. Im using TUN interface on configs. And iptables are on this way: VPN1 - openvpn ip server1 : 192.168.6.0 / ip as client of VPN2: 192.168.5.70 iptables -t nat -A POSTROUTING -s 192.168.6.0 -j SNAT --to-source 192.168.5.70 VPN2 - openvpn ip server2 : 192.168.5.0 iptables -t nat -A POSTROUTING -s 192.168.5.0/24 -j SNAT --to-source EXTERNAL_IP_TO_INTERNET Hope someone help me with this. thanks in advance

    Read the article

  • Integrating external computer into a domain - some recommendations please

    - by TomTom
    Given: * A multi loation company. Every office has local routers that connect to a central VPN capable rouer in a data center. All fine so far. We now need to move a computer off site into a hosting center across the globe, to get it closer to some supplier computers we work for. it will run limited logic but latency is important, and our latency so far is too large. This computer will be in a data center and does no require incoming connections except for adminsitrative purposes, although it needs outgoing connetions. I have no real chance to put one of my VPN routers there, sadly - otherwise I would have no problem. Usage of RRAs is not recommended (we had various probblems there over time). I could deal with it. The computer MUSt integrate into the corporate structure via VPN and join the domain and be fully "tracked" (controlled for performance). What is the best suggestion? So far it looks like my best bets woudl be to log in via RRAS and deal with whatever issues arise there plus uise the local firewall the limit incoming connections to this computer to what is needed (which runs down to an emergency RDP connection allowance). Anyone a better idea?

    Read the article

  • What is the potential for a FUSE mount to destabilize a Linux server?

    - by 200_success
    I'm a sysadmin for a multi-user server, where students in our department have shell accounts. One of our users has requested that we install sshfs on it. I'm debating whether it would be wise to install sshfs as suggested. My main concern is whether a FUSE mount could make our server less reliable. In my experience, bad things can happen to servers when an NFS server suddenly becomes unavailable — the load average shoots up, and you might not be able to unmount it cleanly, to the point where a hard reboot might be necessary. If a FUSE-mounted server suddenly disappears, how hard might it be to clean up the mess? Are there any other likely catastrophes or gotchas I should consider? At least with NFS, only root can mount, and we can choose to mount NFS servers that we consider to be reasonably reliable. Let's assume that our users have no hostile intentions, but might do stupid things accidentally. Also, I'm not really worried about the contents of the filesystems they might mount, since our users already have shell access and can copy anything they want to their home directory.

    Read the article

  • Asynchronous readback from opengl front buffer using multiple PBO's

    - by KillianDS
    I am developing an application that needs to read back the whole frame from the front buffer of an openGL application. I can hijack the application's opengl library and insert my code on swapbuffers. At the moment I am successfully using a simple but excruciating slow glReadPixels command without PBO's. Now I read about using multiple PBO's to speed things up. While I think I've found enough resources to actually program that (isn't that hard), I have some operational questions left. I would do something like this: create a series (e.g. 3) of PBO's use glReadPixels in my swapBuffers override to read data from front buffer to a PBO (should be fast and non-blocking, right?) Create a seperate thread to call glMapBufferARB, once per PBO after a glReadPixels, because this will block until the pixels are in client memory. Process the data from step 3. Now my main concern is of course in steps 2 and 3. I read about glReadPixels used on PBO's being non-blocking, will this be an issue if I issue new opengl commands after that very fast? Will those opengl commands block? Or will they continue (my guess), and if so, I guess only swapbuffers can be a problem, will this one stall or will glReadPixels from front buffer be many times faster than swapping (about each 15-30ms) or, worst case scenario, will swapbuffers be executed while glReadPixels is still reading data to the PBO? My current guess is this logic will do something like this: copy FRONT_BUFFER - generic place in VRAM, copy VRAM-RAM. But I have no idea which of those 2 is the real bottleneck and more, what the influence on the normal opengl command stream is. Then in step 3. Is it wise to do this asynchronously in a thread separated from normal opengl logic? At the moment I think not, It seems you have to restore buffer operations to normal after doing this and I can't install synchronization objects in the original code to temporarily block those. So I think my best option is to define a certain swapbuffer delay before reading them out, so e.g. calling glReadPixels on PBO i%3 and glMapBufferARB on PBO (i+2)%3 in the same thread, resulting in a delay of 2 frames. Also, when I call glMapBufferARB to use data in client memory, will this be the bottleneck or will glReadPixels (asynchronously) be the bottleneck? And finally, if you have some better ideas to speed up frame readback from GPU in opengl, please tell me, because this is a painful bottleneck in my current system. I hope my question is clear enough, I know the answer will probably also be somewhere on the internet but I mostly came up with results that used PBO's to keep buffers in video memory and do processing there. I really need to read back the front buffer to RAM and I do not find any clear explanations about performance in that case (which I need, I cannot rely on "it's faster", I need to explain why it's faster). Thank you

    Read the article

  • C++ Multithreading with pthread is blocking (including sockets)

    - by Sebastian Büttner
    I am trying to implement a multi threaded application with pthread. I did implement a thread class which looks like the following and I call it later twice (or even more), but it seems to block instead of execute the threads parallel. Here is what I got until now: The Thread Class is an abstract class which has the abstract method "exec" which should contain the thread code in a derive class (I did a sample of this, named DerivedThread) Thread.hpp #ifndef THREAD_H_ #define THREAD_H_ #include <pthread.h> class Thread { public: Thread(); void start(); void join(); virtual int exec() = 0; int exit_code(); private: static void* thread_router(void* arg); void exec_thread(); pthread_t pth_; int code_; }; #endif /* THREAD_H_ */ And Thread.cpp #include <iostream> #include "Thread.hpp" /*****************************/ using namespace std; Thread::Thread(): code_(0) { cout << "[Thread] Init" << endl; } void Thread::start() { cout << "[Thread] Created Thread" << endl; pthread_create( &pth_, NULL, Thread::thread_router, reinterpret_cast<void*>(this)); } void Thread::join() { cout << "[Thread] Join Thread" << endl; pthread_join(pth_, NULL); } int Thread::exit_code() { return code_; } void Thread::exec_thread() { cout << "[Thread] Execute" << endl; code_ = exec(); } void* Thread::thread_router(void* arg) { cout << "[Thread] exec_thread function in thread" << endl; reinterpret_cast<Thread*>(arg)->exec_thread(); return NULL; } DerivedThread.hpp #include "Thread.hpp" class DerivedThread : public Thread { public: DerivedThread(); virtual ~DerivedThread(); int exec(); void Close() = 0; DerivedThread.cpp [...] #include "DerivedThread.cpp" [...] int DerivedThread::exec() { //code to be executed do { cout << "Thread executed" << endl; usleep(1000000); } while (true); //dummy, just to let it run for a while } [...] Basically, I am calling this like the here: DerivedThread *thread; cout << "Creating Thread" << endl; thread = new DerivedThread(); cout << "Created thread, starting..." << endl; thread->start(); cout << "Started thread" << endl; cout << "Creating 2nd Thread" << endl; thread = new DerivedThread(); cout << "Created 2nd thread, starting..." << endl; thread->start(); cout << "Started 2nd thread" << endl; What is working great if I am only starting one of these Threads , but if I start multiple which should run together (not synced, only parallel) . But I discovered, that the thread is created, then as it tries to execute it (via start) the problem seems to block until the thread has closed. After that the next Thread is processed. I thought that pthread would do it unblocked for me, so what did I wrong? A sample output might be: Creating Thread [Thread] Thread Init Created thread, starting... [Thread] Created thread [Thread] exec_thread function in thread [Thread] Execute Thread executed Thread executed Thread executed Thread executed Thread executed Thread executed Thread executed .... Until Thread 1 is not terminated, a Thread 2 won't be created not executed. The process above is executed in an other class. Just for the information: I am trying to create a multi threaded server. The concept is like this: MultiThreadedServer Class has a main loop, like this one: ::inet::ServerSock *sock; //just a simple self made wrapper class for sockets DerivedThread *thread; for (;;) { sock = new ::inet::ServerSock(); this->Socket->accept( *sock ); cout << "Creating Thread" << endl; //Threads (according to code sample above) thread = new DerivedThread(sock); //I did not mentoine the parameter before as it was not neccesary, in fact, I pass the socket handle with the connected socket to the thread cout << "Created thread, starting..." << endl; thread->start(); cout << "Started thread" << endl; } So I thought that this would loop over and over and wait for new connections to accept. and when a new client arrives, I am creating a new thread and give the thread the connected socket as a parameter. In the DerivedThread::exec I am doing the handling for the connected client. Like: [...] do { [...] if (this-sock_-read( Buffer, sizeof(PacketStruc) ) 0) { cout << "[Handler_Base] Recv Packet" << endl; //handle the packet } else { Connected = false; } delete Buffer; } while ( Connected ); So I loop in the created thread as long as the client keeps the connection. I think, that the socket may cause the blocking behaviour. Edit: I figured out, that it is not the read() loop in the DerivedThread Class as I simply replaced it with a loop over a simple cout-usleep part. It did also only execute the first one and after first thread finished, the 2nd one was executed. Many thanks and best regards, Sebastian

    Read the article

  • iPhone Image Resources, ICO vs PNG, app bundle filesize

    - by Jasarien
    My application has a collection of around 1940 icons that are used throughout. They're currently in ICO and new images provided to me come in ICO format too. I have noticed that they contain a 16x16 and 32x32 representation of each icon in one file. Each file is roughly 4KB in filesize (as reported by finder, but ls reports that they vary from being ~1000 bytes to 5000 bytes) A very small number of these icons only contain the 32x32 representation, and as a result are only around 700 bytes in size. Currently I am bundling these icons with my application and they are inflating the size of the app a bit more than I would like. Altogether, the images total just about 25.5MB. Xcode must do some kind of compression because the resulting app bundle is about 12.4MB. Compressing this further into a ZIP (as it would be when submitted to the App Store), results in a final file of 5.8MB. I'm aware that the maximum limit for over the air App Store downloads has been raised to 20MB since the introduction of the iPad (I'm not sure if that extends to iPhone apps as well as iPad apps though, if not the limit would be 10MB). My worry is that new icons are going to be added (sometimes up to 10 icons per week), and will continue to inflate the app bundle over time. What is the best way to distribute these icons with my app? Things I've tried and not had much success with: Converting the icons from ICO to PNG: I tried this in the hopes that the pngcrush utility would help out with the filesize. But it appears that it doesn't make much of a difference between a normal PNG and a crushed png (I believe it just optimises the image for display on the iPhone's GPU rather than compress it's size). Also in going from ICO to PNG actually increased the size of the icon file... Zipping the images, and then uncompressing them on first run. While this did reduce the overall image sizes, I found that the effort needed to unzip them, copy them to the documents folder and ensure that duplication doesn't happen on upgrades was too much hassle to be worth the benefit. Also, on original and 3G iPhones unzipping and copying around 25MB of images takes too long and creates a bad experience... Things I've considered but not yet tried: Instead of distributing the icons within the app bundle, host them online, and download each icon on demand (it depends on the user's data as to which icons will actually be displayed and when). Issues with this is that bandwidth costs money, and image downloads will be bandwidth intensive. However, my app currently has a small userbase of around 5,500 users (of which I estimate around 1500 to be active based on Flurry stats), and I have a huge unused bandwidth allowance with my current hosting package. So I'm open to thoughts on how to solve this tricky issue.

    Read the article

  • How to distort the desktop screen

    - by HaifengWang
    Hi friends, I want to change the shape of the desktop screen, so what are displayed on the desktop will be distorted at the same time. And the user can still operate the PC with the mouse on the distorted desktop(Run the applications, Open the "My Computer" and so on). I think I must get the projection matrix of the screen coordinate at first. Then transform the matrix, and map the desktop buffer image to the distorted mesh. Are there any interfaces which can modify the shape of the desktop screen in OpenGL or DirectX? Would you please give me some tip on it. Thank you very much in advance. Please refer to the picture from http://oi53.tinypic.com/bhewdx.jpg BR, Haifeng Addition1: I'm sorry! Maybe I didn't express clearly what I want to implement. What I want to implement is to modify the shape of the screen. So we can distort the shapes of all the applications which are run on Windows at the same time. For example that the window of "My Computer" will be distorted with the distortion of the desktop screen. And we can still operate the PC with mouse from the distorted desktop(Click the shortcut to run a program). Addition2: The projection matrix is just my assume. There isn't any desktop projection matrix by which the desktop surface is projected to the screen. What I want to implement is to change the shape of the desktop, as the same with mapping the desktop to an 3D mesh. But the user can still operate the OS on the distorted desktop(Click the shortcut to run a program, open the ie to surf the internet). Addition3: The shapes of all the programs run on the OS are changed with the distortion of the screen. It's realtime. The user can still operate the OS on the distorted screen as usually. Maybe we can intercept or override the GPU itself to implement the effect. I'm investigating GDI, I think I can find some clue for that. The first step is to find how to show the desktop on the screen.

    Read the article

  • NVIDIA graphics driver in Ubuntu 12.04

    - by user924501
    So my overall goal is that I want to be able to code with CUDA enabled applications. However, upon many days of searching, using installation walkthroughs, and reinstalling countless times after driver failure... I'm now here as a last resort. I cannot get Ubuntu 12.04 LTS to install the NVIDIA 295.59 driver for my GeForce GT 540M NVIDIA graphics card. My main system specs is as follows... (I believe having the Intel processor may be the problem) DELL Laptop XPS L502X Intel® Core™ i7-2620M CPU @ 2.70GHz × 4 Intel Integrated Graphics 64 bit NVIDIA GeForce GT 540M Ubuntu 12.04 LTS All other specs are irrelevant unless I forgot something? Methods Tried: aptitude install nvidia-current (all packages) Results: Nothing really happened. Nothing in the additional drivers menu appeared, nor was the NVIDIA X Server settings application allowing access because it thought there was no NVIDIA X Server installed. Downloaded driver from nvidia.com. Set nomodeset in the grub boot menu through /boot/grub/grub.cfg Went to console and turned off lightdm. Installed the driver, but it said the pre-install failed? (mean anything?) Started up lightdm again. Results: NVIDIA X Server settings still didn't notice it. Even tried to do nvidia-xconfig multiple times. I also went into the config file to make sure the driver setting said "nvidia". aptitude install nvidia-173 (all packages) Results: Couldn't find the xorg-video-abi-10 virtual package. It was nowhere to be found and the ubuntu forums everywhere had no answers. Lots of people were having this problem. This is easily done in windows, simply download the driver and debug in visual studio with no problems at all. I'd like clear step-by-step instructions on how I should go about this. I'm relatively new to linux but I can find my way around pretty well so you aren't talking to a straight-up beginner. Also, if you think another thread may have the answer please post because I was having a hard time looking for my specific type of problem. TL;DR I want to have access to my GPU so I can code with CUDA while in Ubuntu 12.04 on my 64 bit laptop that also has Intel integrated graphics on the processor. Solution: sudo apt-add-repository ppa:ubuntu-x-swat/x-updates && sudo apt-get update && sudo apt-get upgrade && sudo apt-get install nvidia-current

    Read the article

  • CUDA memory transfer issue

    - by Vaibhav Sundriyal
    I am trying to execute a code which first transfers data from CPU to GPU memory and vice-versa. In spite of increasing the volume of data, the data transfer time remains the same as if no data transfer is actually taking place. I am posting the code. #include <stdio.h> /* Core input/output operations */ #include <stdlib.h> /* Conversions, random numbers, memory allocation, etc. */ #include <math.h> /* Common mathematical functions */ #include <time.h> /* Converting between various date/time formats */ #include <cuda.h> /* CUDA related stuff */ #include <sys/time.h> __global__ void device_volume(float *x_d,float *y_d) { int index = blockIdx.x * blockDim.x + threadIdx.x; } int main(void) { float *x_h,*y_h,*x_d,*y_d,*z_h,*z_d; long long size=9999999; long long nbytes=size*sizeof(float); timeval t1,t2; double et; x_h=(float*)malloc(nbytes); y_h=(float*)malloc(nbytes); z_h=(float*)malloc(nbytes); cudaMalloc((void **)&x_d,size*sizeof(float)); cudaMalloc((void **)&y_d,size*sizeof(float)); cudaMalloc((void **)&z_d,size*sizeof(float)); gettimeofday(&t1,NULL); cudaMemcpy(x_d, x_h, nbytes, cudaMemcpyHostToDevice); cudaMemcpy(y_d, y_h, nbytes, cudaMemcpyHostToDevice); cudaMemcpy(z_d, z_h, nbytes, cudaMemcpyHostToDevice); gettimeofday(&t2,NULL); et = (t2.tv_sec - t1.tv_sec) * 1000.0; // sec to ms et += (t2.tv_usec - t1.tv_usec) / 1000.0; // us to ms printf("\n %ld\t\t%f\t\t",nbytes,et); et=0.0; //printf("%f %d\n",seconds,CLOCKS_PER_SEC); // launch a kernel with a single thread to greet from the device //device_volume<<<1,1>>>(x_d,y_d); gettimeofday(&t1,NULL); cudaMemcpy(x_h, x_d, nbytes, cudaMemcpyDeviceToHost); cudaMemcpy(y_h, y_d, nbytes, cudaMemcpyDeviceToHost); cudaMemcpy(z_h, z_d, nbytes, cudaMemcpyDeviceToHost); gettimeofday(&t2,NULL); et = (t2.tv_sec - t1.tv_sec) * 1000.0; // sec to ms et += (t2.tv_usec - t1.tv_usec) / 1000.0; // us to ms printf("%f\n",et); cudaFree(x_d); cudaFree(y_d); cudaFree(z_d); return 0; } Can anybody help me with this issue? Thanks

    Read the article

< Previous Page | 175 176 177 178 179 180 181 182 183 184 185 186  | Next Page >