1. Tear Apart


    from skul4aface / Added

    4,117 Plays / / 1 Comment

    From a functional car to a stripped shell in a day and a half.

    + More details
    • Nexus 30-GPU / 87.6 TFLOP iray cluster


      from jcdeblok / Added

      390 Plays / / 0 Comments

      I've just build a doomsday machine :) By using about ~30 EC2 GPU-instances, each a g2.2xlarge with a nVidia GRID K340 board, to all work on the same single-frame render things get done quite fast! First some numbers: 30 * 1536 = 46080 CUDA cores 30 * 2930 = 87.6 TFLOPs (yes, Terra-FLOPs :) 6.7 kWatt of electricity for GPUs alone. Operating cost 30* $0.11 (spot) = only $3.30 an hour in total. (I you would have to buy 30 of these videocard it would set you back $90,000 ) How it was done: I've build a framework called "Nexus", it provides communication between Max session, it can execute raw MXS remotely or trigger more complex macro type operations complete with callbacks and easy to add plugins. It's using ZMQ sockets (thanks Arthur for the tip!) which makes it quite fast an easy to connect to basically anything. One server runs the Nexus hub and each node sends it's messages to it and it gets relayed to a target-node or to an entire group. Each node can subscribe to a 'topic', it will subscripe to it's own ID, to 'ALL' and the render slaves will additional subscript to a 'servers' group. Each EC2 instances runs a '3dsmax.exe -server' on startup with a special startup-script that does a few things and then sets up channel with Nexus server. Once everything is set I can simply load a maxscene on all the servers at once by setting the target to 'servers' and executing this mxs for example: loadMaxFile @"\\\cloutput\iraytest.max" quite:true After pressing the 'render' button it will do a check on the available render servers and sends each a message with info on what bucket (region) to render, on completion a callback will send the bitmap back to the node that started the render operation. All the buckets get assembled as they come in, right now it sends raw base64-encoded bitmaps wrapped in XML so that's quite slow, it takes more time for the buckets to download then to render but that should be easy to fix! :) Just to make clear, all the rendering is done around the same time the first bucket comes in, all buckets are done in parallel. Now for the brilliant part, if I want to move the camera I can just do that on my workstation, then simply execute a MXS command that sends the new camera-transform to the entire 'servers' group. Each render-node now has the camera updated and when I press render it will reflect the update. And by sending over nodes, or parts of the scene, material changes etc can be updated with a single push of a button. So basically it's a hybrid DBR/ActiveShade technology. Nexus will work on any LAN/WAN (TCP sockets) etc and in due time I'll make Nexus publicly available but it needs some more time in the oven first. It can be used to send assets around the office, control a render farm or even controlling max from a browser.

      + More details
      • Direct accident management sound of the hemi cuda


        from direct accident / Added

        5 Plays / / 0 Comments

        Stay safe people, and enjoy the sweet sounds of this hemi cuda.

        + More details
        • CV Treasures doesn't collect hemis but we would like to


          from Dyna-e International / Added

          1 Play / / 0 Comments

          CV Treasures doesn't collect hemis but we would like to. Love the sounds of the hemi from this cuda. Enjoy! http://vimeo.com/dynaeinternational/cv-treasures-doesnt-collect-semis-but-we-would-like-to

          + More details
          • 1971 Plymouth Barracuda 440-6 Vray render hood up


            from Edward Nahorodny / Added

            18 Plays / / 0 Comments

            This is a model of a 1971 Plymouth Barracuda 440-6. Rendered in Vray with the hood up to display the motor.

            + More details
            • 1971 Plymouth Barracuda 440-6 Vray render


              from Edward Nahorodny / Added

              18 Plays / / 0 Comments

              This is a model of a 1971 Plymouth Barracuda 440-6, Rendered in Vray.

              + More details
              • Burnouts, Wheelies, and Drag Racing


                from Vidimation / Added

                You gotta have horsepower to go fast. This fast paced video comes close to the intensity and feel of being in one of these cars during two events at the Woodburn Dragstrip. There are plenty of Mopars including E-body and A-Body Cudas, Challengers, Dodge Darts, a Plymouth Duster called Lemon Twister and a few others. One of the events was the 2013 Mopar Nationals which explains all the Mopars. There are also dragsters and a Chevy or two. Are You Ready For This? 'Cuda Brothers - Preserving the Barracuda Legend www.cudabrothers.com LIke Us On Facebook www.facebook.com/CudaBrothers2

                + More details
                • GPU implementation of an interactive & real-time implicit fluid solver using CUDA


                  from Nick Avramoussis / Added

                  307 Plays / / 0 Comments

                  GPU implementation of an interactive & real-time implicit fluid solver using CUDA Check out the link below for more info! http://nidclip.wordpress.com/category/cuda/cuda-fluids/ Blog: http://nidclip.wordpress.com/

                  + More details
                  • Pike Leader Setup


                    from Fish Tales Fly Shop / Added

                    498 Plays / / 2 Comments

                    How to build your own Pike or other toothy critter wire leader. Includes a swivel for easy changing of tieable wire, as well as allows the fish to do their crazy spin they always seem to do.

                    + More details
                    • parallel_party - Custom Python module for GPGPU


                      from Timm Wagener / Added

                      185 Plays / / 0 Comments

                      What is it? -------------------------------------------------------------- A general purpose Python multithreading module. Easy and Pythonic access to CPU or GPU parallelization (CUDA) Features -------------------------------------------------------------- - Process Python types (lists etc.) in parallel on the GPU - DCC agnostic - Accelleration structures for minimized uploading to GPU memory or naive complete-process computation Known Issues -------------------------------------------------------------- - Extracting data out of Python types and packaging them again is not threadsafe and therefore has to be done serial. That is a major limiting factor right now, and a big hit for performance. I'm not sure if there is any way to speed this up? Ideas or suggestions are welcome! Specs -------------------------------------------------------------- Wrapping: Boost Python Python Version: 2.6x64, 2.7x64 CUDA: 5.5 Maya Versions: 2012, 2013, 2013.5, 2014 Dislcaimer -------------------------------------------------------------- This is just a test or proof of concept i've done for learning purposes, and is not ment to become any larger or evolve into a serious project.

                      + More details

                      What are Tags?


                      Tags are keywords that describe videos. For example, a video of your Hawaiian vacation might be tagged with "Hawaii," "beach," "surfing," and "sunburn."