Jump to content

lukeiamyourfather

Members
  • Content count

    901
  • Donations

    0.00 CAD 
  • Joined

  • Last visited

  • Days Won

    10

Everything posted by lukeiamyourfather

  1. HQueue Substep caching of geometry NOT WORKING

    Another option if you must cache it out at the SOP level for some reason is to use a Time Shift SOP, cache it out, then shift it back. The frames would still be integers but they'd be whatever fraction of the actual time.
  2. HQueue Substep caching of geometry NOT WORKING

    You can do $SF in the DOP. Don't cache it out at the SOP level.
  3. Caching substeps via HQUEUE

    Cache out at the DOP level and use $SF for the frame number in the file name. Submit that ROP in the DOP to the farm. https://www.sidefx.com/docs/houdini/nodes/dop/rop_dop.html This isn't unique to HQueue. Any queue manager can do the same.
  4. Trying to learn Vex. Major differences from typical OOP?

    What are you wanting to do with VEX? What is your goal? I'm guessing you're new to Houdini so there might be better ways to accomplish your goal depending on what it is.
  5. Compiled Block Nodes Usage

    I'm using them to spawn millions and millions of instances for large environments. Without the compile blocks it takes much longer (like half an hour versus a minute or two).
  6. Open CL error

    You're not using the same operating system or hardware as the original poster and this thread is four years old. Please don't do this. Open a new thread.
  7. Python Install locations in OSX and Linux

    Houdini uses it's own version of Python installed along with the application.
  8. consistent file cache issues...

    Change the OpenCL device to the CPU and try again. Also what hardware are you using?
  9. MSI GL63 8SE laptop

    Assuming you're not using anything that relies on the older version of Qt, I'd try updating the video card drivers (downloaded directly from Nvidia) and update Windows. If that doesn't change anything then maybe try a vanilla install of Windows instead of the manufacturer provided Windows. If you're using something that relies on the older version of Qt then continue with the version of Houdini, with legacy Qt support, that works.
  10. Building a beast Houdini + Redshift mainly

    I don't trust products from ASRock but otherwise I think that's a decent plan. I had a server board from ASRock die three times for the same reason. They replaced it twice under warranty and the third time it was out of warranty (replaced it with a different brand board). That was my first and last board from ASRock.
  11. Pyro renders taking way too long

    Rendering volumes with physically based lighting and shading is incredibly expensive. If you need fast render times then consider deep shadow maps instead of ray tracing for the lighting.
  12. Building a beast Houdini + Redshift mainly

    Threadripper is a great option if you don't need a ton of memory (128 gigabytes or less).
  13. Building a beast Houdini + Redshift mainly

    That's less relevant these days because of the "turbo" mode modern processors have. I used to work on workstations with the Xeon E5-2630 v4 and they would regularly hit 3.5GHz for single threaded tasks. The listed "turbo" is 3.1GHz but it would go well above that with sufficient cooling. Either way I wouldn't go with those processors because they're old but the base frequency isn't an issue.
  14. Building a beast Houdini + Redshift mainly

    I wouldn't use hardware from two generations back. If you're fine with older hardware like that then just buy used at this point.
  15. CloseUp, HiRes Pyro/Smoke renders

    If you're going for wispy smoke (like cigarette smoke) then advect many particles through the simulation and render those. The solver isn't good at retaining details for wispy smoke. If you search the forum there are a couple of examples for doing this. If you're not going for wispy smoke maybe you can share what look you're trying to obtain?
  16. GPU in Houdini

    Yes. In the case of Houdini, like Pyro, it doesn't use multiple OpenCL devices but in theory it could. Probably easier said than done. You could develop your own tools that use multiple GPU (nontrivial and would need appropriate skills to accomplish). No idea about NVLink and OpenCL memory. I would assume the answer is no until someone shows it working with Houdini. Maybe email support? I wouldn't be surprised if someone at SESI has tested this scenario out.
  17. Lockheed Martin

    The team I work with has some full-time openings. We use Houdini for part of our pipeline and we're wanting to grow it. I can answer some general questions but I'm not the hiring manager. 462928BR - Full Time https://www.lockheedmartinjobs.com/job/grand-prairie/multimedia-design-engineer-grand-prairie-texas/694/10153801 462927BR - Full Time https://www.lockheedmartinjobs.com/job/grand-prairie/multimedia-design-engineer-grand-prairie-texas/694/10151727 462583BR - Full Time https://www.lockheedmartinjobs.com/job/grand-prairie/multimedia-design-engineer-grand-prairie-texas/694/10151739
  18. GPU in Houdini

    Houdini will use the GPU for a few things that are OpenCL accelerated like Pyro simulations and FLIP simulations. Using the GPU for things like that relies on the GPU having enough memory to store all of the data so at some point only the CPU will be able to handle the workload. Houdini won't use more than one GPU for any given task. For example the viewport is drawn with a single GPU. A simulation that uses OpenCL will run on a single GPU. Environment variables can be used to define what card to use for what. For Houdini one higher performance GPU with more memory is a better option than multiple GPU. It can in theory but it's up to the developer to decide how to manage multiple devices, memory, etc. For a typical workstation go with the best single GPU that fits in the budget.
  19. Disco Dingo

    Unless there's a new feature or bug fix to get around a show stopping problem I'd stick with LTS releases like 18.04 LTS. The next LTS release will be 20.04 LTS (every two years). By the way the Linux kernel versions are arbitrary. They don't signify anything like major features or changes. They're just another number. https://www.theregister.co.uk/2019/03/05/linux_50/
  20. Lockheed Martin

    The job listings on this other thread (linked below) are still open.
  21. Houdini 17.5.173 Centos PDG cores use

    Linux doesn't have anything to do with it. Some algorithms used by Houdini use a single processor core. In those cases performance can be improved by running multiple tasks concurrently. You would see 100% processor utilization only if there were more concurrent tasks than processor cores. Other algorithms will use all processor cores available. It all depends on the specific scene and how it's built as to how well it will scale to multiple processor cores.
  22. Lockheed Martin (Suffolk, VA)

    There are two positions at Lockheed Martin available in Suffolk, VA. Both are full time positions and on site. I'm posting here because Houdini experience is a huge plus. I'm not the hiring manager but I can answer questions. https://www.lockheedmartinjobs.com/job/suffolk/multimedia-design-engineer-suffolk-va/694/10916693 https://www.lockheedmartinjobs.com/job/suffolk/multimedia-design-engineer-suffolk-va/694/10916704 The facility is very nice and so are the people who work there. They have some information about the facility on the website. https://www.lockheedmartin.com/en-us/capabilities/research-labs/center-for-innovation.html This video shows the kind of work the team does but take this with a grain of salt because it's from projects that are five years old or more. Unfortunately the cool stuff isn't public but I assure you there's cool stuff to work on!
  23. Lockheed Martin (Suffolk, VA)

    United States citizenship is required to get a clearance in most cases. These positions require a secret clearance (shown at the bottom of the job posting) which the company will help employees obtain.
  24. Lockheed Martin

    These positions have been filled but there are more available. I'm going to create a new thread for those.
  25. Workflow pipeline building

    The plugins for Substance are awful. Just export textures from Substance. You're not missing anything there by not using the plugins. I would use Mantra instead of Redshift but out of all the GPU renderers Redshift is the best option if you must have a GPU renderer. Here be dragons... Maxon purchased Redshift last week. They say they're going to leave it alone and it'll be business as usual (and third party application plugins will continue) but time will tell if they stick with that or not. https://www.maxon.net/en/news/press-releases/article/maxon-acquires-redshift-rendering-technologies/
×