Jump to content


  • Content count

  • Donations

    0.00 CAD 
  • Joined

  • Last visited

  • Days Won


DaJuice last won the day on April 29

DaJuice had the most liked content!

Community Reputation

46 Excellent

1 Follower

About DaJuice

  • Rank
    Houdini Master
  • Birthday 03/01/1982

Personal Information

  • Name
    Siavash Tehrani
  1. ACES Rendering output

    In COPs you can use the OCIO Transform VOP inside a VOP COP2 Filter to convert images from ACEScg to srgb or rec709. The DMNK tools also provide a couple of convenient HDAs to streamline that: https://github.com/DominikLingenover/DMNK-Tools As far as external tools, PYCO is very nice: https://mrlixm.github.io/PYCO/standalone/ColorspaceConvert/home/
  2. I would guess the AMD system will be faster at everything, single-threaded and multi-threaded. The Xeons are probably somewhere in the neighborhood of 2.2-2.5GHz base and 3GHz boost, versus the 16-core TR CPUs which can be overclocked to around 4GHz all-core with adequate cooling. And that's setting aside architectural/generational differences and just looking at clock speeds. Also you can use ECC RAM with Threadripper (look at Unbuffered UDIMMs).
  3. It's possible that Pixar will implement their own output LOP (I believe RS and 3Delight already plan to) but for now I think you have to use a USD Render ROP LOP in conjunction with the Render Settings LOP.
  4. Redshift Render Settings Slow?

    Have you tried the automatic sampling in 3.0? You can enable it in the System tab.
  5. LOP's/USD workflow worth it?

    I think a lot of the shortcomings of Solaris at this stage are really more about the current state of USD. If you subscribe to the USD Google group you get an idea of just how much things are in flux on the rendering side.
  6. I don't believe those numbers in the TR specs are hard limits, more like what AMD will guarantee works with appropriately fast memory. For example the "System Memory Specification" for 1st gen was only 2667MHz, but I had no trouble running the HyperX kit (not the same as you linked) at the rated 3200MHz just using the XMP profiles. I did some comparisons of different memory speeds with Cinebench renders when I first built the system, but nothing comprehensive and nothing with Houdini sims. Those kinds of benchmark will probably be hard to come by. It did make a little bit of difference in Cinebench and I think that was comparing the stock 2400MHz speed vs the 3200MHz overclock. I feel like the difference for Houdini sims will be pretty negligible if we're talking 2666 vs 2933.
  7. Yup, that's the one. MSI hasn't bothered updating the specs (nor the BIOS in quite a while), but they have continued to expand the QVL at least. At the time 16GB DIMMs were the largest size available for mainstream and HEDT platforms. Now that has been bumped to 32GB, allowing for 256GB on motherboards with 8 DIMM slots, though I couldn't tell you for sure that all TR mobos support it. According to this forum post, the Taichi does: http://forum.asrock.com/forum_posts.asp?TID=12298&title=256gb-ram-on-x399-professional-gaming-motherboard As for use cases, when we were doing solves with Maya/PhoenixFD, heavier pyro and fluid sims often had to be scaled back because they would eventually creep past 128GB. I'd like to get into more sim stuff in Houdini for learning/personal projects and didn't want to run into that. Hope that helps
  8. I'm using this now: https://www.corsair.com/us/en/Categories/Products/Memory/VENGEANCE-LPX/p/CMK256GX4M8A2666C16 I chose it because... well, it was the only 256GB kit on my motherboard's QVL (MSI X399 Pro Carbon something something). 1st gen TR can be picky with memory and I didn't want to take chances with a non-QVL kit. I've only had it a couple of weeks ago but performance-wise it seems to be doing it's job. The HyperX ran at 3200MHz vs 2666MHz on this set, but in day-to-day tasks I can't tell a difference by feel alone.
  9. Are they both on the QVL for the Taichi? Assuming the HyperX kit is on the QVL it should be fine, and I don't feel like the tighter timings on the other RAM will make a big enough difference to justify the price. For what it's worth, before upgrading to a larger kit I had a 32GB HyperX kit in my 1st gen TR system and it worked without issues (although it was Hynix rather than Samsung b-die).
  10. Studio Workstation

    What Atom said, that amount of memory is inadequate for a workstation nowadays. Everything else looks okay,
  11. Houdini on AMD EPYC?

    Howdy, Is anyone running Houdini on an EPYC system, and if so, have you experienced any issues? I am thinking of the hard crashes that myself and others had with Core Performance Boost enabled on Threadripper machines. I don't know if that is still an issue with Zen2 Threadripper systems or if EPYC even uses the same clock boosting scheme.
  12. Redshift Volume Motion Blur need Help!

    This is a known issue with Redshift. As in, motion blur on volumes does something, but the results are not accurate. From what I understand the devs are working on updating the volume tech to add support for things like multi-scattering, so hopefully we'll see a fix for this as well.
  13. Anti-Aliasing setting in Redshift?

    The Min Samples, Max Samples and Threshold settings under Unified Sampling are your main AA controls. I would check out this tutorial on sampling in Redshift:
  14. Uncheck 'Match Normals with Terrain". The 'Random Up' slider allows you to add some randomness back in.
  15. Houdini Workstation

    They are a mix of Corsair coolers like the H115i I believe. Not a perfect solution as they don't cover the full heat spreader on Threadripper, but they perform well enough and last more than 6 months unlike the Enermax stuff. At home I use a Noctua NH-U14S air cooler.