catchyid Posted February 20, 2020 Share Posted February 20, 2020 Hello, I want to run pyro/fluid sim on a farm which has its own job manager (i.e. not using HQueue). If I want to use this job manager (coalition) to render using Mantra, then I would write some scripts to send each frame to a different machine (i.e. it's doable). However, If I want to sim pyro volumes on the farm (and as far as I understand), I should have a tracker that exchanges volume data on the boundaries between machines --> How easy/difficult to setup If I were to use a different job manager? I still need to research the topic (initially test how things work using HQueue), but I just need a very rough estimate if this task is doable or hard? Thanks, Quote Link to comment Share on other sites More sharing options...
dleonhardt Posted February 20, 2020 Share Posted February 20, 2020 (edited) Hello Khaled It is absolutely doable. How easy it will be in the end depends on which Job Manager you are using. Deadline for example supports submitting the HQueue Simulation ROP directly. Otherwise you need to make sure that: 1. One PC runs the simtracker.py script, you have to set the IP adress of that PC and the port that the simulation tracker runs on, inside your Gas Net Slice Microsolvers or the control null that the shelf tool creates 2. Each PC has it's unique $SLICE variable set from 0- the total number of slices minus 1, which would look like this: Number of Slices:3 PC1: $SLICE=0 PC2: $SLICE=1 PC3: $SLICE=2 The shelf tool takes care of the initial setup. EDIT: I just read now that you are using Coalition as your Job Manager. Edited February 20, 2020 by dleonhardt 1 Quote Link to comment Share on other sites More sharing options...
catchyid Posted February 20, 2020 Author Share Posted February 20, 2020 Thanks Dennis Really appreciate your help ! Just one more question: time-wise, I will test distributed simulation first using HQueue (just make sure I understand how things work), then implement it on our job manager, I "think" two to four weeks is enough to have the system running? I have a strong technical background (both IT and programming), so I think this would be a super safe bid? Thanks Quote Link to comment Share on other sites More sharing options...
dleonhardt Posted February 20, 2020 Share Posted February 20, 2020 1 minute ago, catchyid said: Thanks Dennis Really appreciate your help ! Just one more question: time-wise, I will test distributed simulation first using HQueue (just make sure I understand how things work), then implement it on our job manager, I "think" two to four weeks is enough to have the system running? I have a strong technical background (both IT and programming), so I think this would be a super safe bid? Thanks Yep, should be doable in that timeframe. 1 Quote Link to comment Share on other sites More sharing options...
anim Posted February 20, 2020 Share Posted February 20, 2020 here Jeff describes how to use implement custom distributed sim (towards the end of the video) https://www.sidefx.com/tutorials/h15-masterclass-distributed-simulations/ also TOPs support distributed sims, not sure if it works with all schedulers but may be worth looking into 1 Quote Link to comment Share on other sites More sharing options...
catchyid Posted February 20, 2020 Author Share Posted February 20, 2020 Thanks Tomas Will have a look Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.