Recording Aug 22 Event - Polaris vs Hyperblock


Thank you so much for your overwhelming response. For those who were not able to make it to this event here is the recording. Featuring @rob_marshall and @eschera


Please feel free to post any questions about anything especially Dimension Indexing, Calc Complexity etc.



Miz Logix





  • @Misbah @rob_marshall @eschera 

    Great hour. Thanks for recording. I tried to stay awake but couldn't make it.

    • Elizabeth, will the presentation be available on this post? Would love to get a copy.
    • Rob/Misbah will the best practices, as they are learned, be added to the planual? e.g. when to use summaries (wow, what a difference)
    • Elizabeth, Do list items also take 3x as much space? There's already a 500B overhead just for creating the list item.
    • Elizabeth, Does Polaris ignore a zero if we intend it to be zero? So if I need to count deliberate zeroes, say, inventory stock out, will I be able to get that count?
    • Elizabeth, Emergency Brakes / Cancel Calc! Yeah. Best feature imaginable. When will we get that in the normal workspace?
    • Rob/Elizabeth, Watched the dimensional indexing three times but still can't grasp the concept. Is member and line item synonymous? What is a member? Anyway, can that spreadsheet be published (Check Dims.xlsx). Maybe I just need to reverse engineer this. Can you put the idea of indices in the blueprint as an idea so we can upvote it? Sounds like a good idea. 
  • @JaredDolich 


    1. I doubt it but I will let Rob/Elizabeth answer that

    2. There will be another PLANUAL for Polaris.

    3. Elizabeth?

    4. I think all zeroes are considered as sparse whether there is a formula or not. Rob/Elizabeth?

    5. It's only in Polaris. 

    6. It's slightly tricky. Will try and explain 


  • @JaredDolich @Misbah 


    1. Let me see what I can do about the presentation
    2. Misbah is correct, there will be a new Planual for only Polaris
    3. Let me ping Elizabeth
    4. @Misbah is correct, a zero is a zero and doesn't count towards memory
    5. Correct, currently only in Polaris
    6. Dimensional Index - members is the number of members in a list which is why we are attempting to get the Dimensional Index number in the module blueprint.  Currently in classic, in order to get the Cell Count, you multiply the number of list members by list members of another list.  Instead of the product, the dimensional index is the aggregation of how many bits are needed from a power of 2 perspective.  If I have 2 members in a list, I will only need to allocate 1 index, if I have 3 members, then I will need 2 indexes (2^2= 4 which is greater than 3 - the number of list members.  Take a look at the chart below:

       Another way to think about it, what is the power of 2 needed to be greater than the members in a list - that will be your indexing amount.

    Does that help?



  • @JaredDolichadding on to what Rob shared.


    1) Not at this time

    4) Polaris will still include 0s as part of your calculations.  As Rob mentioned it will not take up memory and is ignored if the results will always be 0 (such as multiplications) but will be included for averages.  Or in the example below, you can have a formula of if 0 then 1 else 0 - BUT I would be very careful to include another condition to know if the zero is a stock out. Otherwise, you will end up creating a lot of unnecessary density.
    5) No plans to introduce this to the Classic engine at this time
    6) Yes please, an idea exchange would be a great way to surface this request more broadly. 
    The basics behind dimension indexing are:
    Each dimension requires a number of indices based on the number of members in the list (see Rob's table).  When you add up the indices required for all dimensions used in your line item the maximum is 64.  If for example, your list has 1,000 members it will use 10 indices because it falls between 512 and 1,024.  If you added 25 members to that list for a total of 1,025, this would require 11 indices since you exceeded the 1,024 max members within the 10 indices.  Each dimension is calculated independently to determine the number of indices that dimension requires.  The combination of the indices required for each dimension cannot exceed 64.