DavidGretzschel
and some conjecture on why that might be and how it might be fixed 
[which as I don't know the code might be extremely wrong]

tested on Plusle:

This is a continuation of what I found in this thread:
https://forums.thebrain.com/post/beta-version-11-0-46-0-on-mac-has-huge-performance-issue-10399657?pid=1310505882
That thread however is about Mac-performance and might be unrelated.
In there I could only test with 1080p, now I could test the performance in 4k.

[note that when I talk about "performance", here I mainly mean:
"How fast will the focus-highlight move from Thought to Thought".
This is my number one performance metric, because if it's too slow, I won't be able to use TheBrain]

findings:
TheBrain is perfectly capable and happy to show me 200 Thoughts at once. This is extremely useful to me.
In the example I can quickly scan all the threads that I have started in this forum at once.

When I zoom in and Thought-names get shortened by ...-abbreviations, then during the zoom-process there will be freeze lasting a couple seconds. Then afterwards the focus-highlight navigation performance will be so slow, as to be unusable. It also drops inputs. So I click up-arrow four times, but only two of the commands will be executed.

https://www.loom.com/share/fc27db1840ff41a2ae0852d14a6518a4

Fundamentally a Thought is just a buch of text on screen. Each Thought has a line connecting it to the Parent-Thought. So add 200 more lines to be rendered. There's no possible way, that should be a performance challenge.
I suspect that something is wrong with the calculation of the ...-abbreviations.

[guess A, mainly that there's something that doesn't scale linearly, not necessarily in the way I illustrate]

So that when those get calculated, TheBrain tries to do collision-detection with other Thoughts. And somehow each Thought has to calculate whether it can collide with all other Thoughts and should be shortened and where.
Let n be a Thought, t be the cost of computing the abbreviation/handling the collision per Thought.
So whenever I add another Thought, it'll mean that each Thought has to compute whether it collides with it.
The computational costs of t(n)= n^2
That's not a problem with n= 20, because t(20) = 0.4k. (which must still be low, as performance is good if I have 20 Thoughts somewhere on the Plex and am zoomed in enough to give them all dots)
But with n=200, t(n) = 40k
So we have only 10 times more Thoughts on screen, but 100 times the computation has to be done.
That would explain why there's a freeze, when I zoom further in. The calculation gets a lot more expensive.
[because presumably, there was so much space to spread out for the Thoughts, that no collision detection/abbreviation was done]

possible quickfix:
Allow the user to disable the abbreviations (and/or collision detection if it's involved) entirely. When Thoughts then display over each other, simply give the left Thought the foreground. Now the user can quickly zoom out and find a view, where everything looks nice.

[guess B, somewhat confident in]
Zooming further out again also freezes, because TheBrain immediately tries to calculate the intermediate zoom-states.

possible quickfix:
Option to only render a new zoom-setting after a 1 second delay. It won't feel responsive and smooth and it will feel a little awkward. But Zoom rarely feels responsive, anyway.

[guess C, this one must be somewhat true]
The focus-highlight is just a small grey circle with a dot in the middle. How can the rendering of where it's supposed to go next possibly lag?
It's not complicated. A three year old would be able to point to the Thought where the circle should go, if I press right.
TheBrain somehow can't figure this out, when it has all of Plusle's resources at its disposal?
A toaster should be able to do this, but somehow this is a problem.
Except the navigation itself probably isn't the problem.
I highly suspect that with each press of an arrow-key, TheBrain tries to rerender the entire Plex, instead of merely rerendering the Thought where the focus-highlight used to be and the Thought where it appears next.

possible quickfix:
Only render the two affected Thoughts, when moving the focus-highlight.

@eeik
Did a 200-Thought Plex like that ever render with decent performance in TheBrain10?

Lenovo Legion Y720 | 32Gb Ram | i5-7300-HQ, 4 cores @2.5 Ghz | GTX 1060 | :{Windows 10 (latest build) |  SSD: Model SAMSUNG MZVLW256HEHP-000L2

Laptop monitor: 15,6 inch, 1920x1080@60Hz
external monitor: 42.5 inch, 3840x2160@60Hz
(I use one or the other, never both at once)
--
Samsung Galaxy Tab S6 256Gb LTE
Samsung Galaxy S10 Plus 256Gb LTE
---
unless otherwise specified, assume 100% scaling
My videos always show the current time & date (look at Windows Taskbar).
---
How to use TheBrain 11 fast (work in progress)

Quote
eeik
@DavidGretzschel , ha! Not a chance! I can only dream of a day where displaying that many thoughts yields the performance that I would expect of a fat client based GPU acceleration capable app. I always bring up the Swiss Music Chart web experience which is buttery smooth but the team doesn't seem to wish to tackle this issue truly head on from a bottom up performance first perspective. I've shared a bunch of content with the team on this topic maybe 8,9,10 months ago or so.

The thing that really cripples me, no line drawing required, is the search results being added to a selection for quick perusal and multi-deselection features that aren't well supported. Add 200 or more thoughts to the selection pane and try to work with them and I'll almost always get a crash...

The crux of the issue, as I'm hypothesizing, is not necessarily the line drawing and collision detection (although, design it to render properly and you calculate out positions from the onset so collision detection may not actually be involved), but rather the overly eager thumbnail caching when you get that many thoughts on screen. Because reasons it seems the plex loads the thoughts and before settling in for the user to keep working it aggressively recalculates all the thumbnails even if you never change your zoom levels or in park the window from it's maximized state. Super inefficient and CPU intensive with what seems like an inadequate in-memory caching approach (capped at 128MB or so? from my recollection of the logs). Every image viewer software worth its salt approaches large libraries of images, these days in potentially raw form which mandates an appropriate, performant, non-UI blocking caching approach which TheBrain has not yet opted to implement properly for reasons unknown. Image caching should never bring down the app ... ever!

Hopefully v12 will address this, codename, Imagery Ida? This is, again, my working hypothesis that remains unproven but not yet disproven behind all of my hard crashes.

//E



Quote
mcaton
We're always working on making TheBrain lighter/stronger/faster/etc.  As is the nature with ANY application or operating system, the more advanced it becomes, the more we (as users) ask of it. I appreciate your thoughts and feedback and I'll share your experiences with the team.

Thank you,
Matt
Quote
DavidGretzschel

@eeik
Thumbnail caching? Like.... preparing the CPU to be quick with spitting out a big mouse-hover image, just in case I would want to hover over one of the Thought icons? Or do you mean something else with image-caching? The thumbnail would just be the TheBrain-forums logo. 200 times.
I can do without icons, if that gives me fluid performance. I don't see how this is more of a challenge zoomed in, than zoomed out.
I'd expect the opposite really. [aside: How do I remove 200 icons?]
Anyway, if I'm zoomed out enough to see all 200 without dots, I'm able to move via focus-highlight. I also selected all of them at once by ctrl+clicking the Child-Gate. Loaded into Selection within half a second. Also could connect all of them to another parent instantly.
Copy and pasting them to my personal Brain takes 30 seconds, which isn't fun at all. But if I zoomed out in preparation and dragged the Plex up (not sure if that matters) it'll not even crash. Copy and delete was instant.

This 200-Thought Plex seems workable for me. I didn't create it to make a performance test, but to get an overview of all the topics I started.
And it really does do the job. The zoomed-out-view without abbreviations is what I would prefer in any case.
It's just a bit awkward to set the zoom settings first and then navigate to the big Thought in question.
I'm obviously none too happy, that there's performance considerations here at all, but I'm cautiously optimistic that this won't be crippling
or force me to adapt too much.

Otherwise, as long as one of the evil navigation bugs forces a restart, using TheBrain is a lot of fun.
I'm getting really fast at it, yet the software can keep up with my thoughts.
And I'm better organized than ever before in my life.

@mcaton
Thank you, that's good to hear.

Lenovo Legion Y720 | 32Gb Ram | i5-7300-HQ, 4 cores @2.5 Ghz | GTX 1060 | :{Windows 10 (latest build) |  SSD: Model SAMSUNG MZVLW256HEHP-000L2

Laptop monitor: 15,6 inch, 1920x1080@60Hz
external monitor: 42.5 inch, 3840x2160@60Hz
(I use one or the other, never both at once)
--
Samsung Galaxy Tab S6 256Gb LTE
Samsung Galaxy S10 Plus 256Gb LTE
---
unless otherwise specified, assume 100% scaling
My videos always show the current time & date (look at Windows Taskbar).
---
How to use TheBrain 11 fast (work in progress)

Quote

Newsletter Signup  Newsletter        Visit TheBrain Blog   Blog       Follow us on Twitter   Twitter       Like Us on Facebook   Facebook         Watch Us on Youtube  YouTube       

TheBrain Mind Map & Mindmapping Software     Download TheBrain Mind Mapping Software