Standin-ception - ram usage test

[Andrew_Wilkins1]

Hello,

I ran a quick test and wanted to ask about the results, since it's not exactly what I was expecting.

default sphere exported to "default_sphere.ass"

test #1:
2550 aiStandin nodes (duplicate / copies) pathed to the same "default_sphere.ass" file.

RAM usage:

00:00:09  4333MB         | peak CPU memory used     4333.43MB 
00:00:09  4333MB         |  at startup              4266.49MB 
00:00:09  4333MB         |  plugins                    8.19MB 
00:00:09  4333MB         |  AOV samples               11.59MB 
00:00:09  4333MB         |  output buffers             1.13MB 
00:00:09  4333MB         |  node overhead              2.35MB 
00:00:09  4333MB         |  instance overhead          0.67MB 
00:00:09  4333MB         |  message passing            0.02MB 
00:00:09  4333MB         |  memory pools              14.55MB 
00:00:09  4333MB         |  geometry                   0.26MB 
00:00:09  4333MB         |   polymesh                  0.03MB 
00:00:09  4333MB         |  accel structs              0.18MB 
00:00:09  4333MB         |  skydome importance map     7.80MB 
00:00:09  4333MB         |  strings                   24.25MB 
00:00:09  4333MB         |  profiler                   0.74MB 
00:00:09  4333MB         |  unaccounted               -4.79MB (note: interactive session)

test #2:
Those same 2550 aiStandin exported to a new ASS file.... "mega.ass":
1x aiStandin pathed to this "mega.ass" file.

RAM usage:

00:00:07  1605MB         | peak CPU memory used     1605.93MB 
00:00:07  1605MB         |  at startup              1536.26MB 
00:00:07  1605MB         |  plugins                   11.98MB 
00:00:07  1605MB         |  AOV samples               11.59MB 
00:00:07  1605MB         |  output buffers             1.13MB 
00:00:07  1605MB         |  node overhead              2.35MB 
00:00:07  1605MB         |  instance overhead          0.67MB 
00:00:07  1605MB         |  message passing            0.02MB 
00:00:07  1605MB         |  memory pools              14.55MB 
00:00:07  1605MB         |  geometry                   0.27MB 
00:00:07  1605MB         |   polymesh                  0.01MB 
00:00:07  1605MB         |  accel structs              0.17MB 
00:00:07  1605MB         |  skydome importance map     7.80MB 
00:00:07  1605MB         |  strings                   24.25MB 
00:00:07  1605MB         |  profiler                   0.66MB 
00:00:07  1605MB         |  unaccounted               -5.77MB (note: interactive session)

---

I would've expected RAM usage to at least be the same or worse, but not roughly 37% less RAM usage? Any ideas what happens here?

All the best,

Andrew

---

Maya: 2022
MtoA: 4.2.1
Arnold: 6.2.0.1

[Andrew_Wilkins1]

Additionally, when setting the "mega.ass" aiStandin - View Port display mode to "shaded" (or any other setting)—we seem to have a view-port display bug?

Only the first 2 spheres appear to show in the view port.

[thiago.ize · Accepted solution]

Arnold is using 67MB and 69MB in both of those tests. That big number you see is due to maya's memory usage before Arnold starts. So test1 is causing maya to load/store much more data, but arnold itself doesn't see that.

[Andrew_Wilkins1]

Hey Thiago,

Right, thanks very much!