FPS-Drop basically doing nothing

Hey everyone! I’ve got a question concerning the performance of TD on my system for you. When I was following along a beginner’s tutorial focused on basic noises I noticed a significant fps-drop: Just one single sparse noise TOP-operator running on a resolution of 256x256 renders my system nearly useless (<15 fps). The problem occurred when I animated one value of “translate” via “absTime.seconds”.
Info tells me the following:

Type: Noise TOP
Total Cooks: 1514
CPU Cook Time: 21.224 ms
GPU Cook Time: 0.000 ms
Cook Frame: 102

Size: 256 256
Aspect: (16:16) (1:1)
Pixels: 65536
Format: 8-bit fixed (RGBA)
Mipmaps: No
Fillmode: Fit Best
GPU Mem this TOP: 0.25 MB
GPU Mem all TOPs: 40.21 MB
Total GPU Mem: 87.10 of 8176.00 MB

I run Touchdesigner on a 16" MacBook pro with a 2,3 GHz i9 processor, 32 GB RAM, and an AMD Radeon Pro 5500M with 8GB of RAM. I know… laptops aren’t your favorite devices here. I run several other pieces of generative imaging software (vuo, resolume, vdmx5) and never experienced something like this even while processing large amounts of stuff in parallel. This makes me scared my system might be inferior for using TD. What do you think?

Hey, can you share your .toe file so we can look at the exact settings you are using on the Noise TOP?

Here you go :relaxed: It gets even worse when I switch off “monochromatic”. Then we are talking about a 2-4 fps slide show.

@malcolm - correct me if I’m wrong, but I don’t think the sparse noise type is GPU friendly. If I change to a GPU-centric noise type my FPS goes back up when I test this toe.

1 Like

@raganmd Yep thats exectly it.

@Friedrichter, you’ll want use the noise types that have (GPU) in their name to generate noise quickly. The other ones are CPU based and very slow.

1 Like

Thank you for pointing me to this one :smile:! I was hoping for an easy solution. I now rewatched elekktronauts awesome beginners tutorial and noticed he uses the sparse noise only for the CHOP-Operator. It’s as most times a thing of looking or listening more closely :sweat_smile:. Thank you for your time.