Subscribe to this thread
Home - General / All posts - New video: Mean Curvature in 2 minutes with GPU, 5 days without
Dimitri


7,413 post(s)
#06-Feb-19 13:45

There's a new video in the Gallery page that does two amazing things: first, it shows how incredibly fast GPU parallelism is for big computations, showing how Manifold can load even 1280 cores to 97% use. Second, even more amazing, it shows how some of those seemingly-obscure Transforms, like Curvature, Mean, can be used in real-life to create very useful and visually appealing displays. :-)

drtees
203 post(s)
#07-Feb-19 18:58

I use LIDAR continually. Looking forward to trying Mean Curvature on some of our projects to see if the enhanced imagery helps to identify potential critical areas.

drtees
203 post(s)
#07-Feb-19 19:29

I used a much smaller LIDAR file than the one in the video. I have far fewer GPU cores and only four CPU cores. Using a radius of 5 instead of a radius of 3, on a LIDAR set encompassing approximately 806 hectares, M9 took slightly over 2 seconds to render the image. Tried the same trick using M8 and got an answer in 0.986 seconds. I was not able in the brief time I had to play with it achieve the same Autocontrast-style image created in M9. Still very impressive.

Dimitri


7,413 post(s)
#07-Feb-19 19:48

Release 8 loads everything memory resident, so for small jobs, 8 can sometimes go faster than 9. That's great, except that when it tops out things go slow and stay slow. 9 has a different system, so even very large jobs go fast, much faster than 8.

Here's an interesting idea: try doing what the first GPU video illustrates, and turn off GPU using the pragma directive

PRAGMA ('gpgpu' = 'none');

It might just run faster without GPU for a small job. :-)

Dimitri


7,413 post(s)
#08-Feb-19 06:47

Forgot to mention: details matter when analyzing timings, especially when comparing two very different systems like 8 and 9. For example, to make sense of timings you have to know the exact GPU card in use. Why? Here's an example:

8 will work with older generation, pre-Fermi GPU cards. 9 won't, as 9 is built around reasonably recent, Fermi or later GPU cards that can work with reasonably recent CUDA.

If you try doing a mean curvature in 8 with an older GPU card, 8 could use it, but if it is pre-Fermi 9 won't use it (you can tell by looking in the Help dialog for 9) and instead will execute the job CPU-parallel. CPU parallelism in 9 is quick enough that it can often look like the GPU is being used. A radius of 5 for anything other than small data probably would be too much computation to mistake CPU parallelism for GPGPU, but if the data is smaller that could be an effect.

lionel

995 post(s)
#08-Feb-19 15:03

some informations about curvative ( a lot of mathematic )

-Understanding curvature rasters

-Using curvature rasters to enhance terrain representation

-Esri Press; Map Use: Reading, Analysis, Interpretation 8 edition (September 19, 2016) review

-LSDTopoTools

-3D Analyst or Spatial Analyst

Manifold changelog


Book about Science , cosmological model , Interstellar travels

Boyle surface fr ,en

Manifold User Community Use Agreement Copyright (C) 2007-2021 Manifold Software Limited. All rights reserved.