Performance of RTX A5000 with Context Capture

Mariusz,

GPUs play a secondary role in terms of performance. There is not much of a speed difference between using 1080ti, RTX2080ti, and RTX 3080. What WILL make a difference is a powerful multicore CPU. For Example AMD Rzen 5900X or 5950x o the latest Intel CPUs.

Hello Adam

For ContextCapture their are two main processes

1. Aerotriangulation (AT) 2. Reconstruction

Of the two, reconstruction takes the greatest amount of time. It primarily uses the GPU. A quality GPU can save hours or even days on a production over lesser GPU

The CPU can enhance the performance of a reconstruction however it has a greater impact on the AT...with ContextCapture the speed (faster) will be a greater impact than number of cores, as not every function of the AT is multi threaded

I would suggest revising your post to avoid confusion

Thanks 

Hello Adam

For ContextCapture their are two main processes

1. Aerotriangulation (AT) 2. Reconstruction

Of the two, reconstruction takes the greatest amount of time. It primarily uses the GPU. A quality GPU can save hours or even days on a production over lesser GPU

The CPU can enhance the performance of a reconstruction however it has a greater impact on the AT...with ContextCapture the speed (faster) will be a greater impact than number of cores, as not every function of the AT is multi threaded

I would suggest revising your post to avoid confusion

Thanks 

Hello,

I did not find a significant difference between GTX 1080TI and RTX 3080 while performing both AT and Reconstruction steps. I have shaved days off processing time by switching from Ryzen 7 3700X to Ryzen 9 5900x

Some examples of applications that I work with:

Edited out 

(give it a few minutes for the textures to load)

I think that having a high end-ish GPU is important and there IS a difference of course. Perhaps the main factor is the amount of RAM available to the GPU?

Anything 8GB and up is great. 

Edit: I am typing it from a workstation running a reconstruction job. I have been monitoring the resources and my GPU's utilization doesn't go past 10-15% max and it typically stays under 5%. 

Edit2: From my own experience, Pix4D benchmarks done by pudget systems also DO apply somewhat to Context Capture given that both software titles use CUDA and apply a multi step approach to photogrammetry reconstructions. 

https://www.pugetsystems.com/recommended/Recommended-Systems-for-Pix4D-207/Hardware-Recommendations

Hi Adam

The two cards you compare are both high end cards..I would not expect a big difference between these

I work in I our reality Modeling team since we acquired Acute 3d over 6 years ago.. what I am explaining comes directly from our development team.. I just do not want people confused 

The cpu example you give is an example of a faster 5900ghz processor outperforming the 3700ghz processor. As I mention speed is preferred over more cores. I have seen this on test machines where cpu is overlooked. But even on these if an inadequate gpu were used (m2000 ect) it woul slow considerably

On another topic I would like to know what uav you use in you profile pic ..did you build it yourself?

Hi Michael,

I agree with your conclusion: A higher end Nvidia GPU is a great choice for CC. 

As to this specific photo: this was a tethered M600 Pro with radio equipment payload. They were used as an "on demand" "towers" retransmitting radio signals. 

Thank you for your input and feedback. 

Sounds cool...

BTW the towers look really good.. thanks for sharing