Support Website Contact Support Blog

Pix4d mapper with GPGPU?

As of today… If money was no object what PC specs would everyone recommend getting to get a 3d photogrammetry model out as fast as possible? I am open for not just one system, but maybe multiple systems working on different collection sets. Also, I was looking to see if there has been any thought to if GPGPUs would do a better job than say an NVIDIA RTX 2080ti. Any help would be much appreciated.

Hi Joshua,

It is always hard to say which hardware is the greatest as at the beginning of considerations like that we need to ask ourselves a question for what kind of purposes do we need a new computer, what kind of data we will be processing on it, what goal we want to archive etc. As you can see the answer depends on many factors, therefore, there is no one solution for each case. Also, not everything was tested and potentially, somewhere out there are still some undiscovered resources/solutions/components combinations. However, I can share with you and the rest of the community some insights about the best hardware for Pix4Dmapper. 

What we do know for now is when configuring a computer, the most important is to have a balanced hardware configuration so the bottleneck does not occur during the processing, thus it is important to have a configuration when one component is not weaker than all others. However, most of the processing is done on the CPU. A faster CPU is the first key to increase processing speed. Pix4D utilize additional cores; not all the steps can be parallelized but by adding additional cores the process will run much faster and you will see a noticeable benefit from it. Clock speed also has a significant impact on project processing time. There is no definitive answer to which has the greatest impact but it is clear that both play a significant role, mostly in step 1 and 2 processing speeds. It was tested that so far Intel’s Core i9 processors offer the best results. 

In terms of processing time and for the maximal performance, it is better to store the images and process the project on SSD. It is not necessary to have Pix4Dmapper installed on the SSD, but it would allow the software to launch faster. Let’s suggest 1TB PCIe.

This depends among other on the number of images, their resolution, size, the overlap between images and processing options that you choose. 128GB of RAM should be fairly enough :slight_smile:

Pix4Dmapper is compatible with any GPU that is compatible with OpenGL 3.2 or above. However, for a faster processing, Pix4Dmapper also uses the processing power of GPUs that are compatible with Nvidia CUDA 9.1 and above (with the latest drivers installed). This allows increasing the performance in Pix4Dmapper especially during step 1 and with large projects. We recommend using GeForce GPU (from the mid-range GTX 1060 on up all perform well) because they are usually cheaper with regards to the performances in comparison with other graphics cards. Titan and Quadro technically work too but they are much slower than GeForce. Just one good graphics card is enough for all your needs. The second seems not to be that helpful. GTX 1080 Ti was the top so far. For now, this is all that we can do for the best performance with GPU and Pix4dmapper.