Metashape 1.5.5 CPU Performance: AMD Ryzen 9 3950XWritten on November 14, 2019 by William George
Metashape, formerly known as PhotoScan, is a photogrammetry program that takes a set of images and combines them to create a 3D model or map. Processing of those images into point clouds and 3D meshes/textures is time-consuming, heavily using a computer's CPU and GPU. With the release of AMD's latest Ryzen 9 3950X processor, which combines a high core count with good per-core performance, we wanted to see how it and its sibling 3900X fare in this application compared to the reigning champ: Intel's Core i9 9900K.
Our full suite of Metashape testing takes many hours to run, especially when doing 2-3 passes per CPU, and we only have a single license for Metashape Pro - so the number of CPUs we could include in this roundup was limited. We've got fairly recent data showing the Core i9 9900K offers the best performance, so this time around we tested only the 9900K and the two new AMD Ryzen 9 processors.
A complicating factor is memory speed support, though: most processors on the market today officially support up to 2666MHz memory, while these 3rd gen Ryzen chips are rated for different maximum memory speeds depending on how many RAM modules are installed. The spread goes from 2666MHz up to 3200MHz (if only using two modules) - but that limit would mean less total memory capacity, which could impact some of our benchmarks. Because of that, I decided to test the Ryzen processors at both ends: with a full set of four 16GB modules at both 2666 and 3200MHz, even though that is technically outside the supported range. We can't sell systems configured that way, because we need to stick to AMD's supported specs, but by having the performance data from both ends of the spectrum we can give our customers a good idea of the rough range they'll end up in.
The Core i9 9900K was also tested with 64GB, but at 2666MHz since that is its maximum official supported speed.
|CPU||AMD Ryzen 9 3950X
AMD Ryzen 9 3900X
|CPU Cooler||Noctua NH-U12S|
|Motherboard||Gigabyte X570 AORUS ULTRA|
|RAM||4x DDR4-2666 16GB (64GB total)
4x DDR4-3200 16GB (64GB total)
|Video Card||NVIDIA GeForce RTX 2080 Ti 11GB|
|Hard Drive||Samsung 960 Pro 1TB|
|Software||Windows 10 Pro 64-bit (version 1903)
Metashape Pro 1.5.5
For benchmarking photogrammetry applications we now have four image sets that we own the rights to, covering both smaller and larger size Model and Map projects. The smaller image sets are included in our public Metashape benchmark, which you can download and run if you want to compare your system's performance to what we measured in our testing.
- Rock Model - 36 photos at 20 megapixels each
- School Map - 51 photos at 18 megapixels each
- School Model - 439 photos at 18 megapixels each
- Park Map - 792 photos at 18 megapixels each
We also split our "model" testing into two branches: one using the building of a dense point cloud to creating the model/mesh, and the other skipping that step and using just the depth maps for mesh creation. These methods utilize the hardware in a computer differently, and we wanted to reflect both approaches in our benchmark. Looking at the results, however, we might drop depth map based model processing because it takes about three times longer, which has a big impact on how many systems we can test per article. Please let us know in the comments section at the end if you have any input on this, especially if you care about that specific set of results.
Each image set was processed 2-3 times per hardware configuration, and the fastest overall result was used for the comparisons below.
Here is a gallery of charts for each of the six test configurations, showing the total time (in seconds) they took to process on each CPU.
A note about the color-coding used here: AMD processors are shown in orange with 2666MHz memory and red with 3200MHz - which is out of AMD's official supported specs at this capacity and speed combination - while all Intel processors are shown in blue.
For those who want to dig further into the differences in how each CPU performed, here is a table showing the times for each step within Metashape on each of the image sets:
The performance between these CPUs is very close in Metashape. The Core i9 9900K is effectively tied with the new Ryzen 9 3950X on smaller projects, if using high-speed memory (3200MHz, technically above AMD's spec for this 64GB configuration). When using the same 2666MHz modules, the Intel processor has a clear lead.
Moving up to the larger projects, however, puts AMD clearly in the lead for the School Model using the dense cloud workflow. For the Park Map, however, the 3950X only came in ahead if it was equipped with the faster RAM.
Based on the charts and analysis above, it is impossible to call out a single, clear winner in Metashape. AMD's new Ryzen 9 3950X does very well, especially when equipped with 3200MHz memory - but doing that either means limiting yourself to 32GB (two 16GB modules) or else exceeding AMD's official memory support by putting in more modules. The 3950X also costs more than either Intel's Core i9 9900K or AMD's Ryzen 9 3900X, and with fairly small performance differences that $200-300 price delta is worth considering.
In the end, though, any of these three processors will give you great performance in Agisoft Metashape!
All of the CPUs we tested here are better optimized for this workflow than AMD and Intel's higher-end processors, which have more cores in total but lower per-core performance. Metashape is very sensitive to that, it seems, and doesn't appear to utilize a lot of cores.
In case you missed it, you can check out our recent GPU performance article. We also make our smaller image sets publicly available to test via our Metashape benchmark, so that you can compare your own system's performance to what we have measured in our articles.
Looking for a Metashape Workstation?
Puget Systems offers a range of workstations that are tailor-made for your unique workflow. Our goal is to provide the most effective and reliable system possible so you can concentrate on your work and not worry about your computer.