RealityCapture 1.0.3: Intel Core i7 & Core i9 vs AMD Threadripper Performance
Written on May 28, 2019 by William GeorgeIntroduction
RealityCapture, like other photogrammetry applications, is built to take a batch of photographs and turn them into digital, 3D models. The many steps involved in that process can take a lot of time, and utilize both the CPU and GPU at different points. We recently put together a benchmark tool for RealityCapture, and in our first article focusing on this software we are going to take a look at how Intel's Core i7 and i9 processors stack up against AMD's Threadripper chips.
Test Hardware
Here is a list of the hardware we tested RealityCapture on. The video card and RAM capacity were kept the same across all platforms, to avoid either of those throwing off the comparison. All results included here are from after the recent Windows 10 security patch addressing MDS vulnerabilities in some Intel processors, but we did see a small increase in Core i7 9700K and i9 9900K performance when we moved to the newer Z390 motherboard - which helped offset some of the performance those chips lost because of the update.
Test Platforms | |
CPU | Intel Core i7 9700K Intel Core i9 9900K Intel Core i9 9900X Intel Core i9 9940X Intel Core i9 9980XE AMD Threadripper 2920X AMD Threadripper 2950X AMD Threadripper 2970WX AMD Threadripper 2990WX |
CPU Cooler | Noctua NH-U12S Noctua NH-U12DX i4 Corsair Hydro Series H80i v2 |
Motherboard | Gigabyte Z390 Designare Gigabyte X299 Designare EX Gigabyte X399 AORUS Xtreme |
RAM | 4x DDR4-2666 32GB (128GB total) 8x DDR4-2666 16GB (128GB total) |
Video Card | NVIDIA GeForce RTX 2080 Ti 11GB |
Hard Drive | Samsung 960 Pro 1TB |
Software | Windows 10 Pro 64-bit (version 1809) RealityCapture 1.0.3.6310 |
Benchmark Details
For testing photogrammetry applications, we 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 RealityCapture 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 - 45 photos at 20 megapixels each
- School Map - 51 photos at 18 megapixels each
- School Model - 278 photos at 18 megapixels each
- Park Map - 758 photos at 18 megapixels each
Each image set was processed 2-3 times on each CPU, and the fastest overall result was used for the comparisons below.
Results Overview
Here are charts for each of the four image sets, showing the total time (in seconds) they took to process on each CPU. The processors are listed in the same order on all charts, regardless of performance, to make it easy to see how they stack up. Intel's chips are shown in blue, with AMD's in red, and the fastest result on each project is highlighted (bold) to ensure it stands out from the crowd.
Detailed Results
For those who want to dig further into the differences in how each CPU performs, here is a table showing the times for each step within RealityCapture on each of the image sets:
Analysis
Interestingly, and unlike other photogrammetry software we have tested, Reality Capture always ran the fastest on a single processor: Intel's Core i9 9900K. Its sibling, the i7 9700K, was close in most cases - just a hair behind on the smaller projects, with the gap widening a bit on the larger ones. Presumably that is due to its slightly lower clock speed, and perhaps its lack of Hyperthreading also contributes to the increased difference with bigger image sets.
All of the higher core count processors did poorly, though, with Intel's Core X and AMD's Threadripper segmented based on processor family more than anything else. Overall, it looks like top clock speed is what matters for RealityCapture rather than the number of cores.
Conclusion
With incredibly steady performance results across all of our benchmark tests, it is easy to make a single processor recommendation: Intel's Core i9 9900K is the chip to get for RealityCapture, hands-down. It was the fastest at processing all four of our datasets.
If you are on a budget and need to save some money, especially if you work mostly with small image sets, the i7 9700K isn't far behind.
Stay tuned in the coming weeks as we take a look at GPU performance, and whether multiple video cards help in RealityCapture. Additionally, once we have all of that data, we will be adding recommended systems for RealityCapture to our website.
Looking for a
Photogrammetry Workstation?
Puget Systems offers a range workstations tailored for Pix4D, Metashape, and RealityCapture. Even in the most demanding situations, our workstations and designed to minimize downtime and allow you to work as efficiently as possible.
It's funny that this site has morphed into perhaps the most complete review site for video, photo and 3d rendering tests to be found anywhere. Obviously the goal was to increase brand awareness of the systems your company sells but whoever came up with the idea of putting together a review site dedicated to the above within the company's sales site was a genius.
If I may offer a suggestion for a test, AMD's APU's, like the 2400g and the upcoming replacements, as well as Intel's Coffee Lake and newer cpu's, have reasonably competent integrated graphics chips, how about a test showing how these processors perform when working with GPU powered video filters, for people that may be amateurs or hobbyists and can't spend thousands on a top of the line system with gobs of ram and a $800 video card.
Some may find it interesting and it may help you guys market a low end, entry level video editing system for the casual user.
Thank you for the kind words!
The video side of things is more Matt's domain, but I think he has done a little bit of work with Intel's onboard graphics in regards to video encoding. I'm sure he will see your comment here as well :)
For a lot of the stuff I work on, though, NVIDIA cards are a requirement. Both Pix4D and RealityCapture require them, though I am going to be testing Metashape on a couple of AMD graphics cards soon... and pretty much all the GPU-accelerated rendering engines (Octane, Redshift, etc) are NVIDIA-only as well. So I have a lot less opportunity to test on-chip graphics in any meaningful way :(
Good detailed comparison, there's not a lot of formal benchmark content going around for these types of applications despite how many users (with generally high end hardware) use them! If you're looking for feedback on expanding this, would be interesting to see the Zen2 AM4 CPUs tested when they're released, plus single and dual Xeon benchmarks relative to these results too?
I do plan to test the upcoming Ryzen 3rd gen processors, if we get samples of some of the higher end models. On the Xeon side, the single-chip Xeon W series should perform very closely to the Core X processors in this article - they are based on the same tech, and have very similar core count and clock speeds. Dual Xeons would just mean even more cores, but lower per-core clock speeds... and given that higher core counts didn't help with either the Core X or Threadripper lines, I don't think we'd see anything meaningful in that testing. Unless the developers of this software make big changes, the main focus of my testing going forward will be on CPU releases that improve clock speed or instructions per clock, since core count doesn't seem to be a big factor in this application.
Excellent review, eager to see how Ryzen 3000 will perform.