Ryzen XT and Core i9 10850K for Photogrammetry: Are They Worth It?Written on August 14, 2020 by William George
TL;DR: Are Ryzen XT & i9 10850K any better or worse for photogrammetry?
AMD's new Ryzen "XT" processors and Intel's new Core i9 10850K are so close in performance to the original models that there should be no perceptible difference in real-world usage. Our advice, for each pair of chips, is to get whichever is more affordable and readily available. These are not necessarily ideal processors for every photogrammetry application, though, so we recommend selecting a CPU based on which of these programs you will actually be using.
Most CPU launches include a lot of rumors, leaks, and fanfare - with reviewers rushing to get content out, system integrators updating their product lines, and enthusiasts doing everything they can to get their hands on one of the new models. The latest processors from AMD and Intel, however, lack almost all of that.
Rather than a new product generation, what we are getting this time around is a handful of very minor updates. From AMD, the new Ryzen 3800XT and 3900XT are a small step up from the existing "X" models - but as far as the specs go, the only real difference is a 100-200MHz increase in the Max Boost Clock. In theory that could boost performance by as much as 2-4%.
On the other side, Intel actually released a slightly slower model - presumably to help alleviate some of the supply issues with the Core i9 10900K. The new Core i9 10850K is essentially the same as the 10900K, only with 100MHz lower base and Maximum Turbo Boost clock speeds. This should translate to around a 2-3% drop in performance, at least theoretically.
|AMD Ryzen 7 3800X||8||3.9 GHz||4.5 GHz|
|AMD Ryzen 7 3800XT||8||3.9 GHz||4.7 GHz|
|AMD Ryzen 9 3900X||12||3.8 GHz||4.6 GHz|
|AMD Ryzen 9 3900XT||12||3.8 GHz||4.7 GHz|
|Intel Core i9 10900K||10||3.7 GHz||5.3 GHz|
|Intel Core i9 10850K||10||3.6 GHz||5.2 GHz|
As far as pricing goes, the AMD Ryzen "XT" CPUs actually have the same MSRP as the normal "X" models. The older models are often on sale, though, so at the moment the "XT" models are effectively a bit more expensive (~$50 more as of this publication). The Intel Core i9 10850K, on the other hand, comes with a slight reduction in price (~$40) compared to the 10900K.
All-told, this means that between the slightly different models from Intel and AMD, you can effectively think of it as paying $50 more for 100-200MHz higher clock speeds. Or, if you want to look at it another way: saving $50 in exchange for the slightly lower frequency.
The question is: does this tiny difference in clock speed even matter? While certainly important, the frequency (either base or Boost/Turbo) is just one factor in what determines the real-world performance of a processor. To answer that question, within the context of photogrammetry applications, we have run all of these chips through benchmarks we have created for multiple programs.
Puget Systems offers a range of powerful and reliable systems that are tailor-made for your unique workflow.
The details of our testbed systems for each processor platform are listed below:
|AMD Ryzen 3rd Gen Test Platform|
|CPU Cooler||Noctua NH-U12S|
|Motherboard||Gigabyte X570 AORUS ULTRA|
|RAM||4x DDR4-2933 16GB (64GB total)|
*All the latest drivers, OS updates, BIOS, and firmware applied as of July 20th, 2020
** RealityCapture 1.1 was released after these results had been collected, and so was not used for this testing
In order to see how each of these CPUs perform in this workflow, we ran our in-house benchmarks on the following photogrammetry applications:
- Agisoft Metashape
We ran both the standard and extended versions of each benchmark, and in the case of RealityCapture ran each one twice to ensure we had an accurate result. This is needed because sometimes our RC benchmark does not properly clock every step, and running it an extra time gives us a second result in case one or the other had such an error.
We'll start off with the results for Agisoft Metashape:
Next up, here are results in Pix4D 4.5.6:
And lastly, RealityCapture 1.0.3 (version 1.1 came out only after we had gotten our results for this article):
For those interested in a deeper dive into the data, here are tables with full results from all three benchmarks:
Analysis & Conclusion
The results above line up almost exactly with what we predicted, based on the small clock speed differences. The Ryzen 7 3800XT and Ryzen 9 3900XT came in a little bit faster (1-6%) than the older X versions in the majority of our tests, and the few times that they didn't the results were so close that they are easily within the margin of error that exists on these benchmarks.
Similarly, on Intel's side of things, the new Core i9 10850K is 0-2% slower than the 10900K - right in line with our predictions. In a couple of tests the 10850K actually came out ahead, but again the difference in those cases was so close as to be within the margin of error.
The bottom line is that AMD's new Ryzen "XT" processors and Intel's new Core i9 10850K are so close in performance to the original models that there should be no perceptible difference in real-world usage. Our advice, for each pair of chips, is to get whichever is more affordable and readily available. However, it is worth noting that these are not necessarily ideal processors for every photogrammetry application. Make sure to select your system specs based on the software you will be running and your own individual workflow.
AMD Ryzen 3800XT/3900XT vs 3800X/3900X for Photogrammetry
Given that the AMD Ryzen "XT" models only have a small 100-200MHz increase in maximum Boost frequency compared to the "X" models, we didn't expect to see much of an increase in performance... and we were right. Across all of our photogrammetry benchmarks, the 3900XT and 3800XT only performed 1-6% faster than the previous models in most tests, and in some cases were no faster at all.
If you can get these new XT versions for the same price as the older X models, then great - but if not, the fractional difference in performance is probably not worth paying more for. It is also worth noting, if you are building your own system, that the new XT models do not include a CPU cooler (not that I'd advise using the stock AMD coolers, personally, but this difference could impact those looking to build on a budget). Moreover, these are not the best-performing AMD processors when it comes to photogrammetry... so make sure to check the wider benchmark results for the specific application(s) you use when selecting your next workstation.
Intel Core i9 10850K vs Core i9 10900K for Photogrammetry
While the Intel Core i9 10850K has a $40 lower MSRP compared to the Intel Core i9 10900K, you don't give up much in the way of performance. We saw at worst 1-2% slower times, but in many cases, these processors are so close as to be indistinguishable.
Supply of the Intel Core i9 10900K processors is very limited, which is partly why Intel launched the Core i9 10850K in the first place, so if you are considering purchasing or building a new system with either of these CPUs we recommend going with whichever one you can get your hands on. You are unlikely to notice the small difference in real-world usage, so there no reason to hold out for one or the other unless the savings available from the 10850K is enough to alter what other specs you can afford. Moreover, these processors are both the top-performing Intel models for photogrammetry right now, though AMD does have faster models for some specific applications.
Puget Systems offers a range of poweful and reliable systems that are tailor-made for your unique workflow.