Puget Systems print logo

https://www.pugetsystems.com

Read this article at https://www.pugetsystems.com/guides/1136
Article Thumbnail

Photoshop CC 2018 CPU Performance: AMD Ryzen 2 vs Intel 8th Gen

Written on April 19, 2018 by Matt Bach
Share:

Introduction

AMD has just released their second generation Ryzen CPUs and while most hardware review sites focus on gaming or general system performance, today we will be looking at how they perform in Adobe Photoshop CC 2018. Not only will we be comparing them to the previous generation Ryzen CPUs, but we will also see how they hold up to similarly priced Intel 8th Gen CPUs.

In this article we will be using our internally developed Photoshop CC 2018 benchmark, which for the first time we are making available for public download. So if you are curious about how your existing PC or Mac stacks up against these CPUs, you can download and run the benchmark yourself to find out.

If you would like to skip over our test setup and benchmark result/analysis sections, feel free to jump right to the Conclusion section.

Test Setup & Methodology

Listed below are the test platforms and CPUs we will be using in our testing:

As we mentioned in the introduction, we will be using our Puget Systems Photoshop CC 2018 Benchmark which is currently available for public download. This benchmark tests a range of general tasks (open/save, resize, content aware fill, etc.), filters, as well as panorama photomerge. Full details on the benchmark are available on the benchmark download page.

Benchmark Results

While our benchmark presents an Overall Score based on the performance of each task, we also wanted to provide the individual result for each task in case there is a specific tasks someone may be interested in. Feel free to analyze the individual results below in more detail if you wish or skip to the next section for our benchmark analysis.

  Ryzen 7
1800X
Ryzen 5
2600X
Ryzen 7
2700
Ryzen 7
2700X
Core i3
8350K
Core i5
8600K
Core i7
8700
Core i7
8700K
Scores                
Overall Score 826.3 893.6 874.0 941.8 848.4 932.0 991.1 1000
General Score 82 87.5 84.5 91.2 84.4 93.1 99.9 100
Filter Score 81 88.9 87.9 95.4 83.5 92.9 99.5 100
Photomerge Score 83.1 94 92.1 97.7 88.4 94.1 96.8 100
                 
Individual Results (Seconds)                
RAW File Open 1.21 1.11 1.19 1.13 1.09 1.01 1.04 1.02
Resize to 500MB 1.43 1.39 1.37 1.36 1.38 1.37 1.33 1.46
Rotate 1.05 1.09 1.09 0.97 1.15 0.92 0.92 0.92
Magic Wand Select 17.91 16.05 17.32 15.76 16.93 15.27 13.01 12.71
Mask Refinement 5.12 4.65 5.09 4.59 3.94 3.78 3.36 3.31
Paint Bucket 2.75 2.35 2.54 2.45 2.49 2.34 1.82 1.92
Gradient 0.33 0.33 0.33 0.29 0.51 0.4 0.38 0.36
Content Aware Fill 15.27 13.91 14.41 13.34 13.45 12.09 11.29 11.07
PSD File Save 5.41 4.99 5.14 4.75 5.17 4.7 4.52 4.48
PSD File Open 3.54 3.22 3.3 3.09 3.04 2.88 2.76 2.70
Camera Raw Filter 5.99 5.68 5.78 5.39 7.06 5.6 4.95 4.89
Lens Correction 19.68 16.82 17.62 16.36 17.01 16.27 15.67 15.36
Reduce Noise 20.12 17.82 18.18 16.76 24.25 20.93 18.8 18.44
Smart Sharpen 24.06 23.17 21.87 19.94 30 22.48 20.25 19.99
Field Blur 17.91 16.48 17.43 16.1 15.51 14.58 14.2 14.24
Tilt-Shift Blur 18.07 16.42 17.33 15.9 16.49 14.94 14.1 14.18
Iris Blur 19.71 17.85 18.87 17.1 19.13 16.86 15.5 15.56
Adaptive Wide Angle 20.16 17.46 17.8 16.42 17.86 17.32 16.53 16.57
Liquify 10.28 12.43 11 10.24 9.88 9.97 9.62 9.65
Photomerge 22MP Images 100.4 85.37 87.4 82.42 87.48 83.62 80.65 78.79
Photomerge 45MP Images 122.13 111.92 113.98 107.43 123.56 114.09 111.79 107.18

Benchmark Analysis

Ryzen 2 2700X 2700 2600X Photoshop Benchmark

Download our Photoshop CC 2018 Benchmark!

Looking at the Overall Score, the first thing to notice is that the new 2nd generation Ryzen CPUs are much faster than the original Ryzen CPUs. Even the Ryzen 5 2600X is almost 8% faster than the Ryzen 7 1800X even though it's MSRP price is less than half. Speaking of the 2600X, it is also interesting to see that it is faster than the more expensive 2700 - likely due to the fact that Photoshop typically runs best with a higher CPU frequency rather than a higher core count.

Compared to the Intel 8th Gen CPUs, however, we still have to give a slight edge to Intel. At the top end, the Ryzen 7 2700X is about 4% slower than the Intel Core i7 8700 even though it is slightly more expensive. The Ryzen 5 2600X is much closer to Intel in terms of price/performance coming in at slightly slower and cheaper than the Intel Core i5 8600K. We didn't test the Intel Core i5 8600 in this article, but if we had to guess we would say that it is likely almost identical to the Ryzen 5 2600X in terms of price and performance.

Conclusion

AMD has made great improvements with the new 2nd generation Ryzen CPUs that really closes the gap between AMD and Intel for Photoshop users. In the $200-250 price range you can easily justify either brand so it is likely going to come down to which platform offers you the features you need and any sales you may be able to find. However, at the higher end the Intel Core i7 8700 edges out over the Ryzen 7 2700X (especially in general tasks) while also retailing for a slightly lower cost. And if you are looking for the best possible performance, the Intel Core i7 8700K is still the overall king and our current go-to recommendation for Photoshop.

Tags: Photoshop, Ryzen, Coffee Lake, 8th Gen, i7 8700K, i5 8600K, i3 8350K, 2700X, 2700, 2600X
Jakub Badełek

Hi Matt, thanks for quick tests! are you planning to have a look at the performance in Lightroom as well?

Posted on 2018-04-19 13:46:09

Hey Jakub. Yes, but it may be a bit. I'm in her middle of revamping our Lightroom testing and really want to get that done first. I expect it will end up similar to this test only with the 2700 and 2700X performing a bit better. Probably right on part with the 8700 on average.

Posted on 2018-04-19 15:32:57
Al Bacione

Ryzen gets benefits from high speeds rams. This comparisons need to be done at 3000/3200Mhz at least. Anyway still interesting, but lacks the mainstream scenario (over 3000Mhz RAM). It is also very important to consider the cpu cooling solution, as the Ryzen, when cooling with a high performance unit, can automatically reach the maximum frequency thanks to XFR "auto-overclock". Intels average OC could be also interesting to conclude the testing comparison. Regards.

Posted on 2018-10-25 08:21:27
Cbed

Really great review. I think I'm going with the 2700x due to the ability to upgrade to future generations on the AM4 Platform.
Trading off a little bit of single threaded performance isn't such a big deal. I do some Multi-thread heavy workloads that make it worth it over the 8700k.
Would be neat to see if running 2700x on faster memory offers some performance improvement since it's rated for it this time around.

Posted on 2018-04-19 18:13:10
Brandon

Was the 8700K system patched for meltdown spectre?

Posted on 2018-04-20 16:09:00

Yea, we're running the latest BIOS and firmware that includes those patches.

Posted on 2018-04-20 16:10:10
bradley parsons

security flaws would not affect how a cpu benchmarks.

https://www.pcworld.com/art...

Posted on 2018-08-15 00:23:55
Mockingbird

You have to take into account that AMD gives you a nice stock cooler like the Wraith Prism right in the box while Intel gives you a box full of air.

Posted on 2018-04-20 16:22:52
Konaha ANBU

I believe this is unnecessary to discuss that considering this as a performance analysis exclusive of discussion about stock cooler although your point is valid.

Posted on 2018-04-20 16:35:48
Mockingbird

Well, he is talking about prices and it would be appropriate to consider the extra cost of cooler necessary for Intel models.

Posted on 2018-04-20 16:58:18

recommending an intel processor for photoshop because this difference in performance is the most stupid thing i have seen. Power consumption, platform price and overall performance like in blender and other applications. When you are using photoshop you might be using other apps in background.

AMD 2700x is the new king.

Posted on 2018-04-20 16:57:50
Mark Harris

sure it is...oh amd fans..so fun

Posted on 2018-10-01 18:06:26
Ned Flanders

Why did you test the Ryzen 2 Sys with 2666MHz DDR4 RAM? It is specified for 2933MHz RAM. Guess that would close the gap.

Posted on 2018-04-20 19:11:25

2933mhz is the maximum supported frequency and is only officially supported if you are using 1 DIMM per channel (so two sticks). We are trying to get confirmation from AMD what the specific speed that us supported by various RAM configurations but so far that is the only official word we have right now. On the original Ryzen CPUs it depended not only on how many sticks you used but also whether they were dual or single rank. So until we get word, we decided to stick with the same speed we use on our Intel based systems.

If I had to guess, I think upgrading to 2933mhz would net another few percent, but not much more than that. DDR4-2666 is also where the sweet spot for maximum stability appears to be at the moment, however, so a few percent may not be worth it even if the CPU and RAM configuration supports it.

Posted on 2018-04-20 19:18:18
Nick Vu

That is incorrect. You have support in X470 chipset for 4x8GB at 3200CL14 and 4x16GB at 3000CL15. (Out of the box support for Asus's x470-Pro)
And depending on the timings and sub-timings it can give 2digit increase in performance considerings few-core workloads benefit the most from infinity fabric speeds.

Posted on 2018-06-05 12:03:24

We just got word back from our contacts at AMD about the official RAM speed support for the 2nd Gen Ryzen CPUs. We couldn't find this information anywhere else, so we put up a quick post with the information: https://www.pugetsystems.co... . Based on what they sent us, in the four stick configuration we tested DDR4-2133 is actually the fastest RAM speed officially supported by the 2nd Gen Ryzen CPUs. That means we are technically overclocking with DDR4-2666, but given how sensitive Ryzen CPUs are to RAM speed I think DDR4-2666 is a very reasonable speed to use. It turns out DDR4-2933 is only recommended if using two sticks of single rank memory on a motherboard with at least 6 PCB layers.

Posted on 2018-06-06 22:38:05

Thanks for sharing the benchmark! For anyone else comparing systems, I have the previous 7th generation Core i7 7700K overclocked to 5Ghz (4.7Ghz for AVX code like Photoshop), 64GB 3200MHz RAM, 1080Ti, and 1TB 960 Pro SSD:

Overall Score 947 Score
General Score 85.9 Score
Filter Score 100.7 Score
Photomerge Score 100.3 Score
GPU Score 88.1 Score
RAW File Open 2 Seconds
Resize to 500MB 2.84 Seconds
Rotate 1.1 Seconds
Magic Wand Select 13.41 Seconds
Mask Refinement 3.45 Seconds
Paint Bucket 2.01 Seconds
Gradient 0.45 Seconds
Content Aware Fill 11 Seconds
PSD File Save 4.46 Seconds
PSD File Open 2.56 Seconds
Camera Raw Filter 5.66 Seconds
Lens Correction 14.31 Seconds
Reduce Noise 20.26 Seconds
Smart Sharpen 25.03 Seconds
Field Blur 14.06 Seconds
Tilt-Shift Blur 13.43 Seconds
Iris Blur 15.18 Seconds
Adaptive Wide Angle 15.69 Seconds
Liquify 7.62 Seconds
Photomerge 22MP Images 77.88 Seconds
Photomerge 45MP Images 107.74 Seconds

Posted on 2018-04-21 21:07:04
Eastman

Hi Matt, thank you to Puget for doing these tests on photoshop. Though useful, the set of tests is mostly automated so they can be easily reproduced between different system setups. It's only half the story for photoshop performance and doesn't show the whole picture. I come from a concept/matte artist background and work in photoshop 90% of the time. My saved files can reach over 2GB in PSB format. The benchmarks I would like to see is how interactive and responsive the software is with user control. For example when you're airbrushing, is there any lag in the brushmarks? After converting a layer into a smart object, I tend to do warping and puppet warp so I can come back and tweak them. Is it smooth or it takes a moment between moving the warp handles around? What about transform operations on a layer (Ctrl-T command) do the corners move smoothly when pulling with a mouse or tablet? How about spot healing brush, finger painting and blending colors, etc
Another feature I use a lot is just panning around by pressing spacebar. I've noticed the refresh rate got worse and worse with each PS CC updates. Is that CPU or GPU dependent?

For me it's all about direct interaction. I don't want any lags in the performance. I don't really reduce noise or blur photos on a daily basis. A good benchmark test I do on systems for responsiveness is to create a 30,000 x 30,000 pixels canvas use a 3000 px airbrush with a reduce spacing of 5%. Then make brush stroke on that canvas and observe the lag. It's just to push the system to the limit. I know these tests are harder to replicate exactly between different systems but they are more indicative of photoshop performance than an image resize will ever be. What GPU/CPUs/Memory combo would reduce that lag. Maybe you could come up with a set of tests like that.

Posted on 2018-04-25 15:07:17

I agree 100% that things like transform, brushes, panning, and things like that would be great to test. Unfortunately, so far I have found it to be pretty much impossible to do in an accurate and repeatable manner. We actually get very similar requests for our Lightroom testing, but so far I haven't had much success developing a way to benchmark that kind of thing. Every few months I get a crazy idea that I think may work, but so far none of them have panned out.

Definitely something I will keep trying to build into our benchmarks, but no promises on when I'll finally get an idea that actually works.

Posted on 2018-04-25 18:56:33
Eastman

Great that you're aware of these areas. Maybe user responsiveness is a factor that could be extrapolated from other sets of synthetic benchmark. But when it comes down to it, a user has to sit down and work with the software to get a 'feel' of it. That's what I'm so curious about. I'm tempted to setup a ryzen 2 and an 8700k system just to test these factors with everything else being equal.

One piece of tidbits that users might find useful is that Photoshop CS6 is much more responsive and smooth than later versions of CC. Panning around in CC jitters more than in CS6. So if you're struggling with large files responsiveness, just switch to CS6. Make sure you don't have any features in your work files missing from CS6 for compatibility.

Again, no way of knowing if it's CPU/GPU or Ram speed...
Could graphic refresh performance be linked to gaming performance?
In this case, Ryzen has less dropped frames than 8th gen intel then again Intel has a higher max frame rate.

Posted on 2018-04-26 11:50:06
a d00d

Something I haven't seen you mention that is very important is that many apps are compiled for Intel-specific optimizations and perform poorly on AMD for that reason. Have you thought of a different, more open-source benchmark that could be compiled for both sets of optimizations so we can get a true oranges--oranges comparison?

Posted on 2018-04-27 09:28:54

We actually focus almost exclusively on testing very specific programs -mostly in the content creation fields. That is the area we sell most of our workstations for and the purpose of our testing is to ensure that our customers are getting the exact right hardware for their needs. We care much more about making sure performance will be as good as possible in those specific programs rather than in a synthetic benchmark.

For more general performance, review sites like Anandtech, Tom's Hardware, and many others do an excellent job. That's where we go ourselves if we want to get a more general take on performance. Honestly, even if we had a reason to do that type of testing it would be hard to justify the extra cost and effort simply due to the fact that those site do such a great job already.

Posted on 2018-04-27 16:24:14
Det Jan

Hi,what was your setting for this problem? Did you take care of the standard boost feature in the bios? Even these guys run into it: https://www.anandtech.com/s...

Posted on 2018-04-30 20:42:10

As far as I understand, they only had an issue because they were forcing HPET to be used by the OS. That is not the default behavior and isn't something we do, so it shouldn't affect our tests (or most others).

Posted on 2018-04-30 21:12:46
Det Jan

Yes, did not read all about it. Just run into some other test where they did not turn on the boost option, could not find it, guess it was a german site. But you did turn it on? Bye the way - great testing you are doing!

Posted on 2018-05-03 08:59:28
Gary H

Did you review/compare/recommend systems that will be running both Lightroom CC 2018 and Photoshop? Can you post link? Thanks. I usually upgrade entire system every 4 years and I don't want to pay for the bleeding edge. Something good, solid, and fast ... but not screaming. I process mostly in LR and Edit using Photoshop, Perfect Effects Pro, and Color EFX Pro.

Posted on 2018-05-03 19:30:14

Mixing software packages is something we tend to recommend doing one-on-one with our sales consultants. There are just so many different ways to mix and match, people have differing priorities, budgets affect choices, etc. The number of variables are simply too much to be able to make a one size fits all. You can give us a call at (425) 458-0273 or send an email to sales@pugetsystems.com

Posted on 2018-05-03 19:41:13
Srijan Roy Chudhury

Hello, thank you so much for application specific benchmarking. We really need it. I have a question and would really be grateful if you would help. I am planning to upgrade my rig but I am confused whether to go for R5-2600 or i7-8770. I am also thinking of installing one RAM module of 16GB for now. What is your opinion about these?
Thanks again.

Posted on 2018-05-18 15:32:31
Chris Miller

I thought you guys were going to include the new imac pro in this year's benchmark? Any plans for a comparison still? Thanks,

Posted on 2018-05-18 21:31:44

How many images are merged in the photomerge test?

Posted on 2018-08-08 20:51:08

Full information on our benchmark is at https://www.pugetsystems.co.... Both the Photomerge tests use 6 images.

Posted on 2018-08-08 23:04:01
bradley parsons

Although the benchmarks are nice, they are also biased. Most benchmarks for the past six years or so on games and software have not fully designed to work on AMD's processors. AMD processors were not great from Bulldozer series and forward...until Ryzen has released.

That is why all games and software are slowly coming to support them fully. (AMD)

Photoshop CC 2018 was released on 18 October 2017. These processors were released in First quarter of 2018. Photoshop is likely still set to run on Intel processors better than AMD's

Posted on 2018-08-14 23:08:36
Mark Harris

So how does that make the benchmarks BIASED??!?!?!!!! They are testing for photoshop with latest versions and current CPUs to show REAL LIFE performance that a user will get right now. How in the world is that biased? Is not like they are using a very specific version of the program where AMD performs worst.
You AMD fanboys always have something to whine about.

Posted on 2018-09-30 18:27:04

Mark, I appreciate your contributions, but we expect respectful discourse here. I deleted your other comment that insulted and called someone names.

Posted on 2018-09-30 18:31:44
Mark Harris

I cant deal with fanboys.

Posted on 2018-09-30 19:58:24
Ray Gralak

bradley parsons
As Matt pointed out before, the most of customers they sell to are looking for workstations for content creation, thus the relative performance of applications like Photoshop is important. When AMD engineers designed and enhanced their processors no one kept AMD from testing all those "Intel optimized applications" (i.e all applications?) you claim caused biased results. And the patches that improve AMD performance on some applications (usually games) usually revolve around using multiple cores more effectively... but the authors of those patches often fail to mention that many of those same patches would also help Intel performance as well. What it comes down to is this... because of Intel's higher per-core performance AMD CPUs need to use more cores to match Intel's performance. At least that is the way it is today.

Posted on 2018-10-16 02:52:13
Владимир Рудаков

you did a great job, no doubt... But any option to test amd 2xxx series with normal memory speed as 2933 and show us the timings? That would be honest to see no diff in speed at all I guess or vice versa amd would be a bit ahead. Thank you anyway

Posted on 2018-09-04 18:16:47