Puget Systems print logo

https://www.pugetsystems.com

Read this article at https://www.pugetsystems.com/guides/1295
Article Thumbnail

Photoshop CC 2019 CPU Roundup: Intel vs AMD vs Mac

Written on December 5, 2018 by Matt Bach
Share:

Introduction

If you are in the market for a new Photoshop workstation, you might be a bit overwhelmed by the hundreds or thousands of different hardware components you can choose from. Even if you are just trying to decide on a CPU, it can still be daunting to find one that not only fits your budget, but will also give you the best performance for your dollar.

In this article, we are going to be benchmarking a wide range of processors from Intel and AMD including the Intel 9th Gen, Intel X-series, AMD Ryzen 2nd Gen, and AMD Threadripper 2nd Gen CPU lineups. In addition, we will be comparing them to a current Mac Pro 12 Core and iMac Pro 14 Core for those that are considering going with a Mac-based workstation rather than a PC.

One thing to note is that we will not be including results for any previous-gen CPUs in this article. At first, we were going to include them but the charts and tables soon got out of hand. Instead, if you want to know how these CPUs compare to previous generations, we recommend checking out the following articles:

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 systems we will be using in our testing:

Shared PC Hardware/Software
Video Card: NVIDIA GeForce RTX 2080 Ti 11GB
Hard Drive: Samsung 960 Pro 1TB M.2 PCI-E x4 NVMe SSD
OS: Windows 10 Pro 64-bit (version 1803)
Mac-based PC Test Hardware  
System: Apple Mac Pro (12 Core) Apple iMac Pro (14 Core)
CPU: 12-core
2.7GHz 30MB of L3 cache
14-core Intel Xeon W
2.5GHz Turbo Boost up to 4.3GHz
RAM: 64GB 1866MHz DDR3 ECC 64GB 2666MHz DDR4 ECC
Video Card: Dual AMD FirePro D700, 6GB of GDDR5 VRAM Radeon Pro Vega 64, 16GB of HBM2 memory
Hard Drive: 1TB PCIe-based SSD 1TB SSD
OS: MacOS Mojave (10.14.1)

For the testing itself, we will be using Photoshop CC 2019 (ver. 20.0.0) along with version 18.10 of our Puget Systems Photoshop CC 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 various scores based on the performance of each type of task, we also wanted to provide the individual results in case there is a specific task someone may be interested in. Feel free to skip to the next section for our analysis of these results.

Benchmark Analysis

Photoshop CC 2019 Benchmark CPU Roundup - Intel 9th Gen, Intel X-series, AMD Threadripper 2nd Gen, AMD Ryzen 2nd Gen, Apple Mac Pro, Apple iMac Pro

Our Photoshop benchmark (which is available for public download) tests a large number of tasks that we split into 4 categories: General, Filter, Photomerge, as well as GPU-accelerated which contains results from both the general and filter tests. With the exception of Photomerge, we also run each test in both 8 bits/channel and 16 bits/channel mode. These results all combine into a single "Overall Score" that represents the overall performance of the system in Photoshop.

If we sort by the overall score, you can see that the Intel 9th Gen CPUs like the Core i9 9900K and Core i7 9700K are the top performers - which is to be expected since Photoshop is not terribly effective at leveraging higher core count CPUs. Next is the iMac Pro 14 Core, followed by the whole slew of Intel X-series processors that were recently released. These X-series CPUs are around 10-15% slower than the Intel Core i9 9900K depending on which model you choose.

Below the Intel X-series are the AMD Ryzen 7 and the AMD Threadripper CPUs. Interestingly enough, while the lower-end Threadripper CPUs are only about 20% slower than the i9 9900K, the two "WX" series models are closer to 27% slower. These models are essentially two CPUs crammed onto a single die which results in additional overhead that can cause performance loss in applications that cannot take advantage of higher core counts. Even with the new "Dynamic Local Mode" feature enabled (which is supposed to assist in this sort of workload) the WX models clearly have some performance issues.

At the very bottom of our list is the old (but somehow still current) Mac Pro. With the Mac Pro, we are about 30% slower than the i9 9900K.

Intel vs AMD vs Mac for Photoshop CC 2019

For Photoshop CC 2019, there is no question that the Intel 9th Gen CPUs are currently the fastest processors available. Even if you were to spend 50% more on a workstation based around an Intel X-series CPU or almost twice as much for an iMac Pro, a Core i9 9900K will still be 7-15% faster.

Compared to an AMD Threadripper system, the performance gap is even larger. Here, you can expect a Core i9 9900K to be anywhere from 20% to almost 30% faster in Photoshop. The AMD Ryzen 7 2700X, on the other hand, does about right for its price. It is a bit more expensive than the Intel Core i5 9600K, but it is also faster by a few percent. This means that if your budget can only accommodate a ~$300 CPU, you likely won't see much of a difference between Intel and AMD for Photoshop.

If you are curious how your current system compares to any of these processors, we recommend downloading and running our Photoshop CC Benchmark which is the exact benchmark used in our hardware articles. And if you are interested in how the latest Intel and AMD processors perform in other applications, be sure to check out our recent Processor articles as we have a number of articles looking at CPU performance in Lightroom, After Effects, Premiere Pro, DaVinci Resolve, and many other software packages.

Tags: Photoshop, Intel 9th Gen, Intel X-series, AMD Threadripper 2nd Gen, AMD Ryzen 2nd Gen, Apple Mac Pro, Apple iMac Pro, 9900K, 9700K, 9600K, 9980XE, 9960X, 9940X, 9920X, 9900X, 9980X, 9800X, 2990WX, 2970WX, 2950X, 2920X, 2700X
ComputahNerd

I notice that both in this test and the AE test the 9960x does a quite better job than the 9940x. Is this because it has a single core speed boost that's the same as 9940x. Meaning that any single thread performance (responsiveness etc) is roughly the same but utilize the extra two cores? Based on the chart, that shows each test, it seems that the 9960x does a better job in general, even on things like rotate, which is weird, as I thought those aren't multithreaded tasks. Could it be that the 9960x just basically is a better tuned CPU, since they have the same clock boost, same wattage drain despite 9960x having two more cores?

last year 7940x scored better than 9760x - So this time around it seems adding the extra two cores doesn't sacrifice performance, in fact it only seem to benefit from it? Since the 9980xe does even better with 18 cores!

Posted on 2018-12-06 07:37:03
ComputahNerd

Hmm, I think I misread the chart, I thought the numbers were seconds on each of the benchmarkings except overall score. But looking at 2950x and Mac Pro, it seems its all scores. So then as assumed, the 9940x and 9960x does pretty much the same in Single threaded tasks, but 9960x gains some extra scores because of the extra 2 cores.

Posted on 2018-12-06 08:09:00

To be honest, the difference in this test and the Ae tests was only about 3% on average which I would really consider to be within the margin of error for these tests. So I think in most situations, you really are not going to notice much of a difference between the two processors. From the i9 9900X on up, all the new X-series CPUs have a max Turbo Boost 3.0 speed of 4.5GHz - which is what the CPU should run at with up to two cores being loaded. Since most things in Ae and Ps are only going to use a handful of cores, it makes sense that all those CPus are going to perform within a few percent of each other.

Posted on 2018-12-10 17:46:25
ComputahNerd

Yeah but this is in fact quite good news for those of us that needs a lot of cores, I think. Because, like myself, I got a threadripper 1950x, and was shocked by the performance drop in Adobe programs. And the same problem seemed to occur on the 79--x series as well, that the more cores you added the worse it would perform in Adobe programs (though not as severe as the AMD HEDT CPUs) But this year, it seems the turboboost is the same across all CPUs, so if one need more cores, one can choose pretty much the amount of cores on the Intel X series as one would like without loosing any performance in AE or Photoshop.

Posted on 2018-12-10 18:18:59

The Turbo Boost 3.0 change is definitely the biggest advantage to these new X-series CPUs. Performance-wise they are only moderately faster than the previous generation, but it is really nice to not have to decide between higher single-core performance or higher multi-core performance. You really just go higher up the stack if you need higher multi-core and don't worry about single-core performance. Really, really nice especially for system integrators like us.

Posted on 2018-12-10 18:26:55
ComputahNerd

Yes, exactly what I would wish for :) I don't care what people say about AMD giving more cores for the money. When you work on these machines every day, this is what it all comes down to, performance on both single and multiple cores.
When you are testing this, are you setting up the programs as priority through the turboboost 3.0 program, or does the machine manage this automatically ?

Posted on 2018-12-10 19:12:54

Turbo Boost 3.0 priority is BIOS level these days actually (unless you have it in "legacy" mode in the BIOS). So core prioritization is all done automatically.

Posted on 2018-12-10 20:14:53
ComputahNerd

Oh so you do not need the annoying Turboboost software at all then?

Posted on 2018-12-10 20:36:16

Nope! Not as long as your motherboard supports the newer "Native" mode which I would be surprised if any X299 boards don't at this point.

Posted on 2018-12-10 23:19:07
Alan Gordon

Fascinating that the 14 core imac pro beats out the 14 core i9 windows. Makes you wonder what Apple computers could do if they kept their chips up to date...

Posted on 2018-12-08 04:59:39
Dragon

You are comparing an x processor to a Xeon. You would have to see the test with an equivalent Xeon to know if the Mac was really faster. There are subtle differences between x processors and Xeons that sometimes make a big difference depending on the software. Not defending any OS here, just noting the difference between processors.

Posted on 2018-12-08 18:53:28

I don't think it is actually Xeon vs Core that is making the difference since these days, the two lines are really the same in terms of performance. Compared to the Core line, the main advantage with a Xeon CPU is that they have support for Registered ECC memory (allowing for up to 256GB configurations), have slightly higher Turbo Boost clocks (although they don't have Turbo Boost 3.0 so in the end it doesn't tend to make much difference), and they are supposedly binned better so they should be a bit more reliable (although both fail so rarely is is hard to tell if this is true).

I think the main reason the 14 core iMac Pro performs better than the i9 9940X on Windows is simply that MacOS is a slimmer and more efficient operating system. Of course, you are paying a hefty premium for that OS (unless you want to get into a Hackintosh setup), so with a PC you can get an even faster CPU to overcome the OS differences. Or faster storage, more RAM, multiple displays, etc. while still coming in at a lower price point than the iMac Pro.

Posted on 2018-12-10 18:09:06
Panacea

Edit: error is mine.

Posted on 2018-12-11 18:41:44
VP

Matt, an artist I work with has some VERY large Photoshop projects. A recent project is 145GB with 890 layers and 255 linked Smart Objects. Her PC is laggy when working on these large files.

Her PC is a 3 year old i7-5960X overclocked to 4.3Ghz with 64GB of RAM. Her system drive is a Samsung M.2 (running Win7 Pro) and her working drive is a 1TB Samsung 960 EVO. Her graphics card is a Quadro K4200. She uses a Quadro instead of a Titan because she has a 10 bit monitor. She runs the latest Photoshop CC version.

In thinking about a new hardware spec for her, I understand that PS does not make efficient use of extra cores, and hence clock speed is more important than core count (for PS). I'm thinking she should get an i9-9900k (for the fastest clock spped per core). But honestly, when I watch over her shoulder, with the Performance tab in Windows Task Manager visible, it's rare that 50% of the CPU is being used. One or two cores might be at 65% - 75%, with the others at 10% - 40%. Yet when she is drawing or masking, the screen is always lagging behind her hand by a few seconds or more. This lack of real time visual feedback makes her work difficult. So my questions are:

1. Do you think 128GB of RAM would help her situation? I've read that The new 9th Gen Intel Core processors memory controller is capable of supporting DDR4 16Gb die density DIMMs which will allow the processors to support a total system memory capacity of up to 128GB when populating both motherboard memory channels with 2 DIMMs per Channel (2DPC) using these DIMMs. I am hoping that there will soon be mobos that support 128GB of ram with the i9-9900k... thereby giving her the fastest clock speed available AND 128GB of ram... if you think more than 64GB would be helpful for her.

2. Do you think a substantially faster GPU with substantially more memory would speed things up for her? ie: if she could live with 8bit instead of 10bit graphics, would a Titan V substantially speed up the visual lag on her machine when working on these huge Photoshop projects?

I'd greatly appreciate your thoughts on this config. All the tests and benchmarks I've found on the web are dealing with much smaller / less complex files. It's hard to find testing info addressing mouse to screen lag on massive files. Many thanks!

Posted on 2019-01-23 01:12:28

You have hit on the issue with large Photoshop files - this is also an issue with applications like After Effects. You want the best per-clock performance (9900K like you said), but also need a ton of RAM. Going through your specific questions:

1) Probably more RAM will help. If you can, I would watch the RAM usage numbers when she opens one of her projects and see how quickly it fills the RAM. If it is immediate, then 128GB of RAM will help, but she very well may benefit from actually having 256GB of RAM. That involves using a Xeon-W, however, which (combined with Registered RAM) is going to get really pricey.

I wouldn't count on the upcoming 32GB DDR4 sticks that are supposed to work on Z390/9900K. They have been talked about for a while now, but I'm personally not convinced that they will actually hit retail - or at least not at a reasonable cost. So if she needs 128GB of RAM, going up to a X299 motherboard with something like a 9900X/9920X is probably the best bet.

2) GPU probably won't make a difference - https://www.pugetsystems.co... . Especially since she is likely very CPU bottlenecked, there likely isn't a need for anything too high end. However, her current K4200 is pretty old, probably equivalent to a modern Quadro P600. If you have the budget, going up to a P4000 would likely help a little bit simply due to the higher VRAM, but I think making the CPU change will be a much bigger impact than upgrading the GPU.

Posted on 2019-01-23 17:41:55

Little update, looks like Anandtech just did a review of the 32GB DDR4 sticks: https://www.anandtech.com/s... . So it actually does exist, and pricing looks very reasonable (same as 2x16GB sticks). No idea when they will actually be available in retail, but at least it looks less like vaporware.

Posted on 2019-01-23 18:19:27
VP

Thank you for your insights Matt... and for the GPU link. I see what you are saying in regards to the GPU. It looks like an 8GB P4000 should get her in the GPU ballpark for PS, while keeping a 10bit display path without breaking the bank.

Yes, the amount of RAM she has allocated to PS (75% of her 64GB) immediately gets used up when she opens one of her large images.

When I watch the Performance tab in Windows Task Manager, it's rare that 50% of the CPU is being used. One or two cores might be at 65% - 75%, with the others at 10% - 40% (usually 10% - 20% for all cores except the main 1 or 2). Is there a better way to gauge CPU use? ie: I'm trying to observe her machine's use in relation to your comment about being likely that she is very CPU bottlenecked

The big question in my mind (for her specific use case) is the RAM vs clock speed trade-off since 256GB of RAM would mean slower clock speeds (ie: machines that can use 256GB of RAM do not have processors available with the clock speed of the 9900k).

If she was willing to go for a Xeon-W and 256GB of RAM, she would not have as fast of a clock speed as a 9900K, but Photoshop could load her whole image into RAM.

Do you think a Xeon W2125 with 256GB of RAM would do more to eliminate the lag (when drawing or masking layers on a 145GB image) than a 9900K with 128GB of RAM (assuming that does become possible)?

Thanks for the anandtech link about the new RAM. For more hope of progress on that front, there's also this article from a couple days ago on anandtech that looks promising (re: 128GB on Z390 boards) as well:
https://www.anandtech.com/s...
It says Asus has validated SL Link's J4BGUS2G8QHBC DDR4-2667 CL19 32GB modules for their Z390 boards.

And on this Asus page, if you click Show All, there is a beta bios for "Supported JEDEC standard DDR4 32GB memory" for the ROG STRIX Z390-F GAMING mobo (BIOS 0803):
https://www.asus.com/Mother...

Posted on 2019-01-25 06:28:38

It is so hard to say how much performance she would get out of having more RAM. My gut feeling is that it will be more than the ~10% performance loss you will get by moving from a Xeon-W. Even if you have enough RAM, however, that much raw data that Photoshop is working with all has to go through the CPU, so I don't think there is any solution that exists today that would give a completely lag free workflow for her projects. Given the crazy cost you would be paying for a Xeon-W with 256GB of RAM, I would lean a bit more towards a i9-9900K with 128GB of RAM (once those sticks hit retail) and invest in a really nice NVMe drive for the disk cache. Either a Samsung 970 Pro or possibly even an Intel Optance 900p. Optane is more expensive, but it has better random performance and much higher endurance which can be great for this type of caching.

Posted on 2019-01-28 17:47:03
VP

Many thanks for your input Matt. That makes sense. I'll tell her to wait until the RAM is available for the 9900k. Hopefully it won't be too much longer...

I did not know that Optane has better random performance and endurance than the Samsung 970 Pro. That's great to know! Her current machine has a Samsung 256GB XP941 M.2 for disk cache. Looks like the Optane random performance is more than 4 times faster!

Posted on 2019-01-29 05:04:48