How much faster is a modern workstation for Adobe Photoshop CC 2017?

By Jos ยท 18 replies
Mar 16, 2017
Post New Reply
  1. Editor’s Note:
    Matt Bach is the head of Puget Labs and has been part of Puget Systems, a boutique builder of gaming and workstation PCs, since the early days. This article was originally published on the Puget blog.

    There is no doubt that a modern workstation will be faster than one that is more than a year or two old, but the advancement of computer technology has certainly slowed in recent years. 10 years ago, it was not uncommon to see massive performance gains generation to generation, but today we often see much smaller 10-15% performance gains whenever a new product is launched. Some areas (such as storage) are still showing massive raw performance gains, but when you benchmark the new products in a real-world scenario even these are not quite as impressive as what we used to see.

    These 10-15% gains per generation are not something most people will get excited about and make it difficult to determine how much of a performance gain you might see if you were to replace your 2, 3, 4, or 5 year old workstation with a new one. To try to answer this question as objectively as we can, we decided to directly compare five workstations using the highest end components over the past 6 years.

    Test Setup

    To benchmark the performance difference between different generations of workstations, we tested Photoshop with the following hardware configurations:

    Available ~January 2017
    • Intel Core i7 7700K (4 Core 4.2GHz)
    • 4x 16GB DDR4-2400 (64GB Total)
    • Nvidia Titan X (Pascal) 12GB
    • Samsung 960 Pro 1TB NVMe SSD
    Available ~August 2015
    • Intel Core i7 6700K (4 Core 4.0GHz)
    • 4x 16GB DDR4-2133 (64GB Total)
    • Nvidia GeForce GTX Titan X 12GB
    • Samsung 850 Pro 1TB SATA SSD
    Available ~June 2013
    • Intel Core i7 4770K (4 Core 3.5GHz)
    • 4x 8GB DDR3-1600 (32GB Total)
    • Nvidia GeForce GTX Titan 6GB
    • Intel 335 240GB SATA SSD
    Available ~April 2012
    • Intel Core i5 3570K (4 Core 3.4GHz)
    • 4x 8GB DDR3-1600 (32GB Total)
    • Nvidia GeForce GTX 580 1.5GB
    • Intel 335 240GB SATA SSD
    Available ~January 2011
    • Intel Core i7 2600K (4 Core 3.4GHz)
    • 4x 8GB DDR3-1333 (32GB Total)
    • Nvidia GeForce GTX 580 1.5GB
    • Intel X25-M Gen 2 160GB SATA SSD

    These configurations were designed first around the various Intel CPU generations (using the fastest CPU available at launch), followed by the fastest GPU that was available at that time. After that, we simply used the highest capacity of RAM supported by the platform along with the SSD model we most commonly sold at that time on Puget machines. On the software side, we will be using Windows 10 Professional and the latest version of Photoshop 2017 on each system to ensure that any performance differences we see are strictly hardware level rather than being influenced by different software versions.

    Unfortunately, since we don't always keep old hardware around once it is obsolete, we had to dig through our warehouse, closets and even home PCs to put together these systems. This means that some configurations are slightly off from what they ideally should be (such as using a GTX 580 instead of a GTX 680 and only a i5 3570K for the April 2012 configuration) but despite these small discrepancies our testing should fairly accurately show how a modern workstation compares to one that is 2, 4, 5, and 6 years old.

    Photoshop Action

    To start our testing, we looked at over 20 different actions including both general task like opening and saving files as well as a variety of filters (blurs, sharpen, lens correction, liquify, etc.) and a few other miscellaneous actions. Because of the high number of different tests, we are going to look at the results as an overall average across all of the tested actions.

    Looking at the average from the 20+ actions we tested, we get a pretty good idea of how much faster a modern workstation is for general Photoshop usage. If you go back just one generation, a system from the middle of 2015 is only going to be about 10% slower than a new system which is not terribly exciting. However, a workstation from 2013 is going to be around 25% slower than a modern workstation which means that if it takes you 10 seconds to open a Photoshop file or apply an effect on a 2013 workstation it should only take around 7.5 seconds on a modern workstation. This isn't anything dramatic, but it will certainly add up very quickly throughout a workday.

    If you go back even further to a workstation from 2011, that machine should be just a bit more than half the performance of a modern workstation. Again, if it takes you 10 seconds to complete an action in Photoshop on a workstation from 2011, a modern system should be able to complete that in about 6 seconds. This is a huge difference in performance!

    Photomerge Results

    To see how well a modern workstation performs when creating a panorama using photomerge, we tested with a set of both 18MP and 80MP images:

    Going back just one CPU generation to 2015, there is only about a 10% difference between a machine from that time and a modern 2017 workstation. However, if you go back to 2013, a workstation from that time is only 64-76% as fast as a modern workstation. If you look at the raw results, you will see that this means a photomerge of 6x 80MP images that takes 6.5 minutes on a 2013 workstation takes just 4 minutes on a new 2017 workstation.

    Going back even further, a machine from 2012 is only about 60% as fast as a modern workstation and one from 2011 is only 43-61% as fast. Again looking at the raw results, where a modern system takes just 4 minutes to do a Photomerge of 80MP images, a machine from 2011 will take a bit more than 9 minutes.

    Merge to HDR Results

    Just like photomerge, we again used a set of both 18MP and 80MP images to see exactly how much faster a modern workstation is:

    Overall, the results are not that much different than the results from the previous two sections. There is only a small difference in performance going back to 2015, but the difference gets larger and larger the further back you go. Especially if you have a workstation from 2013 or older, you can expect pretty large performance gains by upgrading to a modern workstation.


    Averaging the results from the three sections, you can get a great idea of how much faster a modern workstation should be on average compared to ones from previous years. If you would rather think in terms of a new workstation being X times faster than an older system, we also have this data in a slightly different format that you may prefer:

    How much faster is a new 2017 Photoshop workstation?

    • Compared to January 2011 system: 1.8 times faster
    • Compared to April 2012 system: 1.5 times faster
    • Compared to June 2013 system: 1.35 times faster
    • Compared to August 2015 system: 1.12 times faster

    If you have a fairly new workstation that is only a few years old, it probably isn't necessary to upgrade unless your workflow requires the best possible performance. If you have a machine from 2013 or older, however, you could be looking at some very significant performance gains. You might not see a doubling in overall performance even if you have a workstation from 2011 but it will be pretty close and should significant improve your productivity.

    Keep in mind that we are looking strictly at high-end workstation from each time period. If you have an off-the-shelf workstation, you would actually probably see even larger performance gains than we showed since those workstations tend to not be nearly this high-end. For example, back in 2011 a typical workstation likely used platter storage drives rather than an SSD which would make opening and saving files much longer than what we saw in our testing. In addition, the majority of older workstations are not going to have nearly as powerful of a GPU as what we tested with or as much RAM.

    Overall, we hope that this testing helps you decide whether you should upgrade to a new workstation or not. If you found it helpful and want us to do more testing like this please let us know in the comments section. We highly value your feedback and love hearing your suggestions!

    Permalink to story.

  2. Puiu

    Puiu TS Evangelist Posts: 2,662   +1,096

    I was expecting to see some Ryzen results added in too. I use photoshop and I wanted to see just how well it does. It is a more workstation oriented CPU than the mainstream intel I7 CPU.
    pencea and Flebbert like this.
  3. Flebbert

    Flebbert TS Booster Posts: 104   +73

    Yeah this seems like a pointless review...... how does Ryzen perform as it is designed for just that.
    pencea likes this.
  4. Puiu

    Puiu TS Evangelist Posts: 2,662   +1,096

    it's definitely not pointless, but it's true that I wanted more CPUs tested :D
    an i5 with just 4 threads would have been nice too.
    jobeard likes this.
  5. cliffordcooley

    cliffordcooley TS Guardian Fighter Posts: 9,725   +3,700

    I agree.
    I agree again. That is the newest i5 generation would have been all I would ask for.

    It is almost as if they are saying AMD doesn't qualify as a workstation, while using mainstream Intel CPU's for the review. I might would have agreed to that thought before Ryzen.
    Last edited: Mar 17, 2017
    pencea and Puiu like this.
  6. Puiu

    Puiu TS Evangelist Posts: 2,662   +1,096

    it seems the original article was published on February 14, 2017. such unfortunate timing.
    cliffordcooley likes this.
  7. pencea

    pencea TS Booster Posts: 87   +46

    Was expecting to see Ryzen tested too. Disappointing that it was left out.
  8. ZackL04

    ZackL04 TS Addict Posts: 229   +98

    I dont understand why there is an i5 3570k in there vs a i7 3770k.....

    Did you disable hyper threading on all the i7's?
  9. dorsai

    dorsai TS Rookie Posts: 17

    If I was to guess Id say most of the difference between systems comes down to improvements in storage technology and memory. CPU core speed increases not withstanding, DD4 adoption and the SSD evolution into the NVMe specification probably accounts for more than half of the improvement. It's not like Intel CPU's have evolved much in the last few years...
  10. Puiu

    Puiu TS Evangelist Posts: 2,662   +1,096

    the IPC maybe didn't improve a lot but the clocks were raised with every new generation. it's the main reason why you see a 40% difference in these tests between 2011 and 2017.
  11. merikafyeah

    merikafyeah TS Enthusiast Posts: 84   +36

    Elephant in the room: Overclocking

    Bump that 2600K to 4.5 GHz and watch those "years of generational improvements" become moot. Save your money, wait for Ryzen to mature, then make a "real" upgrade.

    Currently using a 2500K myself and I'm practically certain my next upgrade will be the future 2nd gen "Ryzen 8". Motherboard options should be far more plentiful by then.
  12. Stark

    Stark TS Enthusiast Posts: 25   +26

    I had great respect for this site.
    But last few week esp. after Ryzen launch the articles that followed, have deliberate show of intel as better or to say other side of the coin do no justice to Ryzen.
    Skipping Ryzen in this review seems another push esp. beneficiary to intel.
    Proof :
    you see Ryzen 1600x vs 7600k
    R1600x :32 sec
    7600K : 52 Sec
    really disappointed in this type of journalism guys.
    best of luck for future.
  13. cliffordcooley

    cliffordcooley TS Guardian Fighter Posts: 9,725   +3,700

    Did you not read the comments about the article being written before Ryzen release?
    Stark likes this.
  14. Stark

    Stark TS Enthusiast Posts: 25   +26

    Ah sorry about that I did not see that it was released before ryzen launch .
    hope you can make it right though.
    thank you for pointing it out.
  15. LNCPapa

    LNCPapa TS Special Forces Posts: 4,276   +461

    This is exactly the type of knee-jerk, polarizing comment that is non-productive and causes endless turmoil on these types of sites. You don't see what you want to see and suddenly they are biased. People don't even bother to ask for things anymore, just judge the writers or the site. In the future please try to think about things a bit before posting (and not just on this site).

    Not only was this prepared before Ryzen launch, you have no idea what hardware was readily available when this story was being written. You can always append to anything you write and the proper way to have handled this even if you didn't know it was written before the new hardware released would be to ask for it to be added.

    This post applies to a large number of users on this site - it's old constantly seeing 1/4 of the users says TS is biased for team red, 1/4 saying TS is biased for team green, another quarter saying TS is paid by intel, and the last quarter saying TS is too optimistic about AMD. Most of those quarters overlap btw so almost half of you just sit quietly and see how the battle is going to pan out ;)

    Think before you post and keep in mind that TS is not being paid by Intel, AMD, or nVidia. TS is focused on getting users information that might not otherwise be accessible as easily... and they do that for free - sort of (ad revenue pays the bills). Don't be afraid to disagree, but don't be upset that someone may disagree with you or your viewpoint. Just don't take things so personally and if you want to live in the land of conspiracy theories, I think they have dedicated sites for that.
    jobeard likes this.
  16. Stark

    Stark TS Enthusiast Posts: 25   +26

    I am really sorry you feel that way. am not a fan boy but I do feel what I said truly. have been with TS since way long. I do admit to my mistake and apologize.
    But I have reached such conclusion with time and after seeing the comments and the article itself.
    I just want honest opinions. may be my bias to a particular product may seem as apparent but its not. I just want Techspot to succeed in the most consumer friendly way that's all.
    if I have hurt your sentiments I am sorry . Again I accept that my comment may have caused some unintended projection on the sites image and feel free to delete it I don't mind I just what was on my mind, and I do agree what you said on your last paragraph.
    wishing you guys the best of luck in future.
  17. mbrowne5061

    mbrowne5061 TS Evangelist Posts: 747   +357

    I take issue with the 2012 'standard' being an i5, unlike the i7 for every other year's "standard"
    Stark likes this.
  18. LNCPapa

    LNCPapa TS Special Forces Posts: 4,276   +461

    Message wasn't completely directed at you Stark - I greatly expanded on the things I was talking about since it kind of reminded me of the common "you're biased" posts we see around here. Most people think they aren't fan boys even though most are at least a little. I can admit that I lean toward one side more than the other and tend to temper my posts with that in mind. My feelings aren't hurt or anything, I just hope people realize that it's not as deep or convoluted as many make it out to be around here. We all have a passion for technology and we share the things we know and learn with each other.
    jobeard likes this.
  19. jobeard

    jobeard TS Ambassador Posts: 11,158   +986

    Being agnostic on system platforms, I am frequently disappointed in the bias so openly displayed - - all all subjects. Just because my favorite XYZ wasn't mentioned, doesn't mean I should blast away because ABC has a following and was discussed.

    Learn to read and analyze for yourself any topic you read - - practice analytical thinking just to keep your hand in the subject. After decades in software I have discovered that practitioners of technology are frequently biased and intolerant of other opinions :sigh:

    The article is well done and very useful - - and I haven't done PS in decades!

Similar Topics

Add your comment to this article

You need to be a member to leave a comment. Join thousands of tech enthusiasts and participate.
TechSpot Account You may also...