Copy link to clipboard
Copied
Does anyone have any insight into how a top-end Apple system would compare to an equivalent PC system? For example a Mac Studio M2 Max with 96GB RAM vs i9 13900k, 4070ti, and 96GB RAM.
The only benchmark I can find for Photoshop is PugetBench, but this seems to still have the Mac system running through Rosetta. In other relevant benchmarks (Single Core CPU) the PC seems to perform better, but I'm unsure of how this would translate to real-world Photoshop usage. Does Photoshops optimization for the Apple processors make up for its weaker raw computing power?
Also, from an OS standpoint does either platform have a considerable advantage at the moment? In regards to reliability/stability, and color management.
Copy link to clipboard
Copied
The platform is irrelevant. They are both highly optimized and efficient for Photoshop and other Adobe software. That's just down to what you prefer.
In terms of reliability, a more important distinction nowadays is form factor. A laptop is always more risky than a modular desktop system, because there will be many layers of vendor modifications targeted at consumer applications, and this tends to get in the way of advanced applications like Photoshop.
I can only speak for Windows, but you will see here in the forum that 99% of users with problems are using laptops.
Copy link to clipboard
Copied
The OS itself may not be as relevant for performance, but the associated hardware may be. Even between intel and amd cpus, intel seems to have a slight lead in terms of optimization with photoshop. An i9 13900k is 11% faster than a Ryzen 9 7900x (in PugetBench) even though on paper they should be closer in terms of performance. I think this difference may be much more significant with Apple processors.
I've seen videos in other programs where the Apple chips perform much better than one would assume based on specs alone. In this video https://www.youtube.com/watch?v=5dhuxRF2c_w (at around the 6 minute mark) the Mac Studio M2 Ultra is able to stabilize a clip in less than 1/3 the time of a PC (7800X3D + 4090). The M2 Ultra is a faster cpu in a cpu based task, but its not 3x faster (atleast not on paper). This would lead me to believe that the program he is using is more optimized for Apple silicon. Which makes me wonder where Photoshop stands. Since Photoshop is a lightly threaded non hardware intensive program optimizations like this could have a large impact on real world performance.
Copy link to clipboard
Copied
This is simply false. I'm using the same project on a PC workstation with 4 times the amount of RAM and core count compared to my M1 Macbook Pro and the UI in After Effects is so laggy on my PC. It's borderline impossible to use for certain projects. This is in the latest version of AE 2023 and also 2024 and isn't down to slower hardware on my PC.
Copy link to clipboard
Copied
I have an i9-13900K with RTX4080, 64Gb RAM.. nearly empty 2tb 980 Pro for Scratch.
Photoshop is lightly threaded so it tends to be CPU cl;ock speed that counts. I responded to another hardware thread earlier today, and ran Resource Monitor while performing some heavy duty funtions, and I couldn't stress the system no matter how hard I tried. The CPU was peaking at 50 to 60% , but RAM, GPU and drive activity was basically zero.
Even if Photoshop and maybe Illustrator will be your main use, go ask about hardware on the Premier Pro forum
Check out the Hardware links
Copy link to clipboard
Copied
I'd search for benchmarking sites, keeping in mind that the tests may or may not be "real world" or reflective of your type of work or specific image processing operations.
Copy link to clipboard
Copied
As I have mentioned earlier "The only benchmark I can find for Photoshop is PugetBench, but this seems to still have the Mac system running through Rosetta. In other relevant benchmarks (Single Core CPU) the PC seems to perform better, but I'm unsure of how this would translate to real-world Photoshop usage.".
Copy link to clipboard
Copied
As I have mentioned earlier "The only benchmark I can find for Photoshop is PugetBench, but this seems to still have the Mac system running through Rosetta. In other relevant benchmarks (Single Core CPU) the PC seems to perform better, but I'm unsure of how this would translate to real-world Photoshop usage.".
By @romany36267645
The articles I linked to below all have extensive results specifically for Photoshop, but you can change the filters for other apps. The 'systems they sell' link points to systems they consider ideal for Photshop. I chose to use an RTX4080 and a pair of 2Tb Samsung 980 Pro drives, and moved some fast storage over from my old system. I'm hoping to get five years out of this system.
Many years ago a couple of guys from the Premiere Pro Hardware forum had site listing results from hundreds of systems that ran an in depth set of benchmarks they devised for Premiere Pro. They stopped colating results after CS5 because it was a huge job for them. The results were interesting in that the highest placed Mac system was in the 200s. Things may have changed since then, but Windows PCs are easier to configure for high demand workflows, and MUCH cheaper. I remember that Scott Kelby and his buddies collectively known as The Photoshop Guys all used MBPs but Photoshop needs a lot lower system resources than Premiere Pro. So I have stayed with Windows PCs and Workstations through many upgrades.
Copy link to clipboard
Copied
Puget does extensive testing on PC components, but Apple processors are absent from these published comparisons. As I stated before their benchmarking software still has "Mac systems running Photoshop through Rosetta" which could have a significant impact on performance.
Copy link to clipboard
Copied
That's always going to be Puget Systems for me, and I think the OP is aware of them. I can't think of another site that does in depth testing of different hardware using it for content creation apps. In fact usually try to match the specs of the systems they sell when updating.
Copy link to clipboard
Copied
My background is primarily Mac, this is my take on it.
In other relevant benchmarks (Single Core CPU) the PC seems to perform better, but I'm unsure of how this would translate to real-world Photoshop usage. Does Photoshops optimization for the Apple processors make up for its weaker raw computing power?
By @romany36267645
As a still image editor where often only one document is open, there is only so much parallelization that can be done. I think that favors single-core performance in the specific case of Photoshop. Applications built to regularly process many images or frames at once, such as Lightroom Classic, Camera Raw, or video editors, can typically make much more effective use of all available CPU and GPU cores.
Now if we’re talking Mac vs PC, a good PC may be faster in Photoshop. Intel-based CPUs seem to be capable of higher single-core and often also multi-core performance, largely because they have a higher power budget. Apple Silicon is definitely more efficient, but that is mostly a concern for mobile or small desktop machines where there is much less ventilation and cooling. The higher Intel power budget means if you have a tower with great cooling, you should be able to simply pour a lot of wattage into the tower and the CPU/GPU should be able to beat just about any Mac that shows up. Also, if you expect to do much 3D, right now the Mac is not really a contender in 3D against Nvidia graphics cards for PCs.
Where Apple really shines right now is in those mobile and small desktop uses. The Mac laptops can in some cases perform surprisingly well vs Intel for intensive applications, not because the Mac is actually more powerful, but because of performance per watt: To achieve the same level of performance, many PC laptops/mini desktops need power and cooling that are not available, so they slow down either because of not enough watts or too much heat. There are some PC workstation laptops that can solidly beat Mac laptops in sheer performance, but only when they are connected to their 140+ watt power supply plugged into a wall and fans at maximum; not when unplugged.
But CPU is not everything, and when comparing them you have to carefully consider your photo work vs the GPU. For a PC, you’re choosing an amount of RAM + discrete graphics RAM. For a Mac, you have no choice of GPU except core count, and system memory is shared with graphics. But not in the same was as Intel integrated graphics; Apple Silicon is said to manage system vs graphics memory in a much more flexible and dynamic way.
This has two consequences. One is that if you have a PC and Mac with 96GB memory, the PC is really 96GB RAM for the system + whatever amount or graphics memory is on the discrete GPU; and the Mac has 96GB of memory dynamically allocated between system and graphics; so in theory the Mac has less potential memory available for Photoshop because graphics needs some.
But not always. I saw a YouTube video of an interesting edge case where a specific operation went faster on the Mac because the PC graphics card had 16GB graphics RAM, but because the Mac had 64GB of Unified Memory and the application did not need most of it, the Mac could throw much more than 16GB of memory at graphics, so it won the test. But that is an edge case for this discussion, because for photo editing, as long as there is at least 8GB available to graphics, it is rare that the amount of graphics memory would be the limiting factor. But that is an example of how Apple Silicon Unified Memory is not always a drain on system memory and can potentially make more memory available to graphics, on a Mac with far more system memory than applications need at the moment.
The other thing to keep in mind is what parts of Photoshop need what components. Many Photoshop benchmarks are a specific selection of features from a certain point of view, but if you don’t edit that way, you’ll get different results. And some benchmarks represent very traditional workflows, not fully accounting for new features in the last 5 years. For example, the more you edit using features from the first 20 years of Photoshop, the more single-core performance will help. But the more you use features developed during the last 5–10 years of Photoshop, the more the CPU/GPU cores and graphics memory matter. Specifically, Photoshop/Camera Raw features based on AI, such as upscaling and detail enhancing, denoiseing, content removing, healing, etc. are much more dependent on having a good recent GPU. And we know those features are an intensive area of development right now. For AI, graphics in Apple Silicon Macs seem to perform well compared to PC discrete graphics.
So that’s a long rambling way of saying be careful about comparisons, because of architectural differences between current PCs and Macs, and how various components are used by different parts of Photoshop.
A short version of this could be that the more your work can be on a desktop with a beefy power supply and serious cooling system, the more a PC with a high end recent graphics card might result in higher potential maximum Photoshop performance; and the more you want to get a lot of work done fast on battery or in a mini desktop, the more you want a Mac. Or, as in Trevor’s example, if your most intensive work doesn’t come close to pushing either platform to the edge, then either would be more than good enough: Then you just choose the one you like more or that costs less.
Copy link to clipboard
Copied
Thank you for taking the time to reply Conrad, it is greatly appreciated. I won't touch on all the points you've made as I believe the majority of them have already been addressed in my previous posts.
"A short version of this could be that the more your work can be on a desktop with a beefy power supply and serious cooling system, the more a PC with a high end recent graphics card might result in higher potential maximum Photoshop performance; and the more you want to get a lot of work done fast on battery or in a mini desktop, the more you want a Mac. Or, as in Trevor’s example, if your most intensive work doesn’t come close to pushing either platform to the edge, then either would be more than good enough: Then you just choose the one you like more or that costs less."
"might result in higher potential maximum Photoshop performance" This is exactly what I'm trying to find a conclusive answer for. The desktop I've described in my initial post should have better performance but I don't know for certain. I'm sure both of them would be "good enough" but it would be helpful to know which system is truly faster in the real world.
Copy link to clipboard
Copied
If you want maximum performance, you want a balanced system optimized for your purposes.
There's little point in a fast processor if you have bottlenecks elsewhere!
For Photoshop, the major traffic points are disk throughput and the GPU. Those two are the priorities. The CPU is far down on the list. If you have to work off external drives, for instance, you're already set way back and no fast CPU can save you.
Lightroom is a little more CPU-intensive, but no more than a standard i7 handles with flying colors.
This is why building your own desktop is optimal. You get to pick exactly the components you need, no bells and whistles getting in the way. Done with a little consideration, that's also the most solid and reliable workhorse you can get.
Copy link to clipboard
Copied
I would agree that having a balanced system is important. The build I've outlined in my first post is about as optimized as you can get at the moment. i9 13900k, 4070ti, and 96GB RAM. I didn't mention storage, but this would be multiple reliable PCIe 4.0 NVME M.2 drives.
"There's little point in a fast processor if you have bottlenecks elsewhere! For Photoshop, the major traffic points are disk throughput and the GPU. Those two are the priorities. The CPU is far down on the list."
This is simply incorrect. Based on Puget Systems research "While GPU acceleration is gaining traction, right now your choice of CPU is usually going to make a much larger impact on overall system performance."
I recommend you take a look at this article in general, it clears up a lot of misinformation.
https://www.pugetsystems.com/solutions/photo-editing-workstations/adobe-photoshop/hardware-recommend...
They're the only company I know of that extensively tests different PC components and configurations based on Photoshop usage.
Copy link to clipboard
Copied
That's largely outdated. The GPU is where the rapid evolution is these days, if not revolution. Things have changed dramatically in just a few years. With each new version, more and more functionality is moved to the GPU.
Again, that doesn't mean you necessarily need the fastest high-end GPUs, but it does mean you need a GPU that can execute advanced functions reasonably fast. At the moment it's formalized as DirectX Feature Level 12.0, and for ACR/Lightroom a minimum recommended VRAM of 8 GB. Many functions also require Tensor cores optimized for AI tasks to run well - epecially in ACR/Lightroom.
Photoshop will still work with an older/weaker GPU, up to a point - but some functions will be slow, even unbearably slow (like e.g. Denoise).
And I still maintain that any CPU above i7-level is largely wasted for Photoshop. Analyzing proceedings in Task Manager here, I have never been able to provoke any particular load on the CPU. Getting an i7 to peak is extremely difficult in Photoshop, much easier in Lightroom. Yes, there are a few filters that use the CPU more extensively, but they are the exception.
The main load is on disk I/O. The scratch disk is absolutely critical for Photoshop, and tthe bigger your files, the more critical. The scratch disk is Photoshop's main memory, and it needs to be fast. PCIe 4.0 NVMe drives really pay off.
Copy link to clipboard
Copied
The article is not "largely outdated". The information is based on Photoshop benchmarks for current-gen hardware.
"Compared to the high-end GeForce RTX 4090, the much more modest RTX 4070 is within a single percent – even when only looking at GPU-accelerated tasks – and the 4060 Ti is only a couple percent back."
Yes, there are diminishing returns when buying a 13900k over a 13700k and the comparison can be found below. The price difference between the two is about CAD 150 (USD 110), and the slight difference in performance may be worth it for some people. The price difference between a 4070ti and a 4080 however is about $400 and would have a smaller impact on Photoshop performance than the switching from the i9 to the i7.
"This isn't about numbers and benchmarks, it's about practical use." These benchmarks are based on real-world tasks. There are no better ways of objectively measuring performance that I know of.
"It would be nearly impossible to test everything in Photoshop, but our benchmark is designed to test a broad range of tasks in order to give an accurate overall picture of how a system performs in Photoshop.
General Tasks:
Open 18MP .CR2 RAW Image, Resize to 500MB*, Rotate, Magic Wand Select, Mask Refinement, Paint Bucket, Gradient, Content Aware Fill, Save .PSD File, Open .PSD File
Filter Tasks:
Camera Raw Filter, Lens Correction, Reduce Noise, Smart Sharpen*, Field Blur*, Tilt-Shift Blur*, Iris Blur*, Adaptive Wide Angle, Liquify
* denotes GPU-accelerated tasks"
Copy link to clipboard
Copied
The article is not "largely outdated". The information is based on Photoshop benchmarks for current-gen hardware…These benchmarks are based on real-world tasks. There are no better ways of objectively measuring performance that I know of.
…
By @romany36267645
Open 18MP .CR2 RAW Image, Resize to 500MB*, Rotate, Magic Wand Select, Mask Refinement, Paint Bucket, Gradient, Content Aware Fill, Save .PSD File, Open .PSD File
Filter Tasks:
Camera Raw Filter, Lens Correction, Reduce Noise, Smart Sharpen*, Field Blur*, Tilt-Shift Blur*, Iris Blur*, Adaptive Wide Angle, Liquify
* denotes GPU-accelerated tasks"
I think I see what D Fosse getting at. It’s similar to in my earlier reply where I said a realistic comparison depends on the feature mix of the benchmark compared to what you are actually going to use today. I have had similar issues with that benchmark, some examples below:
Open 18MP .CR2: Oh, I bought that 18MP Canon DSLR…in 2007. Today, even though I am slow to upgrade cameras, I no longer regularly use any cameras where the raw file is only 18MP. And I’m just midrange…today, the serious Photoshop users here are all asking about performance processing files from their 48MP, 60MP, etc. sensors, sometimes +100MP with pixel shift resolution boost or panorama merging.
Magic Wand Select: Although this is still popular, even the most basic computers will have no problem with this feature from 1999. The selection performance challenge today is for the more processor intensive AI-based features such as Object Selection, Select Sky/Background, and Select People in Camera Raw. These are all far more intensive for both CPU and GPU, so we all need to know more about this, but they are not yet incorporated into common Photoshop benchmarks.
Gradient: Photoshop just completely renovated gradients around a new non-destructive, easier to edit gradient object. Those tend to be more processor-intensive, and again is so new it is not clear if any benchmark is measuring it yet.
Content-Aware Fill: OK, that is a little more up to date…but not as of last month. The new Generative Fill is easier to use, returns much better results…but requires a mix of on-board and network/cloud processing; probably not part of any benchmark yet.
Save/Open PSD File. This is fine, but if someone really is working with huge Photoshop files, PSD can’t handle it: Where is the test of PSB (Photoshop Large Document Format)?
Filter Tasks: Some of those are still widely used, but many are not. For example, how many of those blurs do you use every day? Liquify is still quite popular though, and good to include.
So those are examples of what I meant by talking about how a benchmark is weighted between what we all used in Photoshop in 2003 and what we are all using now in 2023. The omission of Smart Objects (very processor intensive) and AI features (heavy demand on GPU) are particularly unfortunate. I hope that major benchmarks are working on this, because Photoshop has changed more in the last 5 years than in the previous 25. (The PugetBench page says it was last updated in 2019…4 years ago.)
Copy link to clipboard
Copied
I've taken what you said into account. These benchmarks, while not perfect are all we've got at the moment. Sure a "realistic comparison" on feature mix that I actually use today would be great, but how am I to do that practically?
My original points still stand, and unlike what D Fosse has said the CPU is not "far down on the list" of important components. Most of the new features you've outlined are still processor intensive as you've pointed out.
Copy link to clipboard
Copied
For Photoshop, the major traffic points are disk throughput and the GPU. Those two are the priorities. The CPU is far down on the list. If you have to work off external drives, for instance, you're already set way back and no fast CPU can save you.
By @D Fosse
You keep saying that Dag, but I am sure you are wrong. Chris Coz used to tell us that CPU clock spped was everything for Photoshop. Photoshop is lightly threaded, but I do see a lot of multi thread usage if I try really hard. This a 5000 pixel brush with 1% spacing dragged all over the screen in a 50,000 pixel square 16 bit image.
Something I have found strange with this system is the the CPU idles as around 5.3Ghz, but drops to around 4.9Ghz when loading it with the brush test. GPU is doing nothing.
I tried Nueral Filter Restoration on a 2000 x 3000 pixel photo with a LOT of damage. It only took about two seconds, but still barely put a blip on the GPU
Check out the Puget Systems benchmarks. GPU is still not crucial for Photoshop performance.
Copy link to clipboard
Copied
All right, all right...this is beginning to look like medieval scholastics hurling benchmarks at each other 😉 But I did my bit in leading us down that rabbit hole, so I'll eat up some of it. Mea culpa 🙂
I was making a point: it may be more effective to avoid bottlenecks than to max out choice components. A balanced system, and a real world one. I never said you need an RTX 4090, same argument in fact. I do still insist that if you're going to work with big files, the easiest way to kill that is to ignore the scratch disk.That will kill it much more effectively than a slow CPU.
I stand by most of what I said. But yes, I probably overdid it to make my point clear. Can we go back to work now? 😉
Copy link to clipboard
Copied
"might result in higher potential maximum Photoshop performance" This is exactly what I'm trying to find a conclusive answer for. The desktop I've described in my initial post should have better performance but I don't know for certain. I'm sure both of them would be "good enough" but it would be helpful to know which system is truly faster in the real world.
By @romany36267645
What the replies in this discussion are showing is that the answer to choosing a computer for Photoshop, even within just the PC or just the Mac world, is similar to buying anything else tech-related in that there often is not one conclusive superior solution. Whether you’re buying a car, an appliance, etc. the answer is commonly “it depends.” And that applies here too.
One way to get closer to “conclusive” is to narrow down the scope. From your replies it’s clearer that you’re focusing on a desktop, for mostly Photoshop, and you want the fastest potential Photoshop performance even if your actual workload might not need it. For that specific combination, my suspicion even as a loyal Mac user is that a PC tower desktop with 64–96GB RAM and a top tier Nvidia graphics card is probably going to “conclusively” have faster potential performance than any Mac.
What the Mac Studio is very well suited for is the “big middle” demographic of users who just want an efficient, quiet little desktop that, even in its base model, has more than enough power out of the box to smoothly take care of what most people need to get through in Photoshop. But if the best Mac Studio is up against a well-equipped PC desktop in a brute force contest where the goal is the fastest potential processing of the very largest and most intensive Photoshop files (large pixel dimensions, heavy compositing, Smart Objects, lots of effects…), I expect the Mac Studio to fall well short of PC performance in both CPU and graphics.
Copy link to clipboard
Copied
I find it interesting that it's so difficult to find a performance comparison between the two systems.
"my suspicion even as a loyal Mac user is that a PC tower desktop with 64–96GB RAM and a top tier Nvidia graphics card is probably going to “conclusively” have faster potential performance than any Mac."
I agree, but this is still speculation. Unfortunately seems to be about as good of an answer as we can get at this point. Unless of course, someone takes the time to do some independent testing.
Copy link to clipboard
Copied
look, unless your job is running actions on thousands of files, none of this matters in the least. Most of the time spent in Ps is doing actual work, not waiting for Ps to process anything.
Copy link to clipboard
Copied
I agree, but this is still speculation. Unfortunately seems to be about as good of an answer as we can get at this point. Unless of course, someone takes the time to do some independent testing.
By @romany36267645
And the moment you get it done, you probably have to do it again. Because everything is constantly changing. What about the new Snapdragon X Elite “Apple Silicon killer” CPU announced a couple of weeks ago? But what about the new “Mac Pro desktop performance in a backpack” Apple Silicon M3 Max that was announced shortly after that? And what about the next NVIDIA graphics card? It’s going to keep going back and forth.
In the end, the reliable old rule several of us have pointed out is still the best advice. Forget about the constantly moving “bleeding edge,” and focus on what your own work needs: Just get something you like and can afford, because the midrange and upper midrange of the Mac and PC lines, properly specced, are fast enough for most day to day Photoshop work.
Copy link to clipboard
Copied
I'm Team Apple all the way. In my LONG experience in both IT and graphics, its the Mac systems that tend to have fewer problems with malware, drivers, and usability. My day job is at a Microsoft shop and I used a Windows PC daily for several years before getting a Mac (and an Intel Mac at that) and the PC was just a headache.
These days its mostly down to what you prefer and are used to. Windows users seem to accept problems as the norm (look at all of the driver issues you see just on here) while Mac users tend to demand better reliability.