GPU cores vs RAM

[jpshephard81]

I am wanting to buy a higher spec Mac and I am wondering if LR will benefit from more GPU cores (think Mac Studio) or should I just focus on getting as much RAM (unified memory) as possible?

Thanks

[Sameer K]

Hey, @jpshephard81. Welcome to the Photoshop Community. I'll try to help you decide.

As a baseline, refer to the Lightroom System requirements page: https://adobe.ly/3PhXa8F

I recommend researching independent benchmarks & stress tests run on the specific device with Lightroom.  This conversation is from 2 years ago, and you will find a lot of input from experts and legends of our community: https://adobe.ly/4j0Ep7w

This should help you. Let me know if it does.

Thanks!
Sameer K

(Type '@' and type my name to mention me when you reply)

[Conrad_C]

There's usually little benefit in “maxing out” any component on a modern pro computer.

For example, “getting as much RAM (unified memory) as possible” for a MacBook Pro means 128GB. That costs an extra $1200, but even if you ran a couple other apps alongside Lightroom, most of the 128GB you paid for would go unused. For most Lightroom users, a good amount is between 32GB and 64GB depending on image sizes and workloads, even lower for casual users. Once you have the amount of memory your tasks need, adding more memory doesn’t speed up anything. The people who actually need more than 96GB tend to work with very large databases, very large virtual machines, AI/machine learning models…

If you get more GPU cores, the only features that are faster are those that are GPU-accelerated. I usually use Lightroom Classic so my knowledge might be a little off for Lightroom, but the GPU-accelerated features tend to be photo processing (Detail view), any feature using AI, and probably export.

The people who need the Mac Studio the most are those who frequently do bulk processing of GPU-accelerated features. For example, if you need to push out several shoots a week of 100 images each, and you applied adaptive AI masks, copy them to other images, then export them all, a Studio would save a lot of time. But if most of the time you work with small sets of images, a Studio probably wouldn’t make that much difference because there would be little work to spread across all the cores. Some cores would sit around waiting for something to do.

A special case is AI Denoise. Tests show that Denoise speed generally tracks with the number of GPU cores, so double the GPU cores means Denoise finishes in half the time. That matters a lot to someone who needs to denoise and export 350 high ISO images of a low-light theater or music show.

You might want to review the recent Mac videos on the ArtIsRight channel on YouTube, because he has done a lot of photography software tests (including Lightroom) across several generations of Apple Silicon Macs. You can see how various Mac models vary in performance across different tasks, and which ones tend to depend on CPU, GPU, and memory.