M5 vs M1 Max vs M2 Ultra - Lightroom Test - Shocking Results

[m1maverick]

Do you normally export 750 files from lightroom? Is that part of your normal workflow? What is the relevance of such a task if you (or others) rarely do it.

Running tests that may have little to no bearing on real world usage just turns into noise. How often do people export 700+ images from LR? Is it an hourly task, daily, weekly, once the project over?

The same can be said of Geek Bench yet people use it all the time.

 

[maflynn]

The same can be said of Geek Bench yet people use it all the time.

And also people complain that the synthetic benchmarks fail to convey real world results, so much so that many of the popular tech sites augment their testing with real world usage testing criteria.

 

[RALentz]

...
I have found that the best thing one can do for PS or LR adjustments of large files is get LOTS of ram rather than better processors (within reason). I have had LR and PS take near 100GB memory, which of course uses lots of swap, resulting in lagginess and long pauses. ...

This.

The M2 Ultra setup is the outlier at 64GB of RAM, which in reality is more than twice the M1 Max when you account for system, application, and 'cached files' allotment. If the M2 Ultra can act mostly in RAM then it's memory bandwidth advantage really comes into play, which appears to be reflected in the timings.

Also helps explain the parity between the M1 Max and the M5. Folks cite the memory bandwidth difference, but if swapping is necessary then that distinction is equaled by the similar difference, though in the other way, of the SSD speeds.

This all just comes around to what a tragedy it is that Apple charges as much for an incremental amount of additional RAM as one would pay for a full 32GB on the PC side.

-R

 

[m1maverick]

And also people complain that the synthetic benchmarks fail to convey real world results, so much so that many of the popular tech sites augment their testing with real world usage testing criteria.

I'd say his benchmark is more realistic than GB.

 

[krakman]

Do you normally export 750 files from lightroom? Is that part of your normal workflow? What is the relevance of such a task if you (or others) rarely do it.

Running tests that may have little to no bearing on real world usage just turns into noise. How often do people export 700+ images from LR? Is it an hourly task, daily, weekly, once the project over?

I photograph events such as weddings and conferences, and I regularly export hundreds of JPGS from RAW, approx 2 to 3 times a week, on average, for the last 20 years.

Further more, my network of professional photographer friends, also regularly export hundreds of JPGs from RAW on a similar basis.

So this is a very important data point for some people and I thank the OP for doing this test and providing the information.

 

[Bungaree.Chubbins]

You're right but if it doesn't mimic real world usage, then it largely fails to impart knowledge and just adds to the inane chatter.

Why not test how many Smoots per hour this can do

It might not be relevant to you, but it clearly is to other people. While this particular test might only be relevant to some, it is a data point that, along with a lot of other data points, can paint a picture of generational Apple Silicon performance.

 

[burgman]

Seems a few posters are in pissy moods and wanted to bring others down today. If the tests don't apply in your world, move on.

 

[dmccloud]

It might not be relevant to you, but it clearly is to other people. While this particular test might only be relevant to some, it is a data point that, along with a lot of other data points, can paint a picture of generational Apple Silicon performance.

It's a situational data point. I personally care more about other aspects of M series performance, primarily related to code compilation and ML/AI workloads than I do about Lightroom exports. That being said, I also recognize that my workloads are not representative of the Mac userbase as a whole, who more often than not uses their Mac for basic tasks rather than batch photo processing, coding/development, etc.

 

[Bungaree.Chubbins]

It's a situational data point. I personally care more about other aspects of M series performance, primarily related to code compilation and ML/AI workloads than I do about Lightroom exports. That being said, I also recognize that my workloads are not representative of the Mac userbase as a whole, who more often than not uses their Mac for basic tasks rather than batch photo processing, coding/development, etc.

Yes, and if we take some more photography based situational data points, some for coding, add a few for gaming, running local LLMs, working with video, etc, and we start to build a nice picture, and can start to extrapolate to other areas like my thermal performance modelling.

 

[Alameda]

Well, that literally was my question - how often do you export 700+ images?

It’s helpful information.
I typically open a folder of 500 45mp RAW photos, +\- 250, browse the thumbnails, then perform a ton of cpu intensive edits. My M1 Pro is very responsive, but there are still lags here and there, particularly Export and also previewing noise reduction.

So fixing all of the little 2 second lags here and there would be great, but you can’t easily benchmark 2 seconds.