Skip to main content
Participant
May 31, 2022
Question

Slow Saving 16bit images M1 Max

  • May 31, 2022
  • 3 replies
  • 5409 views

I have a new Mac Studio M1 Max base model with 64g ram and 2 TB HD. Monterey 12.4

I also have a 2019 Macbook Pro 16 2.3ghz 8 core intel i9 with 32g ram and 1 TB HD. Catalina 10.15.7

When doing some tests in photoshop 23.3 saving a 16bit PSD with a couple of layers and around 15 adjustment layers, the image takes over a minute to save on the Mac Studio and exactly the same time on the MBP 16. 

I also did a smart blur test at full quality and the M1 Max took 34 secs with the MBP at 31secs.

I was expecting the M1 Max to be much quicker than the intel MBP, has anyone else had this experience?

I've done these tests with rosetta on and off, plus purged all the presets and a clean install. I've also tried older versions of PS and the Beta ones.

 

This topic has been closed for replies.

3 replies

Earth Oliver
Legend
June 5, 2022

Testing on my M1 Max today and my numbers are nowhere near similar:

v 23.3.2: 6GB saved, 8bit, not-compressed, 2 layer, rbg, .psb:

with background save OFF: 29 secs

with background save ON: 51 secs

and for those who don't know, disabling Background Save makes a huge difference in speed, but for some reason the Ps default pref = slowest save times. Even though most of the cores are available during a save, Ps throttles down when background saving. Ugh.

 

File dims are 45000 x 30000px

davescm
Community Expert
Community Expert
June 5, 2022

@Earth Oliver Do you see a difference between v22 and v23 in terms of speed and file size saving uncompressed, i.e are you also seeing the additional compression on an m1?

All figures above were background saves.

Dave

davescm
Community Expert
Community Expert
June 1, 2022

One thing to look at, if saving and loading speed is important, is disabling PSD file compression in Preferences > File Handling. This increases file size on disk but vastly improves saving and loading time as Photoshop does not need to compress and uncompress the data. For me that is a good trade off and I've worked that way for years.

 

Dave

 

Participant
June 2, 2022

I was doing some tests along this line last night saving large images with compression on and off and the new Mac Studio actually saves in half the time than the intel MBP when PSD compression is disabled.  When it is on there is no difference in save time between the computers which shows it's a fixed bottleneck.

Conrad_C
Community Expert
Community Expert
June 2, 2022

The maximum performance advantages with the M1 Max come from the extra cores for the CPU and GPU, and the specialized video encoders. Apple marketing and benchmark tests around the web are set up to show those off. But like a lot of photo applications, Photoshop doesn’t currently or can’t use all of those, for various reasons.

 

When you save a large file with Photoshop, monitor performance using the CPU History window in Activity Monitor. You will probably find that most of the CPU cores go unused. I think saving is single-threaded in Photoshop. And the thing about the M1 variations from the base M1 through the Pro and Max and Ultra is that single-core performance is about the same. The higher models get faster speeds out of doubling/quadrupling the core counts…but that only helps when an application actually uses all the cores. When an application uses one core on an M1 Ultra, it’s no faster than a base M1.

 

I don’t think the GPU is involved in saving to storage, so all those GPU cores also go unused.

 

The specialized video encoders don’t get used because Photoshop is not rendering to a video file format the encoders were designed for.

 

In addition, although the internal storage of the Mac Studio is capable of some mind-blowing data transfer speeds (like over 5000GB per second), file system overhead and other bottlenecks mean you will rarely see anywhere close to that speed when saving a document from applications including Photoshop. You can monitor read/write speed yourself using the Disk tab in Activity Monitor. When I try it, disk transfer speeds during a Photoshop save are usually in the low hundreds of MB/sec.

 

The M1 Max Mac Studio is a great value overall and I wouln’t mind having one, but the maximum speed benefits happen with applications that can easily divide up their workloads into multiple CPU and GPU cores, such as video editing and 3D. The extra cores save lots of time in  Lightroom when exporting in bulk, because it appears to use the cores to export multiple images in parallel. But Photoshop is not saving multiple images, so it can’t save time that way.

Chris 486
Community Expert
Community Expert
May 31, 2022

Hi David!

 

Both computers you have mentioned are setup to run photoshop very well. Newer/higher spec'ed equipment doesn't always result in faster software speeds. You proccessor is newer but that doesn't mean it's processing faster/more effienciently in every case. You have more RAM but in general, software won't use it just because it's there. Once you hit the recommendtation for software requirements, (photoshop here: Photoshop system requirements (adobe.com) ) it's usually demishing returns on productivity increase.

 

Regarding the M1 chips being slower in your later example. It's a pretty new chip still and I would imgaing that future software and OS updates will be tailored to pull more "speed" out of that chip. It's a great machine that will last you for a good few years of photoshop!

 

Participant
June 1, 2022

Thanks for your reply.

I appreciate that they are both great computers and highly specced, but I've seen numerous reviews comparing these exact models and specifically, in photoshop there were great gains in speed so I'm not sure why I am not seeing these. 

Chris 486
Community Expert
Community Expert
June 1, 2022

That would be tough to say. Most credible test reviews make it a point to mention that "your results may vary". By no means am I saying that reviews are incorrect or that you shouldn't be seeing better results. Just saying that those reviews could be setup to test very specific conditions that don't directly translate to general/real world use. There could be OS/Software settings/configurations that are different then what you are using, updates over time that change performance in the software or OS, What exact hardware they are using, the list goes on...

 

I would first check that the OS/firmware updates, and see if any background programs are running that are sucking up a lot of memory or are cpu intensive. You may have done that but just throwing out a place to start for general troubleshooting. Good luck!