Photoshop + DaVinci Resolve M1 native!!!

[JohnnyGo]

Anyone has seen benchmarks comparing their performances against Intel software ?

I would love to glimpse at the gains these new M1 native versions bring to the table.

So far, so great!

Real-World Test: Photoshop for Apple Silicon is a Game-Changer

Adobe finally released the full version of Photoshop for Apple Silicon, and we have good news: it's fast... really, really fast.

petapixel.com

 

[CheesePuff]

Anyone has seen benchmarks comparing their performances against Intel software ?

I would love to glimpse at the gains these new M1 native versions bring to the table.

So far, so great!

Real-World Test: Photoshop for Apple Silicon is a Game-Changer

Adobe finally released the full version of Photoshop for Apple Silicon, and we have good news: it's fast... really, really fast.

petapixel.com

From the article... very impressive:

What does that mean in real terms? An Intel-based 13-inch MacBook Pro with a 2.3GHz quad-core i7 and 32GB of RAM takes about 2 minutes and 45 seconds to merge a 6-photo panorama shot on the Nikon D850 (full res .NEF Raw files). The M1 Mac mini running optimized Photoshop does this same task in 1 minute and 14 seconds. Even a $6,700 fully-loaded 16-inch MacBook Pro with an 8-core Intel Core i9 and discrete GPU takes 1 minute and 52 seconds.

 

[jeanlain]

From the article... very impressive:

Not that impressive IMO, except the photo merge test.
The GPU test is hugely disappointing, with the M1 not even beating the lowly Intel Iris of the 13" MBP. But something's wrong with that test. It runs much slower under Rosetta, even though Rosetta shouldn't have much impact on GPU performance.

 

[Quackers]

I bought a Lacie HDD a week or two ago and it came with a 1 month subscription to all Creative Cloud/PS/Lightroom and all of that so as soon as I saw it had become M1 native (today) I redeemed it.
I've just installed Photoshop and Lightroom and the installer automatically selected the Apple Silicon versions.
So I've got them but have no idea what to do with it all
I'd better get watching some videos

 

[BootLoxes]

I am wondering how DR compares to FCP in performance. I tried to like FC but its a weird video editor for me. Davinci makes more sense

 

[eltoslightfoot]

Not that impressive IMO, except the photo merge test.
The GPU test is hugely disappointing, with the M1 not even beating the lowly Intel Iris of the 13" MBP. But something's wrong with that test. It runs much slower under Rosetta, even though Rosetta shouldn't have much impact on GPU performance.

I see you must have missed the two paragraphs directly after the chart for which you are concerned:

Unsurprisingly, the M1 Mac mini loses to the competition in raw GPU performance, more-or-less matching the onboard graphics of the quad-core Core i7 that’s in the 13-inch MacBook Pro (full review here). But even with this score working against it, the Mac mini running Apple Silicon-optimized Photoshop managed to get the second highest Overall score we’ve ever seen out of PugetBench.


What’s more, none of the computers we’ve reviewed, not even the most expensive 16-inch MacBook Pro you can buy or the Razer Blade Studio Edition, has ever broken the 100 mark on the PugetBench Photo Merge test. Running optimized Photoshop, the M1 Mac mini hit 130+ in run after run after run.

Impressive is exactly the word. Now if you want to wait three months for the next batch of higher end Apple Silicon, I get that. It will be even more impressive. Forgive me if I enjoy my lightning fast M1 MBA in the meantime.

 

[quarkysg]

The GPU test is hugely disappointing, with the M1 not even beating the lowly Intel Iris of the 13" MBP. But something's wrong with that test. It runs much slower under Rosetta, even though Rosetta shouldn't have much impact on GPU performance.

Could it be that the M1 GPU is not used at all? I suspect Adobe's code is checking for GPUs and not recognising the M1 GPU, falling back to CPU.

I guess someone with a copy running in a M1 Mac, and knows which test is used, could test it and check for CPU and GPU utilisation.

 

[ArPe]

Could it be that the M1 GPU is not used at all? I suspect Adobe's code is checking for GPUs and not recognising the M1 GPU, falling back to CPU.

I guess someone with a copy running in a M1 Mac, and knows which test is used, could test it and check for CPU and GPU utilisation.

Yes it is being used but Photoshop isn’t a native Metal app and still leaning on OpenGL/CL to accelerate Adobe’s Mercury Engine. These API versions are old and depreciating on macOS. On Windows the APIs are up to date and so Mercury make better use of a GPU.

But that really doesn’t mean anything outside synthetic tests. The app performs as good as possible in normal real world use.

 

[jeanlain]

I see you must have missed the two paragraphs directly after the chart for which you are concerned:

I had read these, they do not explain that GPU score. They say this score is not surprising. It is, in the wrong direction. The M1 should absolutely trash the intel Iris. This has been shown in countless reviews.

 

[jeanlain]

Yes it is being used

I'm highly suspicious about the use of the GPU in that particular "GPU" test. As I said, if the test was GPU bound, it should not be affected by Rosetta, which processes openGL/CL/Metal code like a native app.
This is not what we see. On the opposite, the negative impact of Rosetta is even worse than for the other tests.
Or there could be some sort of weird bug with the M1 in this test.

 

[quarkysg]

I'm highly suspicious about the use of the GPU in that particular "GPU" test.

If memory serves, Adobe's software has a bug? with their GPU code that thrashes the M1 Mac's SSD with swap, resulting in folks recommending to turn off GPU acceleration? It could also be this issue where the benchmark was executed with the GPU acceleration disabled.

 

[jeanlain]

If memory serves, Adobe's software has a bug? with their GPU code that thrashes the M1 Mac's SSD with swap, resulting in folks recommending to turn off GPU acceleration?

Apparently, there is such issue, at least with LightRoom (even the native version): https://feedback.photoshop.com/conv...7d5e7f9ae444&replyId=60284c2dfd06b771a28ac00c

 

[ArPe]

I'm highly suspicious about the use of the GPU in that particular "GPU" test. As I said, if the test was GPU bound, it should not be affected by Rosetta, which processes openGL/CL/Metal code like a native app.
This is not what we see. On the opposite, the negative impact of Rosetta is even worse than for the other tests.
Or there could be some sort of weird bug with the M1 in this test.

If you want to test if the GPU is being used disable it in advanced graphics PS prefs. That will enable and disable the compute portion and you can run tests to see if there is a difference.

 

[jdb8167]

If you want to test if the GPU is being used disable it in advanced graphics PS prefs. That will enable and disable the compute portion and you can run tests to see if there is a difference.

You should be able to see if GPU is being used with Activity Monitor. There is a GPU History graph.

 

[leman]

I had read these, they do not explain that GPU score. They say this score is not surprising. It is, in the wrong direction. The M1 should absolutely trash the intel Iris. This has been shown in countless reviews.

There are a couple of possibilities: a) Adobe GPU utilization is poor, b) the test is memory-bandwidth bound. Probably combination of both factors.