Quitting Microsoft going to Mac Mini 4, Now - Software?

Glen Barrington said:

I've been a happy ACDSee Ultimate user on Windows for many years, and I consider myself if not an expert in it's use, then certainly, an advanced user. I can get everything I need from it. BUT, the only ACDSee product for the Mac is just the DAM coupled with the Raw developer, no bit mapped editing. And I've heard it has been a bit problematic on the Mac in the past.

Click to expand...

You clearly know ACDSee well and all your non-destructive edits are in it. Also, other stuff such as keywords, star ratings, etc. If the Mac version is only missing the bit mapped editor then I say get the Mac version and then just add a bit mapped editor such as Affinity, GIMP, etc. I am assuming you would like to not lose all the editing and other stuff, which would probably happen if leaving ACDSee behind.

By the way, did you get an M4 Mini 16gb with a 256gb SSD or 512gb SSD?

Jacques Cornell said:

Zeee said:

When Adobe turned on the Neural engine for that brief time my M1 dropped from 85 to 23. Hope they can resolve the shadow issue but they need Apple for that I think and they won't move fast. For now I'll take quality over quantity.

Click to expand...

What? Is Adobe back to not using ANE? Jeez. DxO has sailed through the chip & OS transitions with just one little update that took only a few days to drop.

Click to expand...

All the troubles with the Neural Engine for years is strange. Apple is totally silent about it. Adobe, DXO, and Topaz have all mentioned it, but have mostly been pretty tight lipped about it. Probably in order to not offend the mammoth Apple.

DXO said a year or two ago they had come up with some sort of workaround in order to use the NE so were able to speed up a lot after having to stop using it for awhile. Topaz has not clearly said much other than it causes problems and that they use it -- maybe they have come up with some way to mitigate the NE problems also. Adobe doesn't say much either, other than they couldn't use it, then last year for a short time they could use it, and then reversed course and said they couldn't use it. I don't know why DXO and Topaz were able to come up with a way to mitigate the NE problems, but big, rich Adobe cannot.

DXO said the NE problem appeared in Ventura (October 2022) and was not in Big Sur and Monterey.

The Nvidia NPU (aka Tensor cores) work well though and are very fast. Adobe can use them, but that is just for Windows users.

256 SSD - if I think the Mac will work for both me and my wife, I'll probably give this one to my wife and then buy whatever I think will work best for me.

But I do believe that ANY minimal base Mac must be at least as photographically competent as a $75 dollar refurbed Dell with a 256 SSD and 8GB of RAM running Linux Mint and raw therapee! If $75 pc can outperform or equal a $600 mac I won't go any further.

She just does shopping, browsing, and keeping household accounts in a spreadsheet. Very much a non technical user.

--
Life is an ongoing learning process. You can stop once you are dead.
http://glenbarrington.blogspot.com/
http://glenbarringtonphotos.blogspot.com/
https://www.flickr.com/photos/130525321@N05/

Is this something a newbie will need to worry about? If I'm going to have to do any troubleshooting, i might as well stick with Linux.

Glen Barrington said:

Is this something a newbie will need to worry about? If I'm going to have to do any troubleshooting, i might as well stick with Linux.

Click to expand...

I have been using this MacBook Air for 3 years with Mac OS 13.6.1 Ventura, running DxO Photolab and many other softs and I had practically no issue. I never heard of Neural Engine. I don't know which softs use it.

bakubo said:

Jacques Cornell said:

Zeee said:

When Adobe turned on the Neural engine for that brief time my M1 dropped from 85 to 23. Hope they can resolve the shadow issue but they need Apple for that I think and they won't move fast. For now I'll take quality over quantity.

Click to expand...

What? Is Adobe back to not using ANE? Jeez. DxO has sailed through the chip & OS transitions with just one little update that took only a few days to drop.

Click to expand...

All the troubles with the Neural Engine for years is strange. Apple is totally silent about it. Adobe, DXO, and Topaz have all mentioned it, but have mostly been pretty tight lipped about it. Probably in order to not offend the mammoth Apple.

DXO said a year or two ago they had come up with some sort of workaround in order to use the NE so were able to speed up a lot after having to stop using it for awhile. Topaz has not clearly said much other than it causes problems and that they use it -- maybe they have come up with some way to mitigate the NE problems also. Adobe doesn't say much either, other than they couldn't use it, then last year for a short time they could use it, and then reversed course and said they couldn't use it. I don't know why DXO and Topaz were able to come up with a way to mitigate the NE problems, but big, rich Adobe cannot.

DXO said the NE problem appeared in Ventura (October 2022) and was not in Big Sur and Monterey.

The Nvidia NPU (aka Tensor cores) work well though and are very fast. Adobe can use them, but that is just for Windows users.

Click to expand...

Thanks. I knew it was just not Adobe that was affected and Apple is part of the issue. There were many discussions about that.

Zeee said:

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Zeee said:

Jacques Cornell said:

Zeee said:

These days with M chip (silicone)

Click to expand...

Silicon is what chips are made of. Silicone increases cleavage.

Zeee said:

there is no separate VRAM. AI eat eats VRAM. It is unified memory now so the more RAM the better. I’ll look at 24. 32 would be better to future proof.

You didn’t mention LrC which is fine. I’ll just use Adobe Denoise AI as an example.

I have a 2019 iMac Intel. 64 GB RAM and 8 GB RAM. Adobe Denoise takes 33 seconds

On Tuesday I had a 2020 MacBook Air M1 with 16GB RAM. It took 85 seconds.

Yesterday I traded it in for a 2024 MacBook Air M3 with 24GB RAM. That dropped to 45 seconds.

A fellow on another site has a Studio Ultra M2 with 64GB RAM. 8 seconds.

When I used Photo AI and DeepPrime XD the numbers were pretty close to my iMac and MacBook M1.

Click to expand...

What size RAW files? Apples to apples here? In my experience, DP processing time scales linearly with pixel count.

Click to expand...

32mp crop. Canon R7.

Zeee said:

Zeee said:

Personally I put the money into RAM, processing power and keep the SSD to 512 GB. I use external drives for both the desktop and laptop to store files.

Click to expand...

Interesting. My M1 Max Studio is 50-100% faster than my M1 mini building 1:1 previews in LRC and processing DxO DeepPRIME. I assumed this was due to the additional GPU cores, but maybe it's also related to RAM: 16GB vs. 32GB. OTOH, my M1 Pro 14" MBP with 16GB is just as fast as the Studio. Go figure.

Click to expand...

That is. Thanks

Click to expand...

When Adobe turned on the Neural engine for that brief time my M1 dropped from 85 to 23. Hope they can resolve the shadow issue but they need Apple for that I think and they won't move fast. For now I'll take quality over quantity.

Click to expand...

What? Is Adobe back to not using ANE? Jeez. DxO has sailed through the chip & OS transitions with just one little update that took only a few days to drop.

Click to expand...

Good for them. As I recall when they first started using it they had colour issues they had to work around.

Click to expand...

IIRC, it took mere days for DxO to provide a fix.

Click to expand...

Colour teaks. Easy

Click to expand...

If it's easy, why has Adobe continued to fail?

Zeee said:

Zeee said:

Zeee said:

For colour I guess is pretty easy to find a work around. Let me know when DXO provides the same advanced masking, DAM

Click to expand...

the main thing keeping me using LRC and PhotoLab together.

Zeee said:

, personalized websites, etc. Then I'll still not be interested

Click to expand...

Yeah, well, let me know when LRC has the same advanced, fast, no-DNG-required NR and sophisticated lens profiles. As for websites, I have them already. Adobe's freebies are nowhere near adequate for my business. Then I'll still use PhotoLab for RAW processing and relegate LRC to DAM.

Click to expand...

No DNG already in ACR. Soon to be in LRC. I don’t know about advanced. They were at it for many years and were selling a product that over sharpened and created false details.

Click to expand...

You're comparing vaporware to a last-generation product.

Zeee said:

Adobe came out and was praised for the more natural look. Even in recent threads. DXO had to make new algorithms with DP XD2s.

Click to expand...

And they work fantastically well, take less time to process, and don't require an intermediate DNG.

Zeee said:

So it’s a matter of personal preference and what the applications are.

Zeee said:

Zeee said:

Zeee said:

My basic M1 Max Studio does DeepPRIME XD2s on a 33MP RAW in 9.8 seconds. That's the average for a batch of 44 files (4 each of ISO 100, 200 through 102,400). Numbers get inconsistent when processing just one or a few files.

Click to expand...

DeepPrime XD took 45 seconds on both my iMac and M1. Photo AI. I'm hobby shooter that travels. I have all day.

Click to expand...

Click to expand...

Click to expand...

--
Event professional for 20+ years, travel & landscape enthusiast for 30+, stills-only.
http://jacquescornell.photography
http://happening.photos

Saint 112 said:

Glen Barrington said:

Is this something a newbie will need to worry about? If I'm going to have to do any troubleshooting, i might as well stick with Linux.

Click to expand...

I have been using this MacBook Air for 3 years with Mac OS 13.6.1 Ventura, running DxO Photolab and many other softs and I had practically no issue. I never heard of Neural Engine. I don't know which softs use it.

Click to expand...

Just for a short overview and you need to worry about anything. Apple dropped the Intel chip around 2019/2020 (I believe) for the Silicone chip and it's now called the M chip. Started with M1 and new Macs are starting to get M4.

Neural processing unit (NPU)

• Apple’s Neural Engine is Cupertino’s name for its own neural processing unit (NPU). Such processors are also known as Deep Learning Processors, and handle the algorithms behind AI, augmented reality and machine learning (ML).

Unified Memory so no more separate VRAM and RAM. It's just RAM now and when getting a new device the more the better. More AI will come.

Pre 2019 my older iMac could barely run Topaz Denoise AI. It looked good in Topaz but awful back in LR. Someone on the Topaz forums suggested I crop my file by 50% and try it. That worked. Had nothing to do with Apple, Topaz or Adobe. It didn't have the horsepower (specs) to use it effectively.

When I got my 2019 iMac I had 8GB VRAM and 8GB RAM installed. I installed the other 56GB RAM myself. I'm glad I got the 8GB VRAM. 6 years later it runs Adobe Denoise AI in 33 seconds. It will never see the Neural Engine because it does not have the hardware. I've read about underpowered devices taking up to 10 minutes. I was lucky to squeak Sequoia in but I think that will the last OS upgrade it will support. Next fall I'll look into a mini or whatever and try to get as close to 64 RAM as I can afford.

I used to say AI eats VRAM which I believe is still the case for PC. I know in the forums people were upgrading their video cards. For Mac AI eats RAM. Consider future proofing when getting a new device.

I would have liked a more powerful MacBook Pro but I could not justify the cost for a travel laptop. I really just wanted a bigger screen. My MacBook Air M1 was 13" and my new M3 is 15". It sure makes a difference. If Apple gets their act together hopefully soon we will get the Neural Engine again. I know Apple is not in an hurry.

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Zeee said:

Jacques Cornell said:

Zeee said:

These days with M chip (silicone)

Click to expand...

Silicon is what chips are made of. Silicone increases cleavage.

Jacques Cornell said:

there is no separate VRAM. AI eat eats VRAM. It is unified memory now so the more RAM the better. I’ll look at 24. 32 would be better to future proof.

You didn’t mention LrC which is fine. I’ll just use Adobe Denoise AI as an example.

I have a 2019 iMac Intel. 64 GB RAM and 8 GB RAM. Adobe Denoise takes 33 seconds

On Tuesday I had a 2020 MacBook Air M1 with 16GB RAM. It took 85 seconds.

Yesterday I traded it in for a 2024 MacBook Air M3 with 24GB RAM. That dropped to 45 seconds.

A fellow on another site has a Studio Ultra M2 with 64GB RAM. 8 seconds.

When I used Photo AI and DeepPrime XD the numbers were pretty close to my iMac and MacBook M1.

Click to expand...

What size RAW files? Apples to apples here? In my experience, DP processing time scales linearly with pixel count.

Click to expand...

32mp crop. Canon R7.

Jacques Cornell said:

Jacques Cornell said:

Personally I put the money into RAM, processing power and keep the SSD to 512 GB. I use external drives for both the desktop and laptop to store files.

Click to expand...

Interesting. My M1 Max Studio is 50-100% faster than my M1 mini building 1:1 previews in LRC and processing DxO DeepPRIME. I assumed this was due to the additional GPU cores, but maybe it's also related to RAM: 16GB vs. 32GB. OTOH, my M1 Pro 14" MBP with 16GB is just as fast as the Studio. Go figure.

Click to expand...

That is. Thanks

Click to expand...

When Adobe turned on the Neural engine for that brief time my M1 dropped from 85 to 23. Hope they can resolve the shadow issue but they need Apple for that I think and they won't move fast. For now I'll take quality over quantity.

Click to expand...

What? Is Adobe back to not using ANE? Jeez. DxO has sailed through the chip & OS transitions with just one little update that took only a few days to drop.

Click to expand...

Good for them. As I recall when they first started using it they had colour issues they had to work around.

Click to expand...

IIRC, it took mere days for DxO to provide a fix.

Click to expand...

Colour teaks. Easy

Click to expand...

If it's easy, why has Adobe continued to fail?

Jacques Cornell said:

Jacques Cornell said:

Jacques Cornell said:

For colour I guess is pretty easy to find a work around. Let me know when DXO provides the same advanced masking, DAM

Click to expand...

the main thing keeping me using LRC and PhotoLab together.

Jacques Cornell said:

, personalized websites, etc. Then I'll still not be interested

Click to expand...

Yeah, well, let me know when LRC has the same advanced, fast, no-DNG-required NR and sophisticated lens profiles. As for websites, I have them already. Adobe's freebies are nowhere near adequate for my business. Then I'll still use PhotoLab for RAW processing and relegate LRC to DAM.

Click to expand...

No DNG already in ACR. Soon to be in LRC. I don’t know about advanced. They were at it for many years and were selling a product that over sharpened and created false details.

Click to expand...

You're comparing vaporware to a last-generation product.

Click to expand...

Whatever. Presenters like Jan Wegener said Adobe was better at first release. Not bad for first release compared to a company that has been at it for a decade. Then he gave DP XD2s the nod. I have links to both videos. You and I both know it will always be leapfrog.

Jacques Cornell said:

Jacques Cornell said:

Adobe came out and was praised for the more natural look. Even in recent threads. DXO had to make new algorithms with DP XD2s.

Click to expand...

And they work fantastically well, take less time to process, and don't require an intermediate DNG.

Click to expand...

I'm happy for you and all the DXO owners.

Jacques Cornell said:

Jacques Cornell said:

So it’s a matter of personal preference and what the applications are.

Jacques Cornell said:

Jacques Cornell said:

Jacques Cornell said:

My basic M1 Max Studio does DeepPRIME XD2s on a 33MP RAW in 9.8 seconds. That's the average for a batch of 44 files (4 each of ISO 100, 200 through 102,400). Numbers get inconsistent when processing just one or a few files.

Click to expand...

DeepPrime XD took 45 seconds on both my iMac and M1. Photo AI. I'm hobby shooter that travels. I have all day.

Click to expand...

Click to expand...

Click to expand...

Click to expand...

--
I roll with pleasing colour

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Zeee said:

Jacques Cornell said:

Zeee said:

These days with M chip (silicone)

Click to expand...

Silicon is what chips are made of. Silicone increases cleavage.

Jacques Cornell said:

there is no separate VRAM. AI eat eats VRAM. It is unified memory now so the more RAM the better. I’ll look at 24. 32 would be better to future proof.

You didn’t mention LrC which is fine. I’ll just use Adobe Denoise AI as an example.

I have a 2019 iMac Intel. 64 GB RAM and 8 GB RAM. Adobe Denoise takes 33 seconds

On Tuesday I had a 2020 MacBook Air M1 with 16GB RAM. It took 85 seconds.

Yesterday I traded it in for a 2024 MacBook Air M3 with 24GB RAM. That dropped to 45 seconds.

A fellow on another site has a Studio Ultra M2 with 64GB RAM. 8 seconds.

When I used Photo AI and DeepPrime XD the numbers were pretty close to my iMac and MacBook M1.

Click to expand...

What size RAW files? Apples to apples here? In my experience, DP processing time scales linearly with pixel count.

Click to expand...

32mp crop. Canon R7.

Jacques Cornell said:

Jacques Cornell said:

Personally I put the money into RAM, processing power and keep the SSD to 512 GB. I use external drives for both the desktop and laptop to store files.

Click to expand...

Interesting. My M1 Max Studio is 50-100% faster than my M1 mini building 1:1 previews in LRC and processing DxO DeepPRIME. I assumed this was due to the additional GPU cores, but maybe it's also related to RAM: 16GB vs. 32GB. OTOH, my M1 Pro 14" MBP with 16GB is just as fast as the Studio. Go figure.

Click to expand...

That is. Thanks

Click to expand...

When Adobe turned on the Neural engine for that brief time my M1 dropped from 85 to 23. Hope they can resolve the shadow issue but they need Apple for that I think and they won't move fast. For now I'll take quality over quantity.

Click to expand...

What? Is Adobe back to not using ANE? Jeez. DxO has sailed through the chip & OS transitions with just one little update that took only a few days to drop.

Click to expand...

Good for them. As I recall when they first started using it they had colour issues they had to work around.

Click to expand...

IIRC, it took mere days for DxO to provide a fix.

Click to expand...

Colour teaks. Easy

Click to expand...

If it's easy, why has Adobe continued to fail?

Click to expand...

bukobo explained it. Read the post.

Jacques Cornell said:

Jacques Cornell said:

Jacques Cornell said:

Jacques Cornell said:

For colour I guess is pretty easy to find a work around. Let me know when DXO provides the same advanced masking, DAM

Click to expand...

the main thing keeping me using LRC and PhotoLab together.

Jacques Cornell said:

, personalized websites, etc. Then I'll still not be interested

Click to expand...

Yeah, well, let me know when LRC has the same advanced, fast, no-DNG-required NR and sophisticated lens profiles. As for websites, I have them already. Adobe's freebies are nowhere near adequate for my business. Then I'll still use PhotoLab for RAW processing and relegate LRC to DAM.

Click to expand...

No DNG already in ACR. Soon to be in LRC. I don’t know about advanced. They were at it for many years and were selling a product that over sharpened and created false details.

Click to expand...

You're comparing vaporware to a last-generation product.

Jacques Cornell said:

Adobe came out and was praised for the more natural look. Even in recent threads. DXO had to make new algorithms with DP XD2s.

Click to expand...

And they work fantastically well, take less time to process, and don't require an intermediate DNG.

Jacques Cornell said:

So it’s a matter of personal preference and what the applications are.

Jacques Cornell said:

Jacques Cornell said:

Jacques Cornell said:

My basic M1 Max Studio does DeepPRIME XD2s on a 33MP RAW in 9.8 seconds. That's the average for a batch of 44 files (4 each of ISO 100, 200 through 102,400). Numbers get inconsistent when processing just one or a few files.

Click to expand...

DeepPrime XD took 45 seconds on both my iMac and M1. Photo AI. I'm hobby shooter that travels. I have all day.

Click to expand...

Click to expand...

Click to expand...

Click to expand...

--
I roll with pleasing colour

Zeee said:

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Zeee said:

Jacques Cornell said:

Zeee said:

These days with M chip (silicone)

Click to expand...

Silicon is what chips are made of. Silicone increases cleavage.

Zeee said:

there is no separate VRAM. AI eat eats VRAM. It is unified memory now so the more RAM the better. I’ll look at 24. 32 would be better to future proof.

You didn’t mention LrC which is fine. I’ll just use Adobe Denoise AI as an example.

I have a 2019 iMac Intel. 64 GB RAM and 8 GB RAM. Adobe Denoise takes 33 seconds

On Tuesday I had a 2020 MacBook Air M1 with 16GB RAM. It took 85 seconds.

Yesterday I traded it in for a 2024 MacBook Air M3 with 24GB RAM. That dropped to 45 seconds.

A fellow on another site has a Studio Ultra M2 with 64GB RAM. 8 seconds.

When I used Photo AI and DeepPrime XD the numbers were pretty close to my iMac and MacBook M1.

Click to expand...

What size RAW files? Apples to apples here? In my experience, DP processing time scales linearly with pixel count.

Click to expand...

32mp crop. Canon R7.

Zeee said:

Zeee said:

Personally I put the money into RAM, processing power and keep the SSD to 512 GB. I use external drives for both the desktop and laptop to store files.

Click to expand...

Interesting. My M1 Max Studio is 50-100% faster than my M1 mini building 1:1 previews in LRC and processing DxO DeepPRIME. I assumed this was due to the additional GPU cores, but maybe it's also related to RAM: 16GB vs. 32GB. OTOH, my M1 Pro 14" MBP with 16GB is just as fast as the Studio. Go figure.

Click to expand...

That is. Thanks

Click to expand...

When Adobe turned on the Neural engine for that brief time my M1 dropped from 85 to 23. Hope they can resolve the shadow issue but they need Apple for that I think and they won't move fast. For now I'll take quality over quantity.

Click to expand...

What? Is Adobe back to not using ANE? Jeez. DxO has sailed through the chip & OS transitions with just one little update that took only a few days to drop.

Click to expand...

Good for them. As I recall when they first started using it they had colour issues they had to work around.

Click to expand...

IIRC, it took mere days for DxO to provide a fix.

Click to expand...

Colour teaks. Easy

Click to expand...

If it's easy, why has Adobe continued to fail?

Click to expand...

bukobo explained it. Read the post.

Click to expand...

Come to think about failure maybe you should think about a company that was 10 years into the game and was outperformed by another's first release. Lit a fire under DXO. Maybe DXO was the cash cow just adding more digits to their NR upgrades, not Adobe as many have said would wind up doing.

Sorry. I have said nothing negative here or in a long while about DXO. I've even said DXO did a good job with the latest release. You drew first blood in this thread.

Zeee said:

Zeee said:

Zeee said:

Zeee said:

Zeee said:

For colour I guess is pretty easy to find a work around. Let me know when DXO provides the same advanced masking, DAM

Click to expand...

the main thing keeping me using LRC and PhotoLab together.

Zeee said:

, personalized websites, etc. Then I'll still not be interested

Click to expand...

Yeah, well, let me know when LRC has the same advanced, fast, no-DNG-required NR and sophisticated lens profiles. As for websites, I have them already. Adobe's freebies are nowhere near adequate for my business. Then I'll still use PhotoLab for RAW processing and relegate LRC to DAM.

Click to expand...

No DNG already in ACR. Soon to be in LRC. I don’t know about advanced. They were at it for many years and were selling a product that over sharpened and created false details.

Click to expand...

You're comparing vaporware to a last-generation product.

Zeee said:

Adobe came out and was praised for the more natural look. Even in recent threads. DXO had to make new algorithms with DP XD2s.

Click to expand...

And they work fantastically well, take less time to process, and don't require an intermediate DNG.

Zeee said:

So it’s a matter of personal preference and what the applications are.

Zeee said:

Zeee said:

Zeee said:

My basic M1 Max Studio does DeepPRIME XD2s on a 33MP RAW in 9.8 seconds. That's the average for a batch of 44 files (4 each of ISO 100, 200 through 102,400). Numbers get inconsistent when processing just one or a few files.

Click to expand...

DeepPrime XD took 45 seconds on both my iMac and M1. Photo AI. I'm hobby shooter that travels. I have all day.

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Click to expand...

--
I roll with pleasing colour

Zeee said:

Saint 112 said:

Glen Barrington said:

Is this something a newbie will need to worry about? If I'm going to have to do any troubleshooting, i might as well stick with Linux.

Click to expand...

I have been using this MacBook Air for 3 years with Mac OS 13.6.1 Ventura, running DxO Photolab and many other softs and I had practically no issue. I never heard of Neural Engine. I don't know which softs use it.

Click to expand...

Just for a short overview and you need to worry about anything. Apple dropped the Intel chip around 2019/2020 (I believe) for the Silicone chip and it's now called the M chip. Started with M1 and new Macs are starting to get M4.

Neural processing unit (NPU)

• Apple’s Neural Engine is Cupertino’s name for its own neural processing unit (NPU). Such processors are also known as Deep Learning Processors, and handle the algorithms behind AI, augmented reality and machine learning (ML).

Unified Memory so no more separate VRAM and RAM. It's just RAM now and when getting a new device the more the better. More AI will come.

Pre 2019 my older iMac could barely run Topaz Denoise AI. It looked good in Topaz but awful back in LR. Someone on the Topaz forums suggested I crop my file by 50% and try it. That worked. Had nothing to do with Apple, Topaz or Adobe. It didn't have the horsepower (specs) to use it effectively.

When I got my 2019 iMac I had 8GB VRAM and 8GB RAM installed. I installed the other 56GB RAM myself. I'm glad I got the 8GB VRAM. 6 years later it runs Adobe Denoise AI in 33 seconds. It will never see the Neural Engine because it does not have the hardware. I've read about underpowered devices taking up to 10 minutes. I was lucky to squeak Sequoia in but I think that will the last OS upgrade it will support. Next fall I'll look into a mini or whatever and try to get as close to 64 RAM as I can afford.

I used to say AI eats VRAM which I believe is still the case for PC. I know in the forums people were upgrading their video cards. For Mac AI eats RAM. Consider future proofing when getting a new device.

Click to expand...

I should point out that bakubo said in another thread that the person that is getting 8 seconds on his studio ultra with 64GB RAM has 60 cores. That is without the Neural Engine working. Another consideration I will add to my desktop replacement.

Zeee said:

I would have liked a more powerful MacBook Pro but I could not justify the cost for a travel laptop. I really just wanted a bigger screen. My MacBook Air M1 was 13" and my new M3 is 15". It sure makes a difference. If Apple gets their act together hopefully soon we will get the Neural Engine again. I know Apple is not in an hurry.

Click to expand...

Zeee said:

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Zeee said:

Jacques Cornell said:

Zeee said:

These days with M chip (silicone)

Click to expand...

Silicon is what chips are made of. Silicone increases cleavage.

Zeee said:

there is no separate VRAM. AI eat eats VRAM. It is unified memory now so the more RAM the better. I’ll look at 24. 32 would be better to future proof.

You didn’t mention LrC which is fine. I’ll just use Adobe Denoise AI as an example.

I have a 2019 iMac Intel. 64 GB RAM and 8 GB RAM. Adobe Denoise takes 33 seconds

On Tuesday I had a 2020 MacBook Air M1 with 16GB RAM. It took 85 seconds.

Yesterday I traded it in for a 2024 MacBook Air M3 with 24GB RAM. That dropped to 45 seconds.

A fellow on another site has a Studio Ultra M2 with 64GB RAM. 8 seconds.

When I used Photo AI and DeepPrime XD the numbers were pretty close to my iMac and MacBook M1.

Click to expand...

What size RAW files? Apples to apples here? In my experience, DP processing time scales linearly with pixel count.

Click to expand...

32mp crop. Canon R7.

Zeee said:

Zeee said:

Personally I put the money into RAM, processing power and keep the SSD to 512 GB. I use external drives for both the desktop and laptop to store files.

Click to expand...

Interesting. My M1 Max Studio is 50-100% faster than my M1 mini building 1:1 previews in LRC and processing DxO DeepPRIME. I assumed this was due to the additional GPU cores, but maybe it's also related to RAM: 16GB vs. 32GB. OTOH, my M1 Pro 14" MBP with 16GB is just as fast as the Studio. Go figure.

Click to expand...

That is. Thanks

Click to expand...

When Adobe turned on the Neural engine for that brief time my M1 dropped from 85 to 23. Hope they can resolve the shadow issue but they need Apple for that I think and they won't move fast. For now I'll take quality over quantity.

Click to expand...

What? Is Adobe back to not using ANE? Jeez. DxO has sailed through the chip & OS transitions with just one little update that took only a few days to drop.

Click to expand...

Good for them. As I recall when they first started using it they had colour issues they had to work around.

Click to expand...

IIRC, it took mere days for DxO to provide a fix.

Click to expand...

Colour teaks. Easy

Click to expand...

If it's easy, why has Adobe continued to fail?

Zeee said:

Zeee said:

Zeee said:

For colour I guess is pretty easy to find a work around. Let me know when DXO provides the same advanced masking, DAM

Click to expand...

the main thing keeping me using LRC and PhotoLab together.

Zeee said:

, personalized websites, etc. Then I'll still not be interested

Click to expand...

Yeah, well, let me know when LRC has the same advanced, fast, no-DNG-required NR and sophisticated lens profiles. As for websites, I have them already. Adobe's freebies are nowhere near adequate for my business. Then I'll still use PhotoLab for RAW processing and relegate LRC to DAM.

Click to expand...

No DNG already in ACR. Soon to be in LRC. I don’t know about advanced. They were at it for many years and were selling a product that over sharpened and created false details.

Click to expand...

You're comparing vaporware to a last-generation product.

Click to expand...

Whatever. Presenters like Jan Wegener said Adobe was better at first release.

Click to expand...

I found it to be a wash in terms of quality. Except that Adobe's processing took much longer, and the whole DNG thing...

Zeee said:

Not bad for first release compared to a company that has been at it for a decade.

Click to expand...

But still not compelling. Also, how long has Adobe been making a RAW processing app with noise reduction built-in? Certainly as long as DxO.

Zeee said:

Then he gave DP XD2s the nod.

Click to expand...

As I do.

Zeee said:

I have links to both videos. You and I both know it will always be leapfrog.

Click to expand...

Prolly. But, Adobe hasn't leaped over the first frog yet. Pun intended. And, despite having years to catch up, Adobe never was able to match Aperture's UI for efficiency and user-friendliness. I mean, the whole modal interface and keyboard shortcuts that change when you switch modes is a classic example of UI dunderheadedness comparable to Microsoft putting "Shut Down" under the "Start" menu.

LRC is a pretty good Swiss Army knife, but I wouldn't choose it for performing surgery.

Zeee said:

Zeee said:

Zeee said:

Adobe came out and was praised for the more natural look. Even in recent threads. DXO had to make new algorithms with DP XD2s.

Click to expand...

And they work fantastically well, take less time to process, and don't require an intermediate DNG.

Click to expand...

I'm happy for you and all the DXO owners.

Zeee said:

Zeee said:

So it’s a matter of personal preference and what the applications are.

Zeee said:

Zeee said:

Zeee said:

My basic M1 Max Studio does DeepPRIME XD2s on a 33MP RAW in 9.8 seconds. That's the average for a batch of 44 files (4 each of ISO 100, 200 through 102,400). Numbers get inconsistent when processing just one or a few files.

Click to expand...

DeepPrime XD took 45 seconds on both my iMac and M1. Photo AI. I'm hobby shooter that travels. I have all day.

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Click to expand...

As an event shooter, I process hundreds or thousands of images per event on 48-hour deadlines. Adobe's whole intermediate-DNG workflow is a complete non-starter, and the fact that they went for it smells of desperation.

Zeee said:

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Zeee said:

Jacques Cornell said:

Zeee said:

These days with M chip (silicone)

Click to expand...

Silicon is what chips are made of. Silicone increases cleavage.

Zeee said:

there is no separate VRAM. AI eat eats VRAM. It is unified memory now so the more RAM the better. I’ll look at 24. 32 would be better to future proof.

You didn’t mention LrC which is fine. I’ll just use Adobe Denoise AI as an example.

I have a 2019 iMac Intel. 64 GB RAM and 8 GB RAM. Adobe Denoise takes 33 seconds

On Tuesday I had a 2020 MacBook Air M1 with 16GB RAM. It took 85 seconds.

Yesterday I traded it in for a 2024 MacBook Air M3 with 24GB RAM. That dropped to 45 seconds.

A fellow on another site has a Studio Ultra M2 with 64GB RAM. 8 seconds.

When I used Photo AI and DeepPrime XD the numbers were pretty close to my iMac and MacBook M1.

Click to expand...

What size RAW files? Apples to apples here? In my experience, DP processing time scales linearly with pixel count.

Click to expand...

32mp crop. Canon R7.

Zeee said:

Zeee said:

Personally I put the money into RAM, processing power and keep the SSD to 512 GB. I use external drives for both the desktop and laptop to store files.

Click to expand...

Interesting. My M1 Max Studio is 50-100% faster than my M1 mini building 1:1 previews in LRC and processing DxO DeepPRIME. I assumed this was due to the additional GPU cores, but maybe it's also related to RAM: 16GB vs. 32GB. OTOH, my M1 Pro 14" MBP with 16GB is just as fast as the Studio. Go figure.

Click to expand...

That is. Thanks

Click to expand...

When Adobe turned on the Neural engine for that brief time my M1 dropped from 85 to 23. Hope they can resolve the shadow issue but they need Apple for that I think and they won't move fast. For now I'll take quality over quantity.

Click to expand...

What? Is Adobe back to not using ANE? Jeez. DxO has sailed through the chip & OS transitions with just one little update that took only a few days to drop.

Click to expand...

Good for them. As I recall when they first started using it they had colour issues they had to work around.

Click to expand...

IIRC, it took mere days for DxO to provide a fix.

Click to expand...

Colour teaks. Easy

Click to expand...

If it's easy, why has Adobe continued to fail?

Click to expand...

bukobo explained it. Read the post.

Click to expand...

Bakubo explained that Adobe has continued to fail, not why.

Zeee said:

Zeee said:

Zeee said:

Zeee said:

Zeee said:

For colour I guess is pretty easy to find a work around. Let me know when DXO provides the same advanced masking, DAM

Click to expand...

the main thing keeping me using LRC and PhotoLab together.

Zeee said:

, personalized websites, etc. Then I'll still not be interested

Click to expand...

Yeah, well, let me know when LRC has the same advanced, fast, no-DNG-required NR and sophisticated lens profiles. As for websites, I have them already. Adobe's freebies are nowhere near adequate for my business. Then I'll still use PhotoLab for RAW processing and relegate LRC to DAM.

Click to expand...

No DNG already in ACR. Soon to be in LRC. I don’t know about advanced. They were at it for many years and were selling a product that over sharpened and created false details.

Click to expand...

You're comparing vaporware to a last-generation product.

Zeee said:

Adobe came out and was praised for the more natural look. Even in recent threads. DXO had to make new algorithms with DP XD2s.

Click to expand...

And they work fantastically well, take less time to process, and don't require an intermediate DNG.

Zeee said:

So it’s a matter of personal preference and what the applications are.

Zeee said:

Zeee said:

Zeee said:

My basic M1 Max Studio does DeepPRIME XD2s on a 33MP RAW in 9.8 seconds. That's the average for a batch of 44 files (4 each of ISO 100, 200 through 102,400). Numbers get inconsistent when processing just one or a few files.

Click to expand...

DeepPrime XD took 45 seconds on both my iMac and M1. Photo AI. I'm hobby shooter that travels. I have all day.

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Zeee said:

Zeee said:

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Jacques Cornell said:

Zeee said:

Zeee said:

Jacques Cornell said:

Zeee said:

These days with M chip (silicone)

Click to expand...

Silicon is what chips are made of. Silicone increases cleavage.

Zeee said:

there is no separate VRAM. AI eat eats VRAM. It is unified memory now so the more RAM the better. I’ll look at 24. 32 would be better to future proof.

You didn’t mention LrC which is fine. I’ll just use Adobe Denoise AI as an example.

I have a 2019 iMac Intel. 64 GB RAM and 8 GB RAM. Adobe Denoise takes 33 seconds

On Tuesday I had a 2020 MacBook Air M1 with 16GB RAM. It took 85 seconds.

Yesterday I traded it in for a 2024 MacBook Air M3 with 24GB RAM. That dropped to 45 seconds.

A fellow on another site has a Studio Ultra M2 with 64GB RAM. 8 seconds.

When I used Photo AI and DeepPrime XD the numbers were pretty close to my iMac and MacBook M1.

Click to expand...

What size RAW files? Apples to apples here? In my experience, DP processing time scales linearly with pixel count.

Click to expand...

32mp crop. Canon R7.

Zeee said:

Zeee said:

Personally I put the money into RAM, processing power and keep the SSD to 512 GB. I use external drives for both the desktop and laptop to store files.

Click to expand...

Interesting. My M1 Max Studio is 50-100% faster than my M1 mini building 1:1 previews in LRC and processing DxO DeepPRIME. I assumed this was due to the additional GPU cores, but maybe it's also related to RAM: 16GB vs. 32GB. OTOH, my M1 Pro 14" MBP with 16GB is just as fast as the Studio. Go figure.

Click to expand...

That is. Thanks

Click to expand...

When Adobe turned on the Neural engine for that brief time my M1 dropped from 85 to 23. Hope they can resolve the shadow issue but they need Apple for that I think and they won't move fast. For now I'll take quality over quantity.

Click to expand...

What? Is Adobe back to not using ANE? Jeez. DxO has sailed through the chip & OS transitions with just one little update that took only a few days to drop.

Click to expand...

Good for them. As I recall when they first started using it they had colour issues they had to work around.

Click to expand...

IIRC, it took mere days for DxO to provide a fix.

Click to expand...

Colour teaks. Easy

Click to expand...

If it's easy, why has Adobe continued to fail?

Click to expand...

bukobo explained it. Read the post.

Click to expand...

Come to think about failure maybe you should think about a company that was 10 years into the game and was outperformed by another's first release. Lit a fire under DXO. Maybe DXO was the cash cow just adding more digits to their NR upgrades, not Adobe as many have said would wind up doing.

Sorry. I have said nothing negative here or in a long while about DXO. I've even said DXO did a good job with the latest release. You drew first blood in this thread.

Click to expand...

What? By saying how Adobe's non-non-destructive DNG-burdened workflow was unusable for my high-volume event photo processing? Gee, sorry. Didn't know you cared.

Zeee said:

Zeee said:

Zeee said:

Zeee said:

Zeee said:

Zeee said:

For colour I guess is pretty easy to find a work around. Let me know when DXO provides the same advanced masking, DAM

Click to expand...

the main thing keeping me using LRC and PhotoLab together.

Zeee said:

, personalized websites, etc. Then I'll still not be interested

Click to expand...

Yeah, well, let me know when LRC has the same advanced, fast, no-DNG-required NR and sophisticated lens profiles. As for websites, I have them already. Adobe's freebies are nowhere near adequate for my business. Then I'll still use PhotoLab for RAW processing and relegate LRC to DAM.

Click to expand...

No DNG already in ACR. Soon to be in LRC. I don’t know about advanced. They were at it for many years and were selling a product that over sharpened and created false details.

Click to expand...

You're comparing vaporware to a last-generation product.

Zeee said:

Adobe came out and was praised for the more natural look. Even in recent threads. DXO had to make new algorithms with DP XD2s.

Click to expand...

And they work fantastically well, take less time to process, and don't require an intermediate DNG.

Zeee said:

So it’s a matter of personal preference and what the applications are.

Zeee said:

Zeee said:

Zeee said:

My basic M1 Max Studio does DeepPRIME XD2s on a 33MP RAW in 9.8 seconds. That's the average for a batch of 44 files (4 each of ISO 100, 200 through 102,400). Numbers get inconsistent when processing just one or a few files.

Click to expand...

DeepPrime XD took 45 seconds on both my iMac and M1. Photo AI. I'm hobby shooter that travels. I have all day.

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Zeee said:

Zeee said:

Saint 112 said:

Glen Barrington said:

Is this something a newbie will need to worry about? If I'm going to have to do any troubleshooting, i might as well stick with Linux.

Click to expand...

I have been using this MacBook Air for 3 years with Mac OS 13.6.1 Ventura, running DxO Photolab and many other softs and I had practically no issue. I never heard of Neural Engine. I don't know which softs use it.

Click to expand...

Just for a short overview and you need to worry about anything. Apple dropped the Intel chip around 2019/2020 (I believe) for the Silicone chip and it's now called the M chip. Started with M1 and new Macs are starting to get M4.

Neural processing unit (NPU)

• Apple’s Neural Engine is Cupertino’s name for its own neural processing unit (NPU). Such processors are also known as Deep Learning Processors, and handle the algorithms behind AI, augmented reality and machine learning (ML).

Unified Memory so no more separate VRAM and RAM. It's just RAM now and when getting a new device the more the better. More AI will come.

Pre 2019 my older iMac could barely run Topaz Denoise AI. It looked good in Topaz but awful back in LR. Someone on the Topaz forums suggested I crop my file by 50% and try it. That worked. Had nothing to do with Apple, Topaz or Adobe. It didn't have the horsepower (specs) to use it effectively.

When I got my 2019 iMac I had 8GB VRAM and 8GB RAM installed. I installed the other 56GB RAM myself. I'm glad I got the 8GB VRAM. 6 years later it runs Adobe Denoise AI in 33 seconds. It will never see the Neural Engine because it does not have the hardware. I've read about underpowered devices taking up to 10 minutes. I was lucky to squeak Sequoia in but I think that will the last OS upgrade it will support. Next fall I'll look into a mini or whatever and try to get as close to 64 RAM as I can afford.

I used to say AI eats VRAM which I believe is still the case for PC. I know in the forums people were upgrading their video cards. For Mac AI eats RAM. Consider future proofing when getting a new device.

Click to expand...

I should point out that bakubo said in another thread that the person that is getting 8 seconds on his studio ultra with 64GB RAM has 60 cores. That is without the Neural Engine working. Another consideration I will add to my desktop replacement.

Click to expand...

It would be interesting to compare performance of DxO DeepPRIME on a $599 M4 Mac mini, with its 38-teraflop Neural Engine and 10 GPU cores to Adobe AI Denoise on the $3000+ Ultra with its two 18-teraflop Neural Engines and 60 GPU cores. Betcha the latter won't be 5x faster.

Zeee said:

Zeee said:

I would have liked a more powerful MacBook Pro but I could not justify the cost for a travel laptop. I really just wanted a bigger screen. My MacBook Air M1 was 13" and my new M3 is 15". It sure makes a difference. If Apple gets their act together hopefully soon we will get the Neural Engine again. I know Apple is not in an hurry.

Click to expand...

Click to expand...

Zeee said:

Saint 112 said:

Glen Barrington said:

Is this something a newbie will need to worry about? If I'm going to have to do any troubleshooting, i might as well stick with Linux.

Click to expand...

I have been using this MacBook Air for 3 years with Mac OS 13.6.1 Ventura, running DxO Photolab and many other softs and I had practically no issue. I never heard of Neural Engine. I don't know which softs use it.

Click to expand...

Just for a short overview and you need to worry about anything. Apple dropped the Intel chip around 2019/2020 (I believe) for the Silicone chip and it's now called the M chip. Started with M1 and new Macs are starting to get M4.

Neural processing unit (NPU)

• Apple’s Neural Engine is Cupertino’s name for its own neural processing unit (NPU). Such processors are also known as Deep Learning Processors, and handle the algorithms behind AI, augmented reality and machine learning (ML).

Unified Memory so no more separate VRAM and RAM. It's just RAM now and when getting a new device the more the better. More AI will come.

Pre 2019 my older iMac could barely run Topaz Denoise AI. It looked good in Topaz but awful back in LR. Someone on the Topaz forums suggested I crop my file by 50% and try it. That worked. Had nothing to do with Apple, Topaz or Adobe. It didn't have the horsepower (specs) to use it effectively.

When I got my 2019 iMac I had 8GB VRAM and 8GB RAM installed. I installed the other 56GB RAM myself. I'm glad I got the 8GB VRAM. 6 years later it runs Adobe Denoise AI in 33 seconds. It will never see the Neural Engine because it does not have the hardware. I've read about underpowered devices taking up to 10 minutes. I was lucky to squeak Sequoia in but I think that will the last OS upgrade it will support. Next fall I'll look into a mini or whatever and try to get as close to 64 RAM as I can afford.

I used to say AI eats VRAM which I believe is still the case for PC. I know in the forums people were upgrading their video cards. For Mac AI eats RAM. Consider future proofing when getting a new device.

I would have liked a more powerful MacBook Pro but I could not justify the cost for a travel laptop. I really just wanted a bigger screen. My MacBook Air M1 was 13" and my new M3 is 15". It sure makes a difference. If Apple gets their act together hopefully soon we will get the Neural Engine again. I know Apple is not in an hurry.

Click to expand...

Uh, you know the Neural Engine is not gone, right? It's just that Adobe hasn't figured out how to make it work, whereas DxO has.

Glen Barrington said:

Is this something a newbie will need to worry about? If I'm going to have to do any troubleshooting, i might as well stick with Linux.

Click to expand...

No, you don't need to worry about it as long as you don't use LrC/Ps -- and even then it is just the AI denoise which is slow, otherwise LrC/Ps run fine. And the LrC/Ps AI denoise still works on Apple, just slower because it cannot use the NE. Although it seems from what DXO, Topaz, and Adobe has said that it is a Neural Engine problem there is very little info about it. All we know is what I wrote already from the comments that the 3 companies have made in the past. It is possible that other companies such as On1, etc. that have AI denoising also have problems, but I have not followed them.

If you don't use Lightroom and Photoshop and their AI denoise then it won't matter to you. As I already wrote, DXO said the problem showed up when Ventura came out in October 2022 so for awhile they told their customers to turn off the Neural Engine and only use the GPU. Then later they came out with a new release and said they had come up with a workaround (probably some sort of kludge) to allow the very fast Neural Engine to be used again even on Ventura and releases after it. And even with the workaround it is pretty widely said that DXO gets the best results. Topaz has said less about all this, just that they had discovered problems with it and separately they have said they use it so probably they have also found a way to mitigate the problems.

Adobe first said they were using the Nvidia Tensor cores and Apple Neural Engine in their AI denoise when it first came out. Then not long after they said they had to disable the NE code and only use the GPU because of quality problems with the NE. Then last year Adobe announced they could finally use the NE and it really speeded up denoise a whole lot, as people on these forums have confirmed -- much faster. Then after just a few months Adobe again said they had to disable the NE code because they found there were still problems with the NE. That is how things stand today.

It is unknown why small, much less rich companies such as DXO and Topaz are able to workaround the NE problems and did it long ago in a short time, but big, rich Adobe for years still can't or won't.

Jacques Cornell said:

Zeee said:

I should point out that bakubo said in another thread that the person that is getting 8 seconds on his studio ultra with 64GB RAM has 60 cores. That is without the Neural Engine working. Another consideration I will add to my desktop replacement.

Click to expand...

It would be interesting to compare performance of DxO DeepPRIME on a $599 M4 Mac mini, with its 38-teraflop Neural Engine and 10 GPU cores to Adobe AI Denoise on the $3000+ Ultra with its two 18-teraflop Neural Engines and 60 GPU cores. Betcha the latter won't be 5x faster.

Click to expand...

Yes, that would be interesting. Does anyone have a link to the raw file that was used? I could try it on my M2 Pro 12/19 MBP 32gb to get the timing in the current LrC. Also, I could try it on the current Topaz Photo AI and Denoise AI.

By the way, in the U.S. the price of an M2 Ultra 24/60/32 Studio 64gb/1tb (base model of M2 Ultra Studio) is $3999, not $3000. The M2 Ultra 24/76/32 Studio 64gb/1tb is $4999. As you say, the M4 10/10/16 Mini 16gb/256gb is $599 (although I have seen it for less sometimes on Amazon, etc.).

Actually, it would be fun and informative to have a thread with a link to a few high ISO raw files from several different brand cameras and we could all compare the timing using various Macs and various software. And then everyone report back their timing results.

Hi

I am a former ACDSee user, I also have extensive experience with CaptureOne, Adobe Lightroom and Photoshop and now DXO Photolab is my main tool.

I have 2 Macs: an M1 Mini with 16GB RAM and 512GB SSD and an M2 Pro MacBook Pro, also with 16GB RAM and 521GB SSD.

Firstly, 16GB RAM is absolutely enough for photography. I invite you to watch this video , my experience is identical, I never notice any slowdown. I have no idea how much swapping is going on but the Mac reads and writes from SSD at 3GB per second, making the issue irrelevant in my opinion.

Regarding disk space it is all a matter of how you organise your storage. I have a NAS and I put as much as I can on it. My 512GB SSD has a usable capacity of 494GB, of which 406 is available.

For the CPU, I find the M2 Pro feels marginally faster than the M1. The M2 Pro slaughters the M1 in benchmarks, but in real use I cannot say it makes a huge difference for me. I use both very happily. It might come from the fact that DXO does use the Neural Engine, which maybe lessens the perceived difference between the two.

To put things in perspective, when I got he M1 Mac mini, I had a PC with the following components: AMD Ryzen 8 cores 16 threads, 32GB RAM, 1TB MVME SSD and an AMD Radeon Pro graphic card. In Lightroom and Photoshop, the M1 was way faster. Not a subtle difference like between the M1 and M2, it really felt like a new generation of hardware.

I would expect a current Mac Mini M4 to be a blast.

Software-wise, Lightroom and Photoshop are the best in my opinion, but I have decided to stop using the subscription, mainly because I don't want to have to continue spending if I ever go through a rough patch.

Of the alternatives, Capture One and DXO are both of very high quality, with different strengths. In the end, I went with DXO and am very happy with it. I probably would be happy if I had chosen Capture One too.

There is the fact that Capture One pushes more towards the subscription model that DXO, who have no subscription in sight.

Capture One subscription model is more honest than Adobe's. It gives you a progressing discount on a perpetual license with each year you keep the subscription. After 5 years you receive a free perpetual license that you can use as a backup plan if and when you want to stop using the subscription.

Good luck with the decisions

--
Stéphane