MacBook Pro M2 Max or M3 Pro for Affinity Photo 2?

[edwardsson]

I do some photography and see myself primarily as a hobby photographer. The photo editing programs I use are mainly Affinity Photo 2, Capture One 23, On1 NoNoise AI 2023 and an older version of Niks Collection (pre-DxO).

Now my old computer, a MacBook Pro from 2012, has become too tired and needs to be replaced with a new and fresh MacBook Pro. The question then is which one should I choose? Right now my choice is between a new MacBook Pro M2 Max, 12C 30GPU 32GB/1TB 14.2" (2023) and a new MacBook Pro M3 Pro, 12C, 18GPU, 36GB/1TB 14.2" (2023).

I will not edit film or anything like that, but only edit images, write texts (Scrivener, Nisus Writer Pro, Mellel), but also use (in addition to previously listed image editing programs) programs such as Affinity Designer and Affinity Publisher.

When it comes to work with image editing programs, is it the new M3 processor that is preferable despite having only 18GPUs or is it the "older" M2 Max processor that is preferable with its 30GPUs. Does anyone know if any of the above image editing programs benefit from more GPU cores or if these programs will benefit more from the new M3 processor (not Max but Pro)?

The amount of RAM is more or less the same. 32 GB for the M2 Max and 36 GB for the M3 Pro, so I guess it's pretty similar (unless the M3 processor has better memory management) in performance. The price is more or less the same, with a small, small advantage for the M2.

Which computer would you recommend for the Affinity Photo2, Designer 2, and Publisher 2?

[v_kyr]

That's a hard/tough decision to take, as the M3 Pro SoC in the new Macbook Pro's are known to be significantly slimmed down now. For example their RAM binding transfer rates (M3 Pro = 192 bits/153.6 GB/s and M2 Pro = 256 bits/204.8 GB/s) have now decreased due to the omisson of a 4'th mem module (see below article reference). All in all the M3 Max SoC looks better here in contrast when compared with a M2 Max SoC, instead of the M3 Pro SoC when directly compared with the M2 Pro or M2 Max SoC.

Here's a partly translation from a german article about the new M3 chips use in the Macbook Pro line now ...

• Apple-Chips: Der M3 Pro ist der diesjährige Verlierer (the original DE article)

Partly translation:

Quote

Apple chips: The M3 Pro is this year's loser

Atypical – Apple is introducing the three new processors M3, M3 Pro and M3 Max in parallel. Not everything about the models is pleasing.

Table of contents

1. Apple chips: The M3 Pro is this year's loser
   • The M3 Max is huge
   • New GPU features: Ray tracing, Dynamic Caching
   • M3 Pro significantly slimmed down

 

The 2023 generation of MacBook Pros is here - and with it a new generation of Apple Silicon. Uncharacteristically, Apple presented the three models M3, M3 Pro and M3 Max in parallel and not in stages.

They are taking the long-awaited leap to 3-nanometer process technology from chip contract manufacturer TSMC. Apple has always been tight-lipped about the exact processor specifications. This year, however, the company is sharing fewer details than ever before: There is no precise information about the clock frequencies, cache sizes or teraflops of the GPUs. However, the data sheets are sufficient for an initial assessment.

The M3 Max is huge

Apple is pushing the limits with the M3 Max with 92 billion transistors. That's 45 percent more than its predecessor, the M2 Max. For comparison: AMD's fastest processors for compact notebooks, such as the Ryzen 7 7840U and Ryzen 9 7940HS, use 25 billion transistors.

Consequently, the advertised increase in performance is also large: the CPU of the M3 Max is said to be 50 percent faster compared to the M2 Max. The increase is primarily due to four additional performance cores, but as with the A17 Pro in the iPhone 15 Pro, Apple has also increased the clock frequencies in all models. 3.8 GHz seems realistic.

Comparison Apple Silicon M3 vs. M2 vs. M1
SoC	M3 Max	M2 Max	M1 Max	M3 Pro	M2 Pro	M1 Pro	M3	M2	M1
production	TSMC 3nm	TSMC 5nm	TSMC 5nm	TSMC 3nm	TSMC 5nm	TSMC 5nm	TSMC 3nm	TSMC 5nm	TSMC 5nm
Size	?	~510mm²	~432mm²	?	~288mm²	~246mm²	?	~145mm²	~120mm²
Transistors	92 billion	67 billion	57 billion	37 billion	40 billion	33.7 billion	25 billion	20 billion	16 billion
CPU
Performance cores	12	8th	8th	6	8th	8th	4	4	4
P-cores L2 cache	?	32 MB	24 MB	?	32 MB	24 MB	?	16 MB	12 MB
P-core clock	?	3.5GHz	3.2GHz	?	3.5GHz	3.2GHz	?	3.5GHz	3.2GHz
Efficiency cores	4	4	2	6	4	2	4	4	4
E-cores L2 cache	?	4 MB	4 MB	?	4 MB	4 MB	?	4 MB	4 MB
E-core clock	?	2.4GHz	2.06GHz	?	2.4GHz	2.06GHz	?	2.4GHz	2.06GHz
GPU
cores	40	38	32	18	19	16	10	10	8th
Shaders	5120	4864	4096	2304	2432	2048	1280	1280	1024
FP32 computing power	?	13.6 TFlops	10.4 TFlops	?	6.8TFlops	5.2TFlops	?	3.6TFlops	2.6TFlops
Number of external displays	4	4	4	2	2	2	1	1	1
Storage
RAM type	LPDDR5-6400	LPDDR5-6400	LPDDR5-6400	LPDDR5-6400	LPDDR5-6400	LPDDR5-6400	LPDDR5-6400	LPDDR5-6400	LPDDR4X-4266
Max quantity	128 GB	96 GB	64 GB	36 GB	32 GB	32 GB	24 GB	24 GB	16 GB
interface	512 bits	512 bits	512 bits	192 bits	256 bits	256 bits	128 bits	128 bits	128 bits
Transfer rate	409.6 GB/s	409.6 GB/s	409.6 GB/s	153.6 GB/s	204.8 GB/s	204.8 GB/s	102.4 GB/s	102.4 GB/s	68 GB/s
System level cache	?	48 MB	48 MB	?	24 MB	24 MB	?	8 MB	16 MB

New GPU features: Ray tracing, Dynamic Caching

There is a nominal increase of 20 percent in the graphics unit. With new features in all three M3 variants, developers can further increase performance if they put in the effort to customize their software.

Rendering programs in particular benefit from ray tracing acceleration. The GPU calculates how light rays behave in virtual space, which only works in real time with hardware acceleration. Apple does not reveal how far the implementation of the M3 generation extends.

Nvidia's ray tracing cores in the RTX GeForce graphics cards, for example, handle the intersection calculations of rays and the rest of the 3D world and run through the necessary acceleration structures (BVH tree). AMD's Radeon graphics cards, on the other hand, rely on a hybrid approach in which the normal shader cores go through the acceleration structures. This saves transistors, but is less powerful.

Apple is also focusing on improvements to the render backend with the M3 chips. So-called dynamic caching is intended to ensure that the GPU only loads data into the local caches that is actually needed. This potentially reduces waiting times, which in turn can increase GPU utilization.

With mesh shader support, Apple brings the rendering pipeline up to date. Developers no longer have to work rigidly with vertex and geometry shaders ; Mesh shaders can be used dynamically and potentially save performance. Nvidia has supported it since the RTX 2000 series, AMD since the RX 6000 generation.

The integrated media engine finally learns the AV1 codec, but can only decode videos with this - for example when you watch videos on YouTube or Netflix series. However, the M3 chips cannot encode in AV1 if you create videos yourself.

M3 Pro significantly slimmed down

The little brothers of the M3 Max are significantly less impressive. Apart from the new functions mentioned above, the normal M3 only receives product maintenance. Thanks to higher clock frequencies and architectural detail improvements, the CPU is said to be up to 20 percent faster than the M2. The same applies to the GPU.

...

Apple, on the other hand, has obviously trimmed the M3 Pro to have the lowest possible manufacturing costs. With 37 billion instead of 40 billion transistors and better manufacturing technology, the M3 Pro is significantly smaller than the M2 Pro.

When it comes to CPU performance, Apple only dares to compare it with the M1 Pro and there are good reasons for that: The CPU cores of the M3 Pro are said to be 20 percent faster than those of the M1 Pro. We remember: Apple already said the same thing with the M2 Pro . This means there is no increase in speed with the M3 Pro.

A look at the CPU configuration reveals why: The M3 Pro has six performance and efficiency cores each. The M2 Pro still had eight performance and four efficiency cores.

...

M3 Pro

Apple is saving the fourth memory module in the M3 Pro. This makes the wearer look asymmetrical. (Image: Apple)

...

The picture continues with the GPU and memory expansion. The former shrinks from 19 to 18 shader clusters (2304 vs. 2432 shaders), the RAM interface from 256 to 192 bits. Thanks to larger memory modules, there is still more RAM (36 vs. 32 GB in the most expensive configurations), but the transfer rate of the M3 Pro falls by a quarter compared to the M2 Pro and even the M1 Pro. This limits the graphics performance in particular.

There is also more RAM in the M3 Max only due to new memory modules with higher capacity. There are four components on a 512-bit wide interface, but in the most expensive configuration they each hold 32 GB.

...

( mma )

Usually I would say the newer the better (aka for M2 vs M3 chips/SoCs), but I'm not sure if that overall also applies here then for the M3 Pro SoC then versus the M2 Max SoC, thus you have decide on your own which one to finally take!

Quote

... (unless the M3 processor has better memory management) in performance ...

According to the above shown it probably won't offer a better/faster memory binding in performance, at least not for the M3 Pro SoC as that one is hardware wise reduced now!

Other than that both should be fine working horses for the Affinity suite.

[edwardsson]

@v_kyr Thank you for the infomation regarding the differences between the M3 Pro and the M2 Max. Highly appreciated.

[v_kyr]

See also some of the Max Tech Youtube comparison videos, which usually give some performance etc. overviews between M3 and M2 and the like.

• M3 Pro vs M2 Pro 14" MacBook Pro - WE WERE WRONG

• M3/Pro/Max MacBooks - Top 13 Problems after 1 Month!

• ... etc.

[edwardsson]

The “problem “was solved by a used MacBook Pro 13” M1 given to me by my oldest son.