M3 Max 14/30 vs 16/40 for Gaming with Benchmarks

[FrankySavvy]

I side by side tested a 16" M3 Max with the 14/30 36gb Ram (Silver) and 16" M3 Max with 16/40 48gb Ram (Space Black). This was specifically for gaming.

I have to say there wasn't much of a difference in my testing with the benchmarks and games I had at my disposal. The 3D Mark Wildlife Extreme was the only benchmark with a notable difference. From my experience the standard Wildlife Extreme was more realistic to my findings. Only a negligible FPS boost overall.

Here is what I found:

3D Mark Benchmarks:

3D Mark Wildlife Extreme Unlimited

M3 Max 14/30 - 24221 Score | 145.0 FPS
M3 Max 16/40 - 31718 Score | 189.9 FPS

3D Mark Wildlife Extreme

M3 Max 14/30 - 19652 Score | 117.7 FPS
M3 Max 16/40 - 19999 Score | 119.8 FPS



Dawn of War II Benchmark (Crossover)
1728 x 1117 Resolution - Ultra Settings

M3 Max 14/30 - 23.25 Avg FPS | 45.05 High FPS | 10.05 Low FPS
M3 Max 16/40 - 24.63 Avg FPS. | 46.62 High FPS | 13.91 Low FPS


Dawn of War III Benchmark (Parallels)
1728 x 1117 Resolution - Ultra Settings

M3 Max 14/30 - 39.54 Avg FPS | 86.96 High FPS | 12.39 Low FPS
M3 Max 16/40 - 40.93 Avg FPS | 95.00 High FPS | 16.82 Low FPS


Dawn of War II Benchmark (Native)
2560 x 1600 Resolution - Ultra Settings

M3 Max 14/30 - 113.92 Avg FPS | 255.74 High FPS | 58.12 Low FPS
M3 Max 16/40 - 103.12 Avg FPS | 235.82 High FPS | 46.71 Low FPS


Dawn of War III Benchmark (Native)
2560 x 1600 Resolution - Ultra Settings

M3 Max 14/30 - 46.76 Avg FPS | 60.64 High FPS | 34.62 Low FPS
M3 Max 16/40 - 44.10 Avg FPS | 60.00 High FPS | 30.00 Low FPS


Shadow of the Tomb Raider Benchmark (Crossover)
1728 x 1117 Resolution - High Settings

M3 Max 14/30 - 81 Avg FPS | 104 High FPS | 62 Low FPS
M3 Max 16/40 - 87 Avg FPS | 109 High FPS | 66 Low FPS


Starcraft II Observed (Crossover) - Standard 1v1
1728 x 1117 Resolution - Extreme Settings

M3 Max 14/30 - 32 Avg FPS | 43 High FPS | 19 Low FPS
M3 Max 16/40 - 37 Avg FPS | 51 High FPS | 19 Low FPS


Starcraft II Observed (Native) - Standard 1v1
1920 x 1200 Resolution - Ultra Settings

M3 Max 14/30 - 37 Avg FPS | 119 High FPS | 22 Low FPS
M3 Max 16/40 - 38 Avg FPS | 119 High FPS | 24 Low FPS


Diablo IV Observed (Crossover) - In Dungeon
1728 x 1117 Resolution - Highest Settings

M3 Max 14/30 - 55 Avg FPS | 80 High FPS | 32 Low FPS
M3 Max 16/40 - 58 Avg FPS | 80 High FPS | 34 Low FPS


For the observed findings there was almost no difference in performance as the games were being played. A few FPS differences here and there in the FPS counter. No differences affected gameplay in anyway, only way I could tell the difference was by observing the FPS counter.

What is amazing is all these games are not optimized for Apple Silicon, that shows strong promise for the future of gaming on the Mac!

In conclusion if you're buying the M3 Max 16/40 for gaming, it really is not worth it in real world use. Save the $500-$600 bucks.

 

[T'hain Esh Kelch]

I like how you write the colors, as if that affected your results.

In any event, everything you have tested is emulated. Why not test something native? We have no idea if the emulation softwares limits the use of GPU cores (Which it appears to!).

 

[Pressure]

By emulating and using a low resolution the burden is on the CPU.

You are CPU limited in all your benchmarks except the 3D Mark Unlimited run.

 

[FrankySavvy]

I like how you write the colors, as if that affected your results.

In any event, everything you have tested is emulated. Why not test something native? We have no idea if the emulation softwares limits the use of GPU cores (Which it appears to!).

I wrote in the colors because I also attached images.

 

[mdhaus72]

I side by side tested a 16" M3 Max with the 14/30 36gb Ram (Silver) and 16" M3 Max with 16/40 48gb Ram (Space Black). This was specifically for gaming.

I have to say there wasn't much of a difference in my testing with the benchmarks and games I had at my disposal. The 3D Mark Wildlife Extreme was the only benchmark with a notable difference. From my experience the standard Wildlife Extreme was more realistic to my findings. Only a slight FPS boost overall.

Here is what I found:

3D Mark Benchmarks:

3D Mark Wildlife Extreme Unlimited

M3 Max 14/30 24221 Score 145.0 FPS
M3 Max 16/40 31718 Score 189.9 FPS

3D Mark Wildlife Extreme

M3 Max 14/30 19652 Score 117.7 FPS
M3 Max 16/40 19999 Score 119.8 FPS



Dawn of War II Benchmark (Crossover)
1728 x 1117 Resolution - Ultra Settings

M3 Max 14/30 23.25 Avg FPS 45.05 High FPS 10.05 Low FPS
M3 Max 16/40 24.63 Avg FPS 46.62 High FPS 13.91 Low FPS


Dawn of War III Benchmark (Parallels)
1728 x 1117 Resolution - Ultra Settings

M3 Max 14/30 39.54 Avg FPS 86.96 High FPS 12.39 Low FPS
M3 Max 16/40 40.93 Avg FPS 95.00 High FPS 16.82 Low FPS


Shadow of the Tomb Raider Benchmark (Crossover)
1728 x 1117 Resolution - High Settings

M3 Max 14/30 81 Avg FPS 104 High FPS 62 Low FPS
M3 Max 16/40 87 Avg FPS 109 High FPS 66 Low FPS


Starcraft II Observed (Crossover) - Standard 1v1
1728 x 1117 Resolution - Extreme Settings

M3 Max 14/30 32 Avg FPS 43 High FPS 19 Low FPS
M3 Max 16/40 37 Avg FPS. 51 High FPS. 19 Low FPS


Diablo IV Observed (Crossover) - In Dungeon
1728 x 1117 Resolution - Highest Settings

M3 Max 14/30 55 Avg FPS 80 High FPS 32 Low FPS
M3 Max 16/40 58 Avg FPS. 80 High FPS. 34 Low FPS


For the observed findings there was almost no difference in performance as the game was being played. A few FPS differences here and there in the FPS counter. No differences affected gameplay in anyway, only way I could tell the difference was by observing the FPS counter.

What is amazing is all these games are not optimized for Apple Silicon, that shows strong promise for the future of gaming on the Mac!

In conclusion if you're buying the M3 Max 16/40 for gaming, it really is not worth it in real world use. Save the $500-$600 bucks.

As mentioned elsewhere, the bigger difference will probably come with native gaming titles…Emulation likely hits a ceiling on many games.

Also, the extra $500-$600 isn’t just for a better chip - It’s also for an additional 16 GB of RAM to take the total to 64 GB. That makes it a really good deal overall for a lot of people.

 

[FrankySavvy]

I will run the dawn of war 2 and 3 benchmarks natively on steam at higher resolutions and report back.

 

[Ascn]

Could you also report on battery consumption please?

 

[FrankySavvy]

Okay so update. Ran a few more benchmarks. With very surprising results.

Dawn of War II Benchmark (Native)
2560 x 1600 Resolution - Ultra Settings

M3 Max 14/30 - 113.92 Avg FPS | 255.74 High FPS | 58.12 Low FPS
M3 Max 16/40 - 103.12 Avg FPS | 235.82 High FPS | 46.71 Low FPS


Dawn of War III Benchmark (Native)
2560 x 1600 Resolution - Ultra Settings

M3 Max 14/30 - 46.76 Avg FPS | 60.64 High FPS | 34.62 Low FPS
M3 Max 16/40 - 44.10 Avg FPS | 60.00 High FPS | 30.00 Low FPS

Seems to be that the 14/30 is running better than the 16/40 in these benchmarks.

Testing Starcraft II Native now.

 

[FrankySavvy]

Additional benchmarks:

Starcraft II Observed (Native) - Standard 1v1
1920 x 1200 Resolution - Ultra Settings

M3 Max 14/30 - 37 Avg FPS | 119 High FPS | 22 Low FPS
M3 Max 16/40 - 38 Avg FPS | 119 High FPS | 24 Low FPS

Okay so playing native Starcraft II there is no discernible difference on 1920 x 1200 Ultra Settings between the M3 Max 14/30 and 16/40.

Added above as well.

 

[Pressure]

Those three games are Intel binaries and run through the Rosetta 2 translation layer. You are still emulating. I believe Starcraft II is even using OpenGL, which Apple deprecated back in 2018 with macOS Mojave.

You need games compiled for Aarch64 and the Metal graphics API to test performance like you try to do.

Do you need help to find games compiled for Aarch64 and Metal?

The 3DMark Unlimited run should make you go "hmm, is something going on here" 🤔

 

[DFBwin]

TYVM for these tests.
I own a 14 inch M1 Pro and use it regularly for gaming, thinking about upgrading in 1-2 years.

How about using the app ‘stats‘ to see the CPU and GPU usage of your systems, to identify the bottleneck?

 

[magbarn]

Additional benchmarks:

Starcraft II Observed (Native) - Standard 1v1
1920 x 1200 Resolution - Ultra Settings

M3 Max 14/30 - 37 Avg FPS | 119 High FPS | 22 Low FPS
M3 Max 16/40 - 38 Avg FPS | 119 High FPS | 24 Low FPS

Okay so playing native Starcraft II there is no discernible difference on 1920 x 1200 Ultra Settings between the M3 Max 14/30 and 16/40.

Added above as well.

As the above poster says, StarCraft 2 is not AS native. It’s still using Rosetta. Appreciate the work you’re doing, but your tests are showing more the overhead CPU limits of Rosetta.

You have to do RE4, Lies of P, or BG3 benchmarks to see true differences. I’m pretty sure WoW is also native.

 

[FrankySavvy]

Those three games are Intel binaries and run through the Rosetta 2 translation layer. You are still emulating. I believe Starcraft II is even using OpenGL, which Apple deprecated back in 2018 with macOS Mojave.

You need games compiled for Aarch64 and the Metal graphics API to test performance like you try to do.

Do you need help to find games compiled for Aarch64 and Metal?

The 3DMark Unlimited run should make you go "hmm, is something going on here" 🤔

On notebook check and mordant.com I found these benchmarks:

Balders Gate III (Native)

1920 x 1080 Resolution - High Settings

M3 Max 14/30 - 90 Avg FPS
M3 Max 16/40 - 96 Avg FPS


3840 x 2160 Resolution - High Settings

M3 Max 14/30 - 50 Avg FPS
M3 Max 16/40 - 60 Avg FPS

So definitely an increase but still I don’t think it’s worth the almost $600 for 6-10fps.

 

[Beau10]

Also, the extra $500-$600 isn’t just for a better chip - It’s also for an additional 16 GB of RAM to take the total to 64 GB. That makes it a really good deal overall for a lot of people.

It's 12 to take to 48.

 

[Beau10]

On notebook check and mordant.com I found these benchmarks:

Balders Gate III (Native)

1920 x 1080 Resolution - High Settings

M3 Max 14/30 - 90 Avg FPS
M3 Max 16/40 - 96 Avg FPS


3840 x 2160 Resolution - High Settings

M3 Max 14/30 - 50 Avg FPS
M3 Max 16/40 - 60 Avg FPS

So definitely an increase but still I don’t think it’s worth the almost $600 for 6-10fps.

In the latter situation it's 20% faster. That is indeed significant, although a shortfall of the theoretical 33% improvement from the sheer number of cores/bus speed increase.

 

[FrankySavvy]

In the latter situation it's 20% faster. That is indeed significant, although a shortfall of the theoretical 33% improvement from the sheer number of cores/bus speed increase.

I don't think it's noticeable enough to warrant the extra cost. I am tempted but for my use case and the fact that there aren't many native AAA games on a Mac to begin with. The base M3 Max seems the way to go.

 

[G5isAlive]

I don't think it's noticeable enough to warrant the extra cost. I am tempted but for my use case and the fact that there aren't many native AAA games on a Mac to begin with. The base M3 Max seems the way to go.

Really do appreciate the work you put into this. I agree with your conclusion about the value add, but people will argue over that point because value add is subjective not objective. And no one likes being told they are essentially stupid (not your words I know) for having a different opinion. For me the best comparisons just give me the data and let me decide for myself whether something is worth the cost or not. Not saying you crossed any line, just sort of a comment on why some might be disagreeing with you.

 

[magbarn]

I don't think it's noticeable enough to warrant the extra cost. I am tempted but for my use case and the fact that there aren't many native AAA games on a Mac to begin with. The base M3 Max seems the way to go.

it would’ve been for me but I wanted 64gb of ram and the base M3 max forces you to jump from 48 to 96 gb of ram which is just way too much.

 

[Beau10]

I don't think it's noticeable enough to warrant the extra cost. I am tempted but for my use case and the fact that there aren't many native AAA games on a Mac to begin with. The base M3 Max seems the way to go.

A 20% bump is what you can expect generation to generation… for only 15% more you get that + a bump in RAM. Yes, at lower res where it’s CPU dominant much less a case can be made and I agree that binned is the value champ overall. If someone won't upgrade for another few generations and is hopeful for the future of Mac gaming there is certainly value going for the upgrade.

That said, OLED is going to be a big boon for gaming (esp. response times) if that comes out by 2026 as rumors indicate might be better conserving capital a bit. Or perhaps not if one wants to wait a bit due to burn-in concerns. I feel like for the most part it's a wash and neither option is a clear winner.

 

[Crow47]

I'm getting significantly higher FPS than you are in Diablo IV via Crossover. I'm running at 2304x1296 (Mac OS desktop scaled resolution), Ultra Settings (no FSR), and it's spending the vast majority of its time at 120FPS connected to a LG C1 OLED via HDMI 2.1. It would happily do the same at 1440p, but I enjoy the fact the computer doesn't have to work as hard at the lower resolution.

I'm happily managing 120FPS in a variety of titles -both Mac/AS-native and via Crossover. 1440P is where it's very happy, but 4K60 is easily acheivable across the board. M3 Max 40c is a bona-fide monster.

Double check performance with Vsync when you're benchmarking these games -my testing so far reveals a strange interaction between VRR displays (ProMotion in the MBP included) and Vsync. Sometimes VSync on works fine, and sometimes it doesn't. I have discovered games I thought performed more marginally actually perform very well, I just have to figure out whether the game likes VSync on or off first. As well, I wouldn't expect an older title like DoW2 to do very well on Crossover -you'll see the best performance using their D3Dmetal translation layer -which only works with DX12.

RE8 on the App Store doesn't like Vsync turned on with the MBP display. Neither does Diablo IV. However, Diablo IV works great with Vsync on the LG TV, however RE8 does not. Seems to be a quirk we get to experience on the edge of the Mac OS gaming experience

 

[FrankySavvy]

I'm getting significantly higher FPS than you are in Diablo IV via Crossover. I'm running at 2304x1296 (Mac OS desktop scaled resolution), Ultra Settings (no FSR), and it's spending the vast majority of its time at 120FPS connected to a LG C1 OLED via HDMI 2.1. It would happily do the same at 1440p, but I enjoy the fact the computer doesn't have to work as hard at the lower resolution.

I'm happily managing 120FPS in a variety of titles -both Mac/AS-native and via Crossover. 1440P is where it's very happy, but 4K60 is easily acheivable across the board. M3 Max 40c is a bona-fide monster.

Double check performance with Vsync when you're benchmarking these games -my testing so far reveals a strange interaction between VRR displays (ProMotion in the MBP included) and Vsync. Sometimes VSync on works fine, and sometimes it doesn't. I have discovered games I thought performed more marginally actually perform very well, I just have to figure out whether the game likes VSync on or off first. As well, I wouldn't expect an older title like DoW2 to do very well on Crossover -you'll see the best performance using their D3Dmetal translation layer -which only works with DX12.

RE8 on the App Store doesn't like Vsync turned on with the MBP display. Neither does Diablo IV. However, Diablo IV works great with Vsync on the LG TV, however RE8 does not. Seems to be a quirk we get to experience on the edge of the Mac OS gaming experience

Thank you for your recommendations. Unfortunately in crossover it gets complicated and trial and error with non native games. I do have a gaming PC as well, for me, i could sell my gaming PC to offset the additional cost for the 40 core but then I’d loose out on many titles that I play that have yet to run on crossover (Microsoft Flight Simulator 2020, Age of Empires IV)

For my use case, I can’t warrant the additional almost $600 for a 10 FPS increase avg while having a gaming PC that plays all games flawlessly.

It seems games that are natively supported by Apple silicon run so well on even the M3, so having even the base M3 Max should more than suffice.

 

[bradman83]

Those three games are Intel binaries and run through the Rosetta 2 translation layer. You are still emulating. I believe Starcraft II is even using OpenGL, which Apple deprecated back in 2018 with macOS Mojave.

You need games compiled for Aarch64 and the Metal graphics API to test performance like you try to do.

Do you need help to find games compiled for Aarch64 and Metal?

The 3DMark Unlimited run should make you go "hmm, is something going on here" 🤔

Minor quibble - Rosetta translates calls made to system APIs into essentially native code the first time an app is launched. Emulation only occurs for the parts of the code that don't make calls to system APIs that can be translated. If an Intel game uses Metal for the graphics then the graphics portion will run at near full speed even if other aspects of the game engine are emulated. A good example of this was DxO PhotoLab - it was able to use the M1's Neural Engine for its denoising engine while the app was still Intel-only because the denoising engine leveraged CoreML APIs. The app spoke to the APIs and the APIs spoke to the hardware natively without the need for emulation for that specific part of the app; performance for that function boosted significantly despite still being Intel code otherwise.

So yes, Metal is extremely important, and ARM-native will absolutely be a boost as well especially for CPU-intensive games like Cities: Skylines, but not as big of one as you might think.

 

[FrankySavvy]

Minor quibble - Rosetta translates calls made to system APIs into essentially native code the first time an app is launched. Emulation only occurs for the parts of the code that don't make calls to system APIs that can be translated. If an Intel game uses Metal for the graphics then the graphics portion will run at near full speed even if other aspects of the game engine are emulated. A good example of this was DxO PhotoLab - it was able to use the M1's Neural Engine for its denoising engine while the app was still Intel-only because the denoising engine leveraged CoreML APIs. The app spoke to the APIs and the APIs spoke to the hardware natively without the need for emulation for that specific part of the app; performance for that function boosted significantly despite still being Intel code otherwise.

So yes, Metal is extremely important, and ARM-native will absolutely be a boost as well especially for CPU-intensive games like Cities: Skylines, but not as big of one as you might think.

Interesting. I wish more developers would come and bring their games over.

 

[FrankySavvy]

I'm getting significantly higher FPS than you are in Diablo IV via Crossover. I'm running at 2304x1296 (Mac OS desktop scaled resolution), Ultra Settings (no FSR), and it's spending the vast majority of its time at 120FPS connected to a LG C1 OLED via HDMI 2.1. It would happily do the same at 1440p, but I enjoy the fact the computer doesn't have to work as hard at the lower resolution.

I'm happily managing 120FPS in a variety of titles -both Mac/AS-native and via Crossover. 1440P is where it's very happy, but 4K60 is easily acheivable across the board. M3 Max 40c is a bona-fide monster.

Double check performance with Vsync when you're benchmarking these games -my testing so far reveals a strange interaction between VRR displays (ProMotion in the MBP included) and Vsync. Sometimes VSync on works fine, and sometimes it doesn't. I have discovered games I thought performed more marginally actually perform very well, I just have to figure out whether the game likes VSync on or off first. As well, I wouldn't expect an older title like DoW2 to do very well on Crossover -you'll see the best performance using their D3Dmetal translation layer -which only works with DX12.

RE8 on the App Store doesn't like Vsync turned on with the MBP display. Neither does Diablo IV. However, Diablo IV works great with Vsync on the LG TV, however RE8 does not. Seems to be a quirk we get to experience on the edge of the Mac OS gaming experience

These are the settings I am running. I am running Diablo 4 through Crossover, without any advanced settings enabled. Whenever I try and enable D3DMetal, the game doest run.

I get a solid 60 FPS Average, its playable but I want to make sure I am not missing anything.

Thanks

 

[Crow47]

These are the settings I am running. I am running Diablo 4 through Crossover, without any advanced settings enabled. Whenever I try and enable D3DMetal, the game doest run.

I get a solid 60 FPS Average, its playable but I want to make sure I am not missing anything.

Thanks

Best practice with Diablo IV and Crossover is to set up your bottle with D3DMetal + Esync. It may work on DXVK but the best performance is with the settings I just mentioned. I can tell you that with Diablo IV, it won't run with just D3DMetal, or just Esync, it needs both of them enabled (see below). MSync does not apply for Diablo IV, do not use it with that game.

Use Crossover to make you a Diablo IV bottle using the built-in recipes if you don't feel like messing with the bottle's individual settings.

I've done some not-insignificant testing with Crossover and my Steam library at this point. If you have any more questions, not just about Diablo IV, feel free to ask!

Also! I highly recommend enabling high-performance fan mode in system settings if you haven't done so already!