M5 gaming

[crazy dave]

Since 3dmark doesn't report clock speeds for Apple systems, I would argue that all the variations are probably due to background processes. Since it can't be proven otherwise 😅.

I've seen it across benchmarks and reviewers. Variations of a few percent are simply expected.

Actually I was wrong about Geekerwan, check my edit that literally just posted, they get 36FPS average for the M5! I was mistaking the 1% lows for the average. Unless I've screwed up again.

Okay I'm pretty convinced most of these outlets are using 2.3 rather than 2.31.

EDIT: unless Geekerwan tested on high not Ultra? I can't read the Chinese and am unsure if the autotranslation is saying 1080p high or 1080p Ultra. It might be high because the FPS for their M4 Pro mini is 54, even more than the 47 you quote. AGGGGGGGGHHHHHHHHH!!!!!!

Okay assuming that Geekerwan tested on high not Ultra we can maybe take a stab at this:

Geekerwan get a 1080p high performance of 36FPS for the M5, NBC reports 34FPS for the M5. Pretty similar, within 6%.

Test Apple MacBook Pro M5 (2025) - Der schnellste Single-Core-Prozessor der Welt

Notebookcheck testet das Apple MacBook Pro 14 M5 mit dem neuen M5-SoC, 32 GB RAM und Nanotexturglas-Mini-LED.

www.notebookcheck.com

NBC reports the non-native 1080p High for 20-core M4 Pro yielded 45 FPS, while Geekerwan reports the M4 Pro (assuming 20-core they don't actually say) to be 54 for the native version. @OptimusGrime when you get a chance can you confirm?

Apple M4 Pro 20-core GPU - Benchmarks and Specs

The 20-core M4 GPU is an integrated graphics adapter designed by Apple that's built into the 14-core Apple M4 Pro SoC. It features 20 "cores" and has access to at least 24 GB of fast 273 GB/s LPDDR5x on-package RAM depending on the configuration. Its gaming performance is at least as good as...

www.notebookcheck.net

For the base M4, Geekerwan reports a score of 27FPS in native while NBC reports 23.7 for emulated.

Apple M4 10-core GPU - Benchmarks and Specs

The 10-core M4 GPU is a graphics adapter built into the 9-core and 10-core Apple M4 processors that has direct access to fast on-package LPDDR5x-7500 RAM (120 GB/s bandwidth) as well as hardware support for ray tracing and mesh shading and other modern technologies. It appears its real-world...

www.notebookcheck.net

Taking the first number that would indicate that the native performance was 20% over emulated. Which would leave another 20% for the M5's improvement over the M4, a much lower increase than almost anything else. Taking the second (more relevant since it's the base M4), native performance only increased by about 14% over emulated and the M5 would've improved over the base M4 by 26% ... similar to Steel Nomad's 28% (actually Geekerwan only saw 23% though I have to say their M4 result was on the high end, again that 6% variation). So maybe that's it and it all makes sense? Hmmmm ...

 

[Icelus]

 

[OptimusGrime]

I've seen it across benchmarks and reviewers. Variations of a few percent are simply expected.

Okay assuming that Geekerwan tested on high not Ultra we can maybe take a stab at this:

Geekerwan get a 1080p high performance of 36FPS for the M5, NBC reports 34FPS for the M5. Pretty similar, within 6%.

Test Apple MacBook Pro M5 (2025) - Der schnellste Single-Core-Prozessor der Welt

Notebookcheck testet das Apple MacBook Pro 14 M5 mit dem neuen M5-SoC, 32 GB RAM und Nanotexturglas-Mini-LED.

www.notebookcheck.com

NBC reports the non-native 1080p High for 20-core M4 Pro yielded 45 FPS, while Geekerwan reports the M4 Pro (assuming 20-core they don't actually say) to be 54 for the native version. @OptimusGrime when you get a chance can you confirm?

53 fps

 

[M4pro]

Baldur’s Gate 3 (from Andrew Tsai)

 

[M4pro]

Assassin's Creed Shadows (from Andrew Tsai)

 

[M4pro]

Control (from Andrew Tsai)

 

[crazy dave]

53 fps

Okay then, it's likely that both NBC and Geekerwan numbers are right and, at least in raster performance, CP2077 simply doesn't get that big of an uplift from M5 or the native version even with the fix. The combined uplift is nice, but it means that the M5 generation will struggle more in CP2077 benchmarks relative to Windows dGPUs than graphics benchmarks or even titles like BG3. Odd I would've bet money on the other way around between BG3 and CP2077.

That Ars Technica looks like it is the bugged version of CP2077 is odd, but maybe it's the 16 not 20-core Pro. I dunno.

 

[OptimusGrime]

Okay then, it's likely that both NBC and Geekerwan numbers are right and, at least in raster performance, CP2077 simply doesn't get that big of an uplift from M5 or the native version even with the fix. The combined uplift is nice, but it means that the M5 generation will struggle more in CP2077 benchmarks relative to Windows dGPUs than graphics benchmarks or even titles like BG3. Odd I would've bet money on the other way around between BG3 and CP2077.

That Ars Technica looks like it is the bugged version of CP2077 is odd, but maybe it's the 16 not 20-core Pro. I dunno.

I’m hoping Andrew Tsai tests the M4 mini to see the difference of 10 core vs 10 core

 

[diamond.g]

Okay then, it's likely that both NBC and Geekerwan numbers are right and, at least in raster performance, CP2077 simply doesn't get that big of an uplift from M5 or the native version even with the fix. The combined uplift is nice, but it means that the M5 generation will struggle more in CP2077 benchmarks relative to Windows dGPUs than graphics benchmarks or even titles like BG3. Odd I would've bet money on the other way around between BG3 and CP2077.

That Ars Technica looks like it is the bugged version of CP2077 is odd, but maybe it's the 16 not 20-core Pro. I dunno.

Isn't BG3 more CPU limited than CP2077.

 

[crazy dave]

Isn't BG3 more CPU limited than CP2077.

That's not necessarily a positive when discussing the uplift between two the M4 and M5. Some CPU limited games showed very little performance increase mostly because the CPU itself didn't improve that much, e.g. NBC had Total War Pharaoh only went up by 22%* (I believe a different reviewer found another TW game to go up by more).

No, the reason I made that earlier statement is that the Mac version of BG3 is only using Metal v2 I think and is missing a bunch of bells and whistles as a result. While I don't think CDPR has yet implemented Metal v4 in CP2077 (they said they will), I believe it using Metal v3. So my inclination would be that CP2077, by virtue of having access to a greater toolbox with more features exposed would've been able to take advantage of the M5 more than BG3. But it's possible as we hypothesize is true with Steel Nomad that BG3 uses more FP16 than CP2077 and the M5 doubled the FP16 resources. Something like that could be at play here. Maybe. As with Steel Nomad, this is conjecture.

*EDIT: "only" and "not big" are of course relative here, normally a 20-30% improvement gen-on-gen (especially when that gen is practically a year later) would be considered quite a strong gaming/graphics improvement!

 

[OptimusGrime]

The difference between MrMacRight and Dametech’s findings is wild. Unfortunately some of the games chosen are different.

 

[M4pro]

Death Stranding Avg FPS (4K, Highest)

M5 MBP: 32fps
M4 MBP: 23fps

From Luke Milani

 

[M4pro]

Cyberpunk 2077 Avg FPS (4K, Ray Tracing Ultra)

M5 MBP: 11fps
M4 MBP: 7fps

From Luke Milani (just a YouTube content creator - not in any way a Tech reporter)

 

[M4pro]

No Man’s Sky
3024x1964 - MetalFX: Performance setting

M5 MBP: 50-60fps

From Blendlogic Tech

 

[MayaUser]

So M5 is finally the base for these more heavily AAA games that offer a decent fps
M1-M3 were ok just for lighter intensive graphics like Dota/LoL etc
Probably fan-less M5 macbook air is out of the question since even with active cooling Mbp cant keep up with the heat in some cases, so for the Mba will be in most cases

 

[Queen6]

So M5 is finally the base for these more heavily AAA games that offer a decent fps
M1-M3 were ok just for lighter intensive graphics like Dota/LoL etc
Probably fan-less M5 macbook air is out of the question since even with active cooling Mbp cant keep up with the heat in some cases, so for the Mba will be in most cases

Mac's are not designed with gaming in mind so the cooling solution & fan curves reflect as silent operation is preferred. Macs Fan Control can solve the issue by allowing the user to manually override the fans.

My M1 MBP will just slowly heatsoak until it's literally hot to the touch on stock cooling with Macs Fan Control it remains cool. Don't need to max out the fans just get them spinning up sooner at 50% - 75%

Q-6

 

[MayaUser]

Mac's are not designed with gaming in mind so the cooling solution & fan curves reflect as silent operation is preferred. Macs Fan Control can solve the issue by allowing the user to manually override the fans.

My M1 MBP will just slowly heatsoak until it's literally hot to the touch on stock cooling with Macs Fan Control it remains cool. Don't need to max out the fans just get them spinning up sooner at 50% - 75%

Q-6

Here we are not talking about whats mac design for...this topic is pure for M5 gaming and still..M5 is the first base mac since M1 that can decent handling more complex gaming
M1 was garbage...you needed the M1 Max to do some decent gaming
Most of people are not using fan control...and for the M5 Macbook Air that i was talking about that it will have difficulties in most cases what fans you control?

 

[MayaUser]

Its clear if M6 cannot come back to M3/M4 heat parameters with its new "2nm" arh its clear Apple needs new cooling system
Im very curious to see the M5 Max in the 16" Mbp how it performs , since M5 Max in the 14" will be limited for sure
For me complex games besides Autodesk Maya for me are still the best way to push the limits of the new systems to see how much improvements we see each generation. Because they push almost everything that SoC offers in real apps

 

[maflynn]

Here we are not talking about whats mac design for...this topic is pure for M5 gaming

And yet the very design of the Mac, has a direct and distinct impact on M5's gaming performance. A for instance, the M4 MBA will perform significantly as in gaming in comparison to the M4 (or M5) MBP, simply because the design of the MBA is inhibiting the M4's performance.

 

[FrostLK8]

Yeah on iPad Pros as far as I can tell, 120 FPS is only achievable in the domains. Wish Apple paid Mihoyo to make a native Mac version, preferably without their invasive anticheat

Yeah, feelsbadman, looks like either Apple will never manage to keep thermals low enough for Genshin to actually pull off smooth 120, or Mihoyo will never optimize it enough to do that for real. 60 maxed is playable smoothly on both M4 and M5, but you'd need about double the FPS of M4 to be able to play smooth 120 on M2, and that won't be the case here.

Edit: To add, I see reports now that WildLife Extreme stress test (20 min throttle test) results are almost identical between M4 and M5, which brings into question whether there's any real difference in games like that at all between the two, and the 1.6x claim might be more for apps natively made for Mac rather than iPad games. Feels pointless I ordered one atm after still having he M4 in perfect condition, but yeah.

 

[maflynn]

Apple will never manage to keep thermals low enough

Apple has always valued quiet running/low fan noise over cooler running. They're perfectly content in having their macs run in the 80s, 90s, given the threshold for damage is > 100c

 

[Queen6]

Here we are not talking about whats mac design for...this topic is pure for M5 gaming and still..M5 is the first base mac since M1 that can decent handling more complex gaming
M1 was garbage...you needed the M1 Max to do some decent gaming
Most of people are not using fan control...and for the M5 Macbook Air that i was talking about that it will have difficulties in most cases what fans you control?

You will have difficulties gaming on any passively cooled notebook. Buying a Mac to game is questionable, buying an Air pointless. Design is a factor whether you like it or not, if the cooling solution is incapable (as was the case on many Intel based Mac's) active or passive the only recourse is for the system to throttle... Therefore physical design is a major factor.

Cyberpunk 2077 Raytracing Ultra @ 2560x1600

Right tool for the right job and all, or maybe design isn't a factor...

You need Max spec for modern AAA a 16" MBP or a Studio. M5 when it comes will be better, equally not going to deliver massive gaming performance. I looked; it is what it is.

As for M1 being garbage you might want to revisit some of the Intel MBP's. M1 was a revelation on launch, and they remain very capable for most common tasks today. It's not like they got slower with age. I game on mine with ease, equally my expectations are balanced playing older titles under modern source port engines.

Those that opt to try Macs Fan Control will likely benefit, as their MPB will remain cooler and less likely to throttle down.

Q-6

 

[Queen6]

Yeah, feelsbadman, looks like either Apple will never manage to keep thermals low enough for Genshin to actually pull off smooth 120, or Mihoyo will never optimize it enough to do that for real. 60 maxed is playable smoothly on both M4 and M5, but you'd need about double the FPS of M4 to be able to play smooth 120 on M2, and that won't be the case here.

Edit: To add, I see reports now that WildLife Extreme stress test (20 min throttle test) results are almost identical between M4 and M5, which brings into question whether there's any real difference in games like that at all between the two, and the 1.6x claim might be more for apps natively made for Mac rather than iPad games. Feels pointless I ordered one atm after still having he M4 in perfect condition, but yeah.

Not a solution, more of a workaround. Take manual control of the fans, stock fan curve is very conservative. Fans will barely move until 85C-90C. By this point while playing a game heatsoak is well under way. Mac's have a complex monitoring system with multiple flags to reduce SOC performance.

Key is to prevent the overheat in the first place, increasing baseline fan speed goes a long way. You can try and see, costs nothing.

Q-6

 

[maflynn]

Not a solution, more of a workaround. Take manual control of the fans, stock fan curve is very conservative. Fans will barely move until 85C-90C.

No question, and the lack of thermal headroom on the M4 Pro Mini, was one of the major reasons why I returned it in lieu of a M4 Max Studio - the other reason was the poor gaming performance of the m4 Pro.

While some folks have an issue of my running crossover and complaining about the M4 performing poorly, the M4 Max performs great and running crossover is really the only way to play AAA games on the mac. So seeing the M5's GPU performance increase as much as it is, bodes well for people wanting to game with Crossover.

 

[keksikuningas]

Apple has always valued quiet running/low fan noise over cooler running. They're perfectly content in having their macs run in the 80s, 90s, given the threshold for damage is > 100c

we were talking about the iPad Pro