Mac Studio 4k benchmarks (or: Is it fast?)

Support Mac Technical Support
1

Is the Mac Studio an upgrade? What kind of performance should I expect? How fast is it?

All values obtained at 4k resolution. THIS IS NOT A PROFESSIONAL REVIEW.

Charts first, details after.

Charts

Valdrakken

Valdrakken Performance GraphValdrakken

The nursery!

Ardenweald

Ardenweald Performance GraphArdenweald

This is one of the hardest zones to render. We have large swathes of layered terrain, foilage, and SFX. We see the steepest dive here between q=4 and 6; whereas the other zones lose about 1/3rd performance, Ardenweald loses a staggering 57%.

The compute graphics setting does incur a significant performance cost in this zone.


Garrison

Garrison Performance GraphGarrison

One of my favourite places to be; terrain, buildings, NPCs, and foilage fill the view. This is one of the game's heaviest pre-Ardenweald zones.

The Studio turns in smaller numbers relative to other zones at the lower quality levels.

Thoughts

We see the Studio is playable at all quality settings, although the Ardenweald q=10 scenario does present a counterpoint. Render scale and compute settings are superb sacrificial lambs, ready to be launched into the gaping maws of the framerate spirits.

The 5700XT reaches but is artificially constrained at 120FPS at the lowest quality settings. This is an outlandish scenario which will not affect players.

If you own the Mac Studio, you'll be happy wiling your time away in World of Warcraft too.

Bonus Charts

From the Shadowlands...

Garrison Combo

The above chart integrates my previous set of benchmarks (see: https://us.forums.blizzard.com/en/wow/t/quick-4k-m1-mac-mini-vs-2012egpu-benchmarks/796697). Don't take the numbers too seriously; that was a different version of the game (Shadowlands!), and Christmas decorations were still up.

The q=6-10 numbers retrace their steps on the 2012 mini; DF however raises the floor.

The M1 mini is not appropriate for 4k WoW.

What about CPU gains?

2018 and 2012 comparative performance

What happens if we improve CPU performance by pairing the Radeon RX580 with the 2018 mini? Evidently, not a lot.

Besides the CPU headroom, the 2018 mini features also a Thunderbolt 3 link while the 2012 mini is equipped with Thunderbolt (1); I was hoping to see a clear improvement here. This suggests this graphics card is the bottleneck. Unfortunately, I was unable to test the 2012 mini with the 5700XT.

The numbers weren't included in the regular charts as they were virtually identical, aside from anomalies; I verified VSync and Target FPS were disabled, but this combo was capped at 60FPS.

I would definitely run the 2018 mini over the 2012 mini; it doesn't as much sound like a banshee screaming in my ear; runs versions of operating system more able to support WoW (and other modern software) for the foreseeable future; is noticably quicker at a variety of tasks; and comes in a very tidy dark grey shell.

Benchmarking Details

Benchmarking was performed in static areas; there are too many variables in PvE content, or highly-trafficked spaces, to pin down reliable numbers.

Each scene was loaded with the default camera angle and character placed in the same position. The client was given some time to stabilise with each quality change before the numbers were recorded.

I eyeballed the numbers; where the framerates would often fluctuate, I attempted to generate a rough average, biasing common ranges and values. If the client would spent nearly all its time at a specific value, I just recorded that value.

Configurations!

Devices

2012 Mac mini 2018 Mac mini Mac Studio
(base Ultra)
  • OSX 10.15.7
  • i7 2.3GHz quad
  • 16GB RAM
  • 1TB (2x500GB) SSD
  • AMD Radeon RX580 eGPU via TB1
    Release: 2017
  • Speed Design mini cooling platform
  • macOS 12.6.2
  • i7 3.2GHz hex
  • 32GB RAM
  • 500GB onboard SSD
  • AMD Radeon 5700XT eGPU via TB3
    Release: 2019
  • Speed Design mini cooling platform
  • macOS 12.6.2
  • M1 Ultra
    20 CPU cores
    48 GPU cores
  • 64GB RAM
  • 1TB onboard SSD

WoW configuration

  • Full UI reset - deleted cache, wtf, interface pre-bench
  • VSync OFF
  • 100% render scale
  • Fullscreen (Windowed) Mode
  • FPS cap disabled
  • Target FPS disabled
  • Anti-Aliasing set to CMAA 2
  • Physics set to Player & NPCs
  • Verified FSR & FFX enabled
  • All other configurations left to default
  • UI turned off
  • Same identical installation of WoW was copied across every device

Shared configuration

  • 4k display - LG 27MU67 (save for Radeon RX580 entries -- this was connected to a 4k TV)
  • Steinberg UR22 audio interface connected
  • Launcher closed but running
  • All other software unloaded:
    In a normal professional review, these machines would be wiped and the operating systems reinstalled. I'm not a professional and didn't want to spend days on this, so I just unloaded or uninstalled what I could.
2

Retaining post for future updates.

3

Thanks for providing this analysis. I have been on the fence about buying a Mac Studio. I think this convinced me to hold off until it is updated with M2 chips. Was hoping for better performance at max settings in some of those difficult zones.

2 Likes
4

You’re very welcome – I’m glad it helped you reach a decision! There’s a lot of anecdotal discussion and apples-to-oranges comparisons so I wanted to bring something concrete to the table.

I’m happy with the Studio’s performance (but of course, I already own the device); I do expect the loaded M2 Max to be a real contender, especially given it’s landing with both architectural improvements and increased GPU core count (over the M1 Max).

All of that for less money.

5

It’s a shame supply shortages putting stuff so far behind. We were supposed to get the M2 pro and max mbp like 5 month, mini probably too. we’d probably be a lot closer to m2 studio at this point, now i can just see them skipping it entirely and waiting for m3. and the mac pro which was gonna be an awesome machine with more than twice the power of studio, will now likely be same as studio but in a bigger case.

6

Nice review! But just curious, why are you using CMAA2? I would say on a 27inch 4K AA is not really needed and all the AA modes especially CMAA2 are costing way too much FPS.

7

Thanks Epoeu!

Two reasons really:

  • I do find a little AA improves visuals, particularly when you have hard edges in high-contrast situations.
  • It does add a little more stress to the system, valuable I think when benchmarking equipment.
8

A note, on apple silicon, CMAA2 actually costs more performance than MSAA 4x, test it out yourself. not only does MSAA look better, it performs better. This is due to fact CMAA2 and CMAA have performance issues on tile based gpu rendering techniques (what apple silicon uses). I use CMAA2 on my intel mac since it’s more performant there, but MSAA on M1 Max

1 Like
9

I do think also MSAA 2x looks better than CMAA2. MSAA 4x is too much for my M1 Max and I do use FSR these days but on a 83% and resample sharpness on 1.
Anyways with AA on still had some scenario’s which would cause stuttering but it seems now I am running Mac OS Ventura 13.2 most of them seems gone and it feels butter smooth again (60FPS).

1 Like