LTT Watercooled a Mac Studio...

[Agent OrangeZ]

Cooler temps... no real performance improvement...

 

[lcubed]

did the acoustic signature go from whine to gurgle?

 

[mikeka]

Doubles as a bong?

 

[Pressure]

According to Alex the only difference were after hours and hours of stress testing, where the normal Mac Studio started revving its fans and the heatsink became heat saturated at the default 1300 RPM fan speed.

Something I have never been able to do on my unit, even with over night renders.

There were no performance differences, at all.

 

[Feek]

There were no performance differences, at all.

Good to know and I'm not too surprised. When I first set mine up, I used Macs Fan Control but as soon as the temperature started increasing for pretty much any reason, the fan speed ramped up. I decided I didn't want that so removed it.

Now, with it under the desk, it's effectively completely silent and on the few occasions I drive it hard enough for the temperature to climb, it's still silent. I was doing some work yesterday evening, glanced up and saw the temps reported as about 55°C but it was still dead quiet.

 

[pastrychef]

I have a real weakness for watching train wrecks.

 

[Killerbob]

Super cooling, i.e. water-cooling, is for overclocking CPUs. Sure it is also for quiet operations, but Macs are generally quiet computers, already more quiet than even water-cooled computer.

 

[haralds]

It's an Intel thing, out of place with Apple Silicon - like a cooler in the snow.

 

[Wowfunhappy]

There were no performance differences, at all.

That's not quite the case. They show in the video that the watercooled version did score consistently higher in benchmarks. It was just a very minor difference—not even close to being worth the effort. (But that's not really what the video was going for—they also say many times to please not try this at home!)

Super cooling, i.e. water-cooling, is for overclocking CPUs. Sure it is also for quiet operations, but Macs are generally quiet computers, already more quiet than even water-cooled computer.

Yeah, it would be really interesting if they could unlock the clock speed...

 

[pastrychef]

That's not quite the case. They show in the video that the watercooled version did score consistently higher in benchmarks. It was just a very minor difference—not even close to being worth the effort. (But that's not really what the video was going for—they also say many times to please not try this at home!)

My unmodded Mac Studio did better than both of theirs in Cinebench.

 

[Pressure]

That's not quite the case.

30 points difference when the score is 12000 is safely within margin of error. It’s so insignificant that we might talk 5 nanoseconds faster if at that.

That’s just taking being pedantic to a whole new level.

Cooling is just not the limiting factor when it comes to Mac Studio.

 

[Boil]

It's an Intel thing, out of place with Apple Silicon - like a cooler in the snow.

The current laptop-class SoCs we have seen may not really need liquid cooling, but a 420W M2 Extreme desktop-class SoC might benefit...?

 

[Wowfunhappy]

30 points difference when the score is 12000 is safely within margin of error. It’s so insignificant that we might talk 5 nanoseconds faster if at that.

It’s not margin of error if the difference happens consistently across multiple tests, every single time.

I suppose it could be chalked up to silicon differences, however. In hindsight, they really should have run the test before creating the mod. Shame they didn't.

 

[pastrychef]

Has there ever been any indication that Cinebench causes throttling on Mac Studios?

 

[Pressure]

Has there ever been any indication that Cinebench causes throttling on Mac Studios?

I don't know of any workload that throttles a Mac Studio. The entire video was just a "tongue in cheek" prank to annoy someone.