Disappointing for sure. I would like to know exactly where the weak point is with the new thermal design. Looks to me like the heatsink is large enough but maybe low airflow is still an issue.
2018 mini 3.2ghz I7 temps?
- Thread starter archer75
- Start date
- Sort by reaction score
Disappointing for sure. I would like to know exactly where the weak point is with the new thermal design. Looks to me like the heatsink is large enough but maybe low airflow is still an issue.
My mini is setup and I'm doing an encode with handbrake, H.265 10bit. I have the 3.2ghz i7.
I'm also streaming a 4k movie via plex to 2 iPhones and a previous gen MacBook Air.
It shows the CPU is running at 3.6ghz with a temp of 97c. Temp is bouncing between 94-99. Clock speed goes from 3.5 to 3.7
The mini is a little warm to the touch but not even remotely close to hot. The air coming out of it is warm.
I'm also streaming a 4k movie via plex to 2 iPhones and a previous gen MacBook Air.
It shows the CPU is running at 3.6ghz with a temp of 97c. Temp is bouncing between 94-99. Clock speed goes from 3.5 to 3.7
The mini is a little warm to the touch but not even remotely close to hot. The air coming out of it is warm.
Last edited:
Any chance you could test some h.265 encoding with handbrake?
That's what I did. h.265 10bit.
How many FPS do you get with just 8-bit? And are you using x265 or quick sync?
Doesn't the T2 handle H265 vs quick sync/
x265. I don't even see an option for quick sync like there is on the windows version. Average FPS of 9.3 And that's at 10bit.
My PC is also doing an encode with handbrake and has a avg fps of 2.8. Not using quick sync. CPU is a Intel 5820, also a 6 core.
I've matched the settings on handbrake for both PC and Mac for comparison.
[doublepost=1541727512][/doublepost]
It's not using the T2 chip as the CPU is maxed. And it doesn't have an option for quick sync.
Last edited:
Thanks! One last question for comparison. Are you doing 1080p to 1080p or 4K to 1080p encoding?
This is 1080p to 1080p on both my PC and mini.
Nice thanks much!
Thanks for providing these results. I think that the results are a bit underwhelming regarding the cpu speed. It seems to me the turbo speeds are not sustained (all core turbo boost speeds should be upto 4,3 ghz for the 8700b I guess). It would be nice for comparison sake if you could run a cinebench r15 cpu test.
StellarVixen
macrumors 68040
Underwhelming, indeed. Barely above base speed, and it hits above 90 degrees.
Sad...
Sad...
I wouldn't call it underwhelming at all! It's close to hitting the thermal limit and yet it's still operating above it's base clock speed of 3.2ghz. I'd call that amazing. And it's not throttling down to it's base speed. Also amazing.
It does jump in the 4ghz range while doing other tasks that don't continuously push the limit of the temp.
If temp was an issue it could throttle down to it's base clock speed. It could throttle lower. But it doesn't. So clearly both Intel and Apple are happy with the temps where they are at. For me the case never got hot. Just barely warm.
And it's 3x faster than my 6 core 5820 at encoding. And my PC is watercooled. So all in all i'd call it a win.
Of course this is me encoding video for hours. If you don't push it like I do you'll never see those temps and you'll never hear the fan.
[doublepost=1541776587][/doublepost]
It runs so cool at idle that you really don't need the fan. In the 20 to 30c range. Same as my watercooled PC.
[doublepost=1541776657][/doublepost]
No. Despite marketing claims most thermal paste these days is pretty much the same. Only minor differences.
Last edited:
StellarVixen
macrumors 68040
I still think that just under 100 degrees Celsius is just too much.
I can't complain. The cooling system in the 2018 Mini lets the CPU run at max temperature and TDP continuously:
3.7GHz with all threads loaded is 100MHz faster than the 2018 i9 MBP after the latest thermal throttling fix. My 2018 i7 Mini runs quieter fully loaded than my 2014 i7 Mini did when fully loaded. And it has 3x as many threads.
The takeaway from this is that:
1) Apple has properly implemented the cooling specifications for the CPU to achieve maximum in-spec performance
2) Apple is not (unsurprisingly) overclocking the CPU
It's ok for the CPU to run at 100C. It's specifically designed by Intel to run at up to that temperature.
[doublepost=1541779296][/doublepost]Also, geeze, synthetic benchmark caveats aside, this little thing is a beast:
3.7GHz with all threads loaded is 100MHz faster than the 2018 i9 MBP after the latest thermal throttling fix. My 2018 i7 Mini runs quieter fully loaded than my 2014 i7 Mini did when fully loaded. And it has 3x as many threads.
The takeaway from this is that:
1) Apple has properly implemented the cooling specifications for the CPU to achieve maximum in-spec performance
2) Apple is not (unsurprisingly) overclocking the CPU
It's ok for the CPU to run at 100C. It's specifically designed by Intel to run at up to that temperature.
[doublepost=1541779296][/doublepost]Also, geeze, synthetic benchmark caveats aside, this little thing is a beast:
Last edited:
Clock speed isn't everything. I'd still go with the i7. It will still be faster.
One video someone posted here showed a guy who was pushing his i5 and said it was very hot to the touch. Mine is barely warm. CPU temps were the same between the two so not sure why the difference in case temps.
Be careful with looking at it like that. There is no realistic scenario (whether single or multicore) where the i5 will be faster than the i7 with all other things being equal. The i7 has higher base and boost clocks with the same TDP as the i5.
The thing about hyperthreading (and where a lot of people are confused by it) is that it is less about increasing overall raw performance and more about increasing performance per watt. I know this sounds like semantics (as the two are usually interchangeable), but the way I phrased it reflects the design philosophy of the technology. Hyperthreading lets a CPU more efficiently utilize all of its inherent processing capability per clock cycle.
In a typical workload, for each cycle and core of a CPU there will be some processing units (whether integer, floating point, vector, instruction decode or I/O) that are idle because the current instruction doesn't use it. Regardless of whether those idle units are calculating or not, they are still spun up (with some exceptions) and adding to the total power usage of the CPU. Hyperthreading lets the CPU try to utilize the idle units by running pieces of other threads at the same time that use different processing units.
The net result of this is that in multicore workloads, the CPU is able to do more with roughly the same amount of TDP. That is why the 6-core i7 in the Mini is benchmarking about 20-30% faster than the 6-core i5 in multicore workloads.
TL;DR: don't look at the i5's "stable" 3.9Ghz with some workload as being comparable to the i7's "stable" 3.7GHz with a much different (and more intense workload). The i7 still has higher throughput.
Last edited:
I'm not expecting the i5 to be faster. I'm expecting it to be a little cooler than the i7 based on these findings and hoping for fans to spin up a little less often with it.
I really care about quiet operation, just not enough to get an i3.
I'm well aware of the fact that the i7 will be a little faster in cases that make use of hyperthreading.
Last edited:
I'm not expecting the i5 to be faster. I'm expecting it to be a little cooler than the i7 based on these finding and hoping for fans to spin up a little less often with it.
I really care about quiet operation, just not enough to get an i3.
I'm well aware of the fact that the i7 will be a little faster in cases that make use of hyperthreading.
Ah, my apologies - I misunderstood your context. It's a shame there's no modern, simple GUI app for OSX that allows one to selectively disable cores, hyperthreading or underclock. Then you could guarantee your maximum noise floor and not be stuck with a slower machine.
This may be up your alley though, if you're comfortable with a little bit of tinkering (and unsigned binaries):
https://geekgiant.in/how-to-undervolt-a-mac-to-increase-its-battery-life/