I haven't, not yet. But i've been moving so much data over to the drobos connected to it and setting things up that I haven't spent much time yet actually using it yet.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
2018 mini 3.2ghz I7 temps?
- Thread starter archer75
- Start date
- Sort by reaction score
Set the screen update and sampling resolution resolution in the power gadget to 100ms, quit everything else that is open and load a single core. You'll see the frequency oscillate back and forth between 4.3GHz and 4.5GHz.
Set the refresh too high though and that also counts as a load, so this is admittedly not very scientific
Nice! It turns out my prior test were flawed because I had activity monitor running - which itself launches 6 threads.Set the screen update and sampling resolution resolution in the power gadget to 100ms, quit everything else that is open and load a single core. You'll see the frequency oscillate back and forth between 4.3GHz and 4.5GHz.
Set the refresh too high though and that also counts as a load, so this is admittedly not very scientific
Turning it off, and running a single instance of yes, and I can start to see 3.8 Ghz (max turbo) popping up on the 4K iMac.
This 2016 4K iMac is pretty amazing I have to say. Loading up all 8 cores, and it doesn't drop below 3.7Ghz, and fan is barely audible. Only if i stress the internal GPU does it get noisy (and slow down a bit). I'll be interested to see how the 2018 mini compares.
Could these high temp reading be the reason I have had to return 2 I7 2018 mins's after running a long handbrake encode? Both failed after at 10 hour encode to h.265
100C is normal max temp for all intel cpus. Hitting that temp is not a problem. The CPU will reduce speed as required to make sure it doesn't overheat.
Tjunction is the temperature of the active surface of the processor, as detected by an array of temperature sensors that are part of the CPU's circuitry.
Tjmax is the maximum allowed Tjunction, and is hard-wired by Intel. It's one of the few safety measures in Intel CPUs that Apple hasn't been able to bypass*. If Tjmax is exceeded, the CPU first tries reducing voltage and clock speed a little bit. If that fails, it cycles the clock 8 microseconds on and 24 microseconds off, and "throttles" the memory and GPU.
For the i7 Mac Mini 2018's processor (i7-8700B) Tjmax = 100 °C.
There's another CPU safety measure hard-wired by Intel, Ttrip, which is undisclosed and varies from individual processor to individual processor, which triggers shutdown. It's generally above 120 °C and below 160°C.
Operating temperatures in the range 90 °C - 100 °C shorten the life of the CPU through _gradual_ damage. 10 hours at 100 °C isn't normally long enough to cause CPU failure on the newer Intel chips.
*(And why would Apple want to bypass Intel's safety features? To get maximum benchmark scores without fan noise in an enclosure that's on the small side for a TDP 65W processor.)
[doublepost=1542484491][/doublepost]
Neither "cooler is better" nor "hotter is better" is normally true.
That said, the computer I'm writing this on has spent a lot of the last several years running at 100 °C, and the CPU seems to still be fine (knock on wood), even though the trackpad and motherboard are getting slightly dodgy with age and abuse.
The recent Intel CPUs are remarkably resistant to high temperatures, but overclockers who push their CPUs to the most extreme limits of speed see measurable degradation of CPU function over the course of a couple weeks of sustained load.
Tjmax is the maximum allowed Tjunction, and is hard-wired by Intel. It's one of the few safety measures in Intel CPUs that Apple hasn't been able to bypass*. If Tjmax is exceeded, the CPU first tries reducing voltage and clock speed a little bit. If that fails, it cycles the clock 8 microseconds on and 24 microseconds off, and "throttles" the memory and GPU.
For the i7 Mac Mini 2018's processor (i7-8700B) Tjmax = 100 °C.
There's another CPU safety measure hard-wired by Intel, Ttrip, which is undisclosed and varies from individual processor to individual processor, which triggers shutdown. It's generally above 120 °C and below 160°C.
Operating temperatures in the range 90 °C - 100 °C shorten the life of the CPU through _gradual_ damage. 10 hours at 100 °C isn't normally long enough to cause CPU failure on the newer Intel chips.
*(And why would Apple want to bypass Intel's safety features? To get maximum benchmark scores without fan noise in an enclosure that's on the small side for a TDP 65W processor.)
[doublepost=1542484491][/doublepost]
I'd put emphasis on the word "maximum". The viewpoint that seems to be endorsed by Intel is that temperatures near Tjmax are perfectly fine for _brief_ temperature excursions under peak load. Sustained operation at Tjmax doesn't seem to be recommended. For the i7-8700B the recommended Tjunction for sustained operation seems to be 82 °C or slightly above.
Neither "cooler is better" nor "hotter is better" is normally true.
That said, the computer I'm writing this on has spent a lot of the last several years running at 100 °C, and the CPU seems to still be fine (knock on wood), even though the trackpad and motherboard are getting slightly dodgy with age and abuse.
The recent Intel CPUs are remarkably resistant to high temperatures, but overclockers who push their CPUs to the most extreme limits of speed see measurable degradation of CPU function over the course of a couple weeks of sustained load.
Short answer: While 100 °C is way too hot for sustained use, I don't think it's the reason your Mac Minis failed. See my post above.
10 hour? My handbrake H.265 encodes on the mini take 3 hours. Same settings on my PC they take 10 hours. My settings are such that the resulting video is indistinguishable from the original. What settings are you using? Or is that multiple files encoding in a batch?
I have queued up several for an encode and let it run over night. Temps in the upper 90's and no issues.
What’s the source of this claim? Is it your assumption or do you have any empirical data?
Yes, Blu-ray rips.
start with one of the h.265 presets. After I set it up I will save the preset as the default.
On the summary tab I change the format to MKV.
Dimensions tab should have cropping at automatic.
Video tab:
video encoder to H.265 10-bit (x265)
Framerate - same as source
check variable frame rate
Constant quality at 18. I change this to 20 for DVD or animated movies
Encoder options:
slow
tune - none
profile - main 10
level - auto
Subtitles I will burn in. I use makemkv to rip and I select forced subtitles. So if there are any then I burn them in to the image so there's no subtitle file to deal with. Some players have issues with these.
For audio I do passthrough. I only rip the HD audio tracks. TrueHD or DTS-MA.
What I am trying to figure out is why the file sizes end up about 1gb larger on my encodes on the Mac compared to my PC. All handbrake settings are the same.
The i7 still has twice the amount of threads as the i5. So even if they both operate around the same frequency, the i7 should perform 20-30% faster for multi-threaded applications. However, many have reported 4.2 - 4.3GHz speeds on this forum if you look around, and I've seen a few posters have reached 4.4 - 4.5GHz speeds.
So as other have stated, the 4.6GHz frequency is only possible if a single core is active. Pretty much every synthetic and real-world test I've seen on these forums are pushing their mini's to use more than a single core. Many of them are using all 6 core's. The max turbo frequency decreases with the addition of each core, so for example, on two cores the max turbo frequency the processor will allow is 4.5GHz, on three it is 4.4GHz, on four it is 4.3GHz. I've seen a few synthetic benchmarks on this forum showing a steady 4.3GHz clock speed, indicating that if at least four cores are being used, the chip is actually performing exactly as expected. The chip cannot exceed this speed (no matter how great the cooling is) unless it's using less than four cores.
On the topic of excessive heat - of course temperatures reaching 90°C is not ideal, but it's also important to note that these chips typically last a very long time. Some processors can last over 20 years, and even if excessive heat decreases that lifespan by 50%, that's still 10 years of usage under heavy loads.
Thanks for the reply, what file sizes are you seeing between the PC and Mac?
[doublepost=1542578355][/doublepost]
Are you using this for 4K discs or 1080p?
One example is Hotel Artemis, PC 4.2gb and Mac 5.3gb. Both look like the original, both with HD audio. Both settings are the same in handbrake.
I'm still testing in an attempt to figure out why the difference.
I don't know what you are doing but just wanted to say, don't take a 1080p bluray to 4k. It's stupid. It's not 4k, increasing the resolution doesn't help and your TV and/or streaming device is going to up convert 1080p anyways. Again, I'm not saying that you are doing that. I don't know. But for others, don't. It's dumb.
In the early days of 4k in the "scene" this is what they were doing. And adding fake HDR in to it too. It's pointless and just creates larger file sizes for no purpose.
can you please share your settings as I'm struggling to maintain the quality while reducing file size a lot
Interesting and relevant writeup here regarding CPU power draw and TDP rating:
https://www.anandtech.com/show/13591/the-intel-core-i9-9900k-at-95w-fixing-the-power-for-sff
https://www.anandtech.com/show/13591/the-intel-core-i9-9900k-at-95w-fixing-the-power-for-sff
I congratulate you on your fine use of handbrake , my endcode started with handbrake estimating a 2 hour completion. As time went on the estimate from handbrake grew to 10 hours and that's how long it took.
My mac mini i7 even does heat up to 70° C when it's in sleep. The case is quite warm too even though it slept for 10 hours straight.
This week I'll apply new thermal paste on the CPU. This might not fix the source of the problem (which seems to be partial software related), but at least it could help to reduce the temperature spikes a little when the CPU is under heavy load.
And (i'm not kidding) I'll saw a 14cm square hole in my table where I'll add a 5V PWM-controlled 14cm Noctua fan with air filter to cool my Mac mini from below. I'll remove the round plastic thing from the bottom so that the air can directly flow into the Mac. The Mac itselft will float on a square of door isolation foam on the table so that the air won't leave on the sides before getting into the Mac.
Another benefit will be that the air in the Mac gets filtered before. This should protect from future heat problems due to dust. Later I'll use an Arduino based PWM controller which sets the fan depending on the mac minis temperature sensors that I'll read out via SMC.
This week I'll apply new thermal paste on the CPU. This might not fix the source of the problem (which seems to be partial software related), but at least it could help to reduce the temperature spikes a little when the CPU is under heavy load.
And (i'm not kidding) I'll saw a 14cm square hole in my table where I'll add a 5V PWM-controlled 14cm Noctua fan with air filter to cool my Mac mini from below. I'll remove the round plastic thing from the bottom so that the air can directly flow into the Mac. The Mac itselft will float on a square of door isolation foam on the table so that the air won't leave on the sides before getting into the Mac.
Another benefit will be that the air in the Mac gets filtered before. This should protect from future heat problems due to dust. Later I'll use an Arduino based PWM controller which sets the fan depending on the mac minis temperature sensors that I'll read out via SMC.
That will be highly device and display dependent, so I’d recommend anyone copying your settings to test for themselves. It shouldn't be necessary unless you have a 10 bit panel and something in the chain isn’t converting the 8 bit source properly.