I am seeking advice on investing in a new machine that does away once and for all with apparently GPU-related crashes and shutdowns. The support team has now told to me that, after reaching out to Apple with regard to suspecting AMD processors as the culprit, Apple confirmed that the issues are indeed caused by these GPUs and they are working on a solution, which will probably require some time.
1. The bug / behavior you have encountered
Constant crashes (freezing > screen going green > app closing and having to log back into my user account, no usable file left)
as well as complete shutdowns, with the iMac just turning off. Have to hard reboot.
This issue happens across all models, input and output formats regardless of length or other processes running in the background.
2. Your system profile
2020 iMac 27’’ Retina display
AMD Radeon Pro 5400 4 GB
40 GB RAM
3. Your log files (Help > Logging > Get Logs for Support)
Shared these with the support team already on a regular basis for a long time.
4. Any screenshots as necessary
With the issues getting worse and worse and many hours of processing time wasted, I am thinking of ending the whole hassle by getting myself a Mac mini M2 as a secondary machine, exclusively for running my VEAI projects pretty much 24/7.
Now here comes the question: From my understanding the Mac minis use a built-in GPU specifically designed by Apple. So without AMD, this should safely eliminate the issue, right? BTW, it does NOT seem to be caused by overheating because the crashes persisted in spite of using a fan acceleration tool keeping my GPU temperature consistently below 90.
Has anyone already done the test and is able to confirm? Should I spend the money? Or is there an even better solution for a new machine?
Hmm OK, so if I get you right you are saying a Mac mini would be a safe bet because it eliminates the overheating issue by design? I am more than willing to believe that, BUT:
I have had crashes or shutdowns too when my CPU/GPU monitoring apps didn’t record an unusually high temperature. Also, I’ve had successful encodes with the temperature suspiciously high (like 95 °C+) throughout.
In short, I am still unsure the heat is really the issue. Or could it be just very short temperature spikes instantly causing crashes without anything fast enough to record them? I mean, if the monitoring app crashes too, there will be no report displayed afterwards showing me the death stroke …