My iMac Pro keeps suddenly shutting down every now and then.
At first I thought I was drawing too much power from the outlet.
But I get the error code “-108” in the terminal. I guess that stands for “too little RAM” or something.
I suspect the graphics RAM, since according to iStats 98% of it is used consistently. Now I have set the graphics memory consumption to 40% in the Flame Setups. Just to test it.
But as soon as a batch is opened, the RAM fills up again to 98%.
Does Flame Setups not work?
Or has anyone experienced something similar?
Most of the time the iMac just comes back after something like that and I can continue.
But every now and then I have to boot it up properly and restart the programs.
This makes me a little worried about the data integrity.
I would reset the PRAM and SMC first to see if that resets the hardware. Test it across different batches on all versions of flame, and hopefully that fixes it. If not you need to take it to the apple store because that’s the only place that can fix the imac pro. Overall it’s not ideal to use one of these machines but if you have to make sure you’re backing up your data regularly.
95C is not too hot if the CPU is working hard. It’s pretty hot if there is nothing going on, however. My non-flame imac idles at 52C. On the lower edge there are a series of vents. Turn everything off and run a vacuum cleaner hose across those. That is the fluff that Marcus_M was talking about. You can’t really see it at a glance.
Also, I run an app called smcFanControl. It gives me a temp readout and it allows me to turn the fans up more if need be. Having a desk fan blow on the back of the machine helps keep the temp down as well is you are in a hot environment.
I would definitely save and back up your projects. Always best practice. I would do some reading about SMC and PRAM online since it’s usually the first thing to try when troubleshooting a mac. You usually hold down certain keys when the computer is restarting and it resets electrical and hardware functions.
Again, applecare and being able to take these machines to the applestore is very important for imac pros. Some of the common problems with these computers aren’t fixable, so if something goes wrong then it’s kind of like your car getting totaled and having insurance pay for it or not.
Also be sure to use a battery backup, also best practice.
I was able to narrow down the problem to the external monitor. When this is not connected, everything runs fine. Flame even renders faster.
It doesn’t matter if it is connected directly to the iMac via USB C or DP, or via TB daisy chained through one of the raids.
I suspect that the macOS own scaling for third party monitors together with large flame batches is too much for the graphics card. But something on top has to be incorrect. Friends have pretty much the identical setup and don’t report such problems.
A technician advised me to install macOS fresh. If the problem persists after that, I’ll have to take it in for service.
But until then, I can at least work. Even if only with one display
On the temps - CPU max generally speaking is 100degC, though keeping them at 95-96degC is as high as I would go. GPU may be different. A whole lot of GPUs list 85degC as the max. Either way it’s up there. That said, both the CPU and GPU monitor their own temps are are supposed to throttle workload to not exceed limits.
The external monitor maybe a symptom, not a root cause. Taking it off, lessens the workload on several components and pressure on memory utilization, the GPU included. So disconnecting the monitor may just keep you below the critical threshold indirectly.
Macs generally don’t do well with heat and video workloads. I was always under the impression that Apple designed for graphic designers, not video editors. There have been cases in the colorist world where the workload easily exceeds the design limits/stress tests Apple aims for. It’s worse for the laptops and iMacs, which prioritize form over heat management. Back in 2016 I lost a MBP because keeping the lid closed and two external monitors got it hot enough during long renders that the GPU expanded and contracted from the temp changes that solder eventually cracked and caused random system crashes. Apple eventually issued a recall on the issue, but it’s been a long struggle. Maybe one of the reason they’re happily pushing the M1 because the can control the whole system better.
All said, high temps can damage your hardware and eventually lead to crashes. In themselves they shouldn’t cause a crash as the system self-monitors. Out of VRAM is much more likely to cause at least an app if not a system crash, and there is less infrastructure for allocating and sharing it. It may be worth see if you run multiple apps that are competing, or if there are settings you can use to reduce VRAM usage.