John Galt wrote:
I doubt there is any way to segregate any particular hardware address space in macOS due to its implementation of ALSR since Lion (at least). Apple used to discuss it in Apple Platform Security but that passage no longer appears.
Memtest86+ and Memtest86 (two different utilities, first one is open source, the second is from a commercial company) are not running within macOS, but from a bootable USB stick which only has a bootloader to put the memory test utility into memory. These utilities are only utilizing about 10MB of RAM for the actual utility. They are also not subject to macOS in any way so those ALSR concerns are irrelevant.
History of Memtest86 and Memtest86+:
https://en.wikipedia.org/wiki/Memtest86
There is another "memtest" command line utility available which can run within macOS (most people will use Rember which provides a GUI wrapper...very outdated version of the app too...it is the only pre-compiled binary version I know of for macOS). This utility that runs within macOS will try to reserve as much contiguous memory as it can for its test area (user can specify the amount of RAM as well, but the utility must still be able to reserve a contiguous memory zone). The memory test run within macOS is extremely limited because only a small portion of memory can actually be tested plus there is always a chance another app could possibly modify memory within the test zone which makes it much less reliable. I don't know if ALSR has any effect on this utility running within macOS since I've rarely seen (perhaps never seen) any memory test utility report a failure when run within any OS (macOS/Windows/Linux).
I have found that Memtest86 (non-plus version) has been pretty good for checking memory. Of course, there is always a chance of a false positive which is why it is important to be able to test another identical computer to see if it also has the same error.
I have utilized Memtest86 (non-plus version) (v7.5 and v8.x, v9.x) on 2015 - 2019 Apple laptops without seeing any issues probably because these Intel systems are pretty much the same as non-Apple systems since the memory controllers are on the CPU these days lessening the chances of odd issues here.
I have had some recent computers (both Apple & Dell) where I suspected a memory issue, but could not prove it since the Apple Diagnostics passed. The problems with these computers were intermittent. Some times it will report a memory failure with a bit flip just like the OP showed here, but other times the test will just freeze the system which I have found is just as useful to identify a hardware issue (especially if it is easily repeatable) as long as it can be confirmed another identical model does not have the same issue with the test.
I'm a little skeptical of Memtest86+ right now since developement on it only resumed a few months ago (after being dormant for a decade), but if Memtest86 also shows the same errors, then I would think it is safe to say that the memory on this Mac is bad. The Apple Diagnostics utilize similar test algorithms which Memtest86 and Memtest86+ also utilize. These apps generally write a specific pattern to the memory area, then confirm the memory still holds the same values at a later time. Sometimes the check is made quickly, while other times it may wait minutes to make sure the memory cells holds the pattern. Any potential programming/test issues would revolve around using any special characteristics of the memory controller. Basic memory test functionality has been around for decades, but has evolved into a few more advanced memory tests.
I never put 100% faith into any diagnostic app or test, but they can be a useful tool when you are having intermittent odd issues and everything else has been eliminated. It is nice being able to confirm a hardware issue instead of wondering if you have overlooked some software or configuration issue.
If Memtest86 also shows memory errors, then I agree with the OP that this Mac very likely has bad memory. If only Memtest86+ is showing these errors, then I would suspect a possible issue with that utility.
The results of Memtest86+ do seem a bit odd to me since the supposed bit flips seem a bit random which is unusual since most memory failures I have seen will tend to have the same bits failing. The bit either sticks on or it sticks off. I don't see that in the results shown by the OP.