This thread has established that firmware updates are only applied when performed from an "internal drive" (at least as it stands today). In the 2013-2014 MacBooks, the PCIe NVMe drives show up as "external" at boot time. You can verify this in your configuration by holding down the option key when starting your updated MacBook: your new PCIe NVMe boot drive will very likely show up in an orange colour (internal drives show up in silver/grey). To apply EFI firmware updates, you will need to keep the original SSD around.I guess I thought that the flash was a one-off. Are you saying that each point upgrade of the OS is going to be followed by me having to flash or lose the ability to hibernate? Removing my logic board each time seems a lot of hassle so the £45 is probably worth it.
If you want to (or must) keep up with EFI firmware updates, you will likely need to perform the procedure for each new firmware update: Reinstall the original drive, update firmware (as part of a macOS update, or manual), modify the firmware, reinstall your NVMe drive, and update macOS. There is a possibility that Apple will allow updates to new macOS versions on systems with current EFI firmware (just like they could block updates on systems with modified EFI firmware, but that's a different story).
So, yes, the modifying your EFI firmware/system ROM can be a one-off, if you are ok to stay on this one modified version. I just want to point out there are some unknowns about these upgraded MacBooks.
That was the solution for me - perhaps thats a good clarification in the upgrade guide.Alternativly, once in the os, go to settings and then startup disk, click the lock, password, then choose the boot volume, and click restart.
Not that I have seen. The test series would not be comparable in a lot of cases (different MacBooks, different drives, different firmwares; when it comes to battery life, different load cycles). The spot checks you did with BlackMagic are in the expected range with your new SSD; if you do not have any kernel panic, your configuration is probably good for now, with this release of macOS. FIO is a great tool to baseline performance - but then we still need to agree on test that are not only comparative but also make sense for day to day work.Is there an established way to stress test the machine to make sure it's working as expected?