I would have several suggestions, and not all are critical or "future proofing" but I feel are needed features anyway. Note that these are not thrilling ideas so much as generally sound ideas for the market:
1. Continue the trend of safely covering the electronics on motherboard PCBs, SSDs, HDD, and other electronics so ESD handling practices become less critical. If you read in PC parts reviews about the number of DOAs, you can get a feel for how many ESD events are happening. The SATA SSDs got better reliability from the onset not just because of no rotating parts, but people could not touch the electronics! I have read that the M.2 SSDs have a lower reliability over their SATA SSD cousins. It is a high probablity that is due to the exposed PCB parts. All of these failures for motherboards and parts that get returned drive up the cost of for all of us.
2. More thought into signaling to what in the past have been "dumb" subsystems such as the PSU. If the PC makers would do a full FMEA study, they would be able to identify what scenarios need better protection. For instance, if the PSU detected that it had a power loss, it could signal the motherboard to make mitigations. The CPU and other hardware could draw back power usage, and storage devices could make a last save and then go to idle.
3. DRAM ECC across the board for all systems. Intel has used this as a cash cow for so long that people readily accept that it is a cost adder (like: brainwashed). It isn't! The circuitry needed for this is laughably simple. In almost every other subsystem in your PC, there is error correction already there: HDDs (disk to controller), SATA interface (parity), PCIe bus, SSD (ECC and RAID), and nearly all communication systems (Ethernet, USB, video, etc.). Note that DDR5 has it built into the standard at the die level, so this is coming soon anyway. AMD has this as standard on many of their CPUs, and it has been their philosophy for decades. I built a desktop Athlon system in 2006 that had ECC. Worked great.
4. Like above, add device statistics like SMART to detect failures are accumulating. If there are memory errors occurring at the same address, the OS should let the user know. If core 3 of the CPU keeps having to shut down on thermal and there no heavy load, the user should know. And SMART errors creeping up in the HDD should also be communicated to the user.
5. Subsystems should be "hardened" against cyber attacks. This means that the hardware philosophy should be to not trust the other hardware on the PCB. If your audio CODEC chip is trying to flash the firmware in your USB controller, it should not be allowed. What can the user do that is in hindsight not a good idea, but they do it anyway? Pick up a USB drive in the parking lot and plug it in - and it bricks your whole system. How: it dumps negative 200VDC into your +5V bus. D'oh.
6. I believe the PC industry should consider marketing to gamers and overclockers differently than to PC builders / hobbyists. With some much of the new bios functions going into tweaking CPU and memory timing, very few other features are being added for hobbyists. I got a used Dell and was surprised at the features and testing that could be done within the bios itself. HP is the same way.
I know, not a flashy list, but there are some very usable ideas there anyway.
- Charles