Yeah, you could drop out of protected mode through software on the 386 (or use virtual x86 mode.)

The A20 line hack was still emulated on modern systems as recently as the chipsets in use in 1999. No idea on current ones; may still.

--

Performance degradation over time is a general issue on SSDs, but unless you move around really huge files a lot, or are benchmarking, most people won't notice. They especially won't notice on SATA2 controllers that won't go above ~270MB/sec to begin with.

The Samsung 840 Pro I've got in my laptop presently has the problem worst of any model I've used, although the problem was pretty bad on the earlier Intel drives (X25M). Much less noticeable with the Intel 320 and various Sandforce-based drives I've used.

Probably won't hurt any, but it can, and it won't help at all.

Using the manufacturer-provided "trim free space" tool is usually good enough; Samsung and Intel are readily downloadable. I don't find it's necessary on Sandforce-based drives, although the Intel tool will work with their Sandforce-based drives (520,330,335, etc.)

It's not hard to be faster than an rotational disk, unless your main criteria is sequential write speed. For most people worrying about speed differences between different SSD models (except the very oldest) or the performance degradation is an irrelevance.

Which is talking about 4 year old SSDs.

A "degraded" SSD these days will still write at 100MB+ per second, and with an IOPS rate of around 1000. Both of these are very slow by the standards of SSDs, but the former is as good as a spinning laptop drive, and the latter about 5x better.

For most people, that's plenty good enough. If you've got particularly demanding workloads, it may not be.

3,000-5,000 writes (which is actually more the last generation of drives -- it's down to 1,000-3,000*) is an average. Blocks can fail WAY faster, or last way longer. Some spots will inevitably fail sooner; all drives have spare space.

(* using standard measures; a lot of controllers are starting to incorporate signal processing that should make the same flash usable for more writes.)

The basic effect is there on almost all SSDs, to some degree or another. The level of impact is much less visible with normal workloads on modern controllers.

Under a normal workload, no.

If you're overwriting a significant fraction of the SSD multiple times daily, maybe. I write a 20-40gb dataset out to my laptop daily. I don't worry about it.

Write limits tend to fail gradually in detectable ways.

Very good points. Also, manually partitioning so as to give a larger spare space (or even just making sure your drive doesn't fill up, and using TRIM) will decrease write amplification and improve the effective write lifetime.

In most cases, yes -- most drives provide either the total writes (which can be divided into the size of the drive) or an estimate of the usable lifetime, via the SMART variables. How to check it depends on the model -- the way it's reported as a raw SMART variable value is often unreadable.

For Intel drives, the Intel Toolbox reports it. For Samsung drives, ditto Samsung Magician.

For a lot of other drives, it's worth checking if the manufacturer has a utility, but you may need to use something like Smartmontools (http://sourceforge.net/apps/trac/smartmontools/wiki ) and figure out how to interpret the variables yourself.