Disk has to be in sufficient repair (and config) so as to be accessible by DOS, and of course - BIOS initially! (then via the small multithreaded OS within SR). SR then checks memory with some difficult bit-patterns, checks HD controller and decides which ECC loops it can disable, whether there's an "engineering cylinder" etc. - can't recall if it looks at FAT lost, linked chains etc. (?) I think.. but not sure that, if there's a FAT anomaly, SR will send you out to repair that (via chkdsk or the others you mention).

I gather from much past lore that Win is particularly crash-prone on any cluster containing a flaky byte/block - which hadn't been marked Bad. In one case I verified this re a local install which was failing in bizarre ways. Had him run SR, which found a bad cluster or maybe a couple. Install proceeded OK afterwards. Not Proof of the posit, but close enough for me.

Yes, SR is mainly about massaging magnetic domains (and also recovering data from now-Bad places) via iteration and in a neat built-in utility which you can watch doing its recovery thing. In course though, it can find marginally-bad controllers or weird main memory (or HD cache mem maybe).

I haven't personally needed that "DynaStat" level of recovery but have a couple first-person hearsays about its utility - one HD brought very near a powerful electromagnet, for one. All the data restored - along with the lo-level and hi-level formats ie it would boot normally (so MBR too, if that Was damaged!)

My allegiance to it is in the fact of its literally rewriting ALL the magn. domains in each cylinder = a fresh lo-level "reformat" along with all the rest. This is the only *practical* lo-level rewrite I can imagine accomplishing re today's HD configurations, and the lack of tech info as would let you do it the hard way (which natch destroys all content).

As to the effectiveness of the magnetic domain "scrubbing" Steve claims - you'd have to read his blurbs and decide for self. I think SR's largest SYAs were in the era of linear head positioners which drifted or got sloppy, thus screwing up the S/N of the output signal. In 'voice-coil' head drives of today, feedback positioning centers the head on-track. But the magnetic domains are where the data lives - so SR ain't obsolescent even.

If Steve never writes anything else very clever, SR is his Pulitzer IMhO.


Ashton