Right, part of where this originally came from was back in 2009/10, when everyone was uber paranoid about any writes to SSDs - so a whole set of b stupid things were recommended through ignorance; including, as the most heinous example, putting the page/swapfiles onto HDD to save on writes. Now, for those who took an active interest in discussing things (the OCZ forum used to be great for that at the time), most of this was debunked very quickly, but you still had 'proper' sites recommending things that really didn't matter (& were often severely detrimental to performance) even a year or two later. That said, as the site you've linked to are not using that argument in what they've written, let's look at the thing properly. So, firstly, to be a little more comprehensive than earlier, SuperFetch learns your usage patterns of what you're likely to use & when & does 2 key things - 1. it tells the Windows Defrag to position the bits of the most frequently used software that are actually loaded together at the beginning of the drive - which is obviously meaningless for modern SSDs (& hence why the Windows Defrag, as with, for example, reasonably modern versions of Diskeeper & Perfect Disk, simply don't do this), but is clearly useful for HDDs of any speed. [NB when the article states that "this often involves writing data to the drive that Windows is installed on", there's this, which doesn't apply to SSDs, & the small database files for the cache - so it's *not* writing significant amounts.] 2. it preloads s/w into physical memory based upon your usage patterns throughout the day -> though always giving priority to what you're actually doing at a given moment &, when necessary, dumping data out of the cache so that the machine's still using physical memory for whatever you've loaded. [NB it's a bit more complicated than that, & also does a whole set of stuff with switching priorities between programs & optimising hibernation, but they're the 2 key things to note. Otherwise it's got nothing at all to do with the swap &/or pagefile.] So, respectively - 1. it is complete nonsense to suggest that contiguously arranging the most used data will not have a significantly material benefit on a HDD in terms of loading that data - & the site's clearly not comprehended everything that SuperFetch does. 2. &, whilst there naturally is a difference between HDD & SSD access & data transfer speeds, pre-loading s/w to a ram cache to then load it from there when needed, at a low priority, which is what SuperFetch does also, is obviously slower than loading in real time a HDD or SSD. Pros & Cons. So, on the plus side then with a HDD, arranging the most used data to be contiguous & at the beginning of the drive is clearly a plus... ...& it more efficiently uses the memory in your system to pre-load what you're most likely to be wanting to use at a given time... ...as opposed to (depending on ram vs what's actually loaded naturally) potentially leaving the majority of the ram unused for the significant amounts of time. [NB So, quickly looking at the current memory usage in task manager (in Win10) - there's ~7.6GB actually in use & ~24.2GB free - but 16.7GB (of the 24.2GB) is currently used for caching s/w (ie SuperFetch & Prefetch).] Whereas the only real potential negative is that the machine may use slightly more power if it preloads data that you then don't use... ...though, with HDDs this potentially could be offset both d.t. quicker load times d.t. contiguous reading & by having preloaded more s/w that you did use & so not spinning the drive up & down as much - though i can find no comparative data on this either which way. Will You Notice A Difference With A SSD? Okay, so whilst it's faster, whether you'd personally notice the difference with a SSD is a separate matter entirely, & all i could suggest is having a play with it on for several days (so it learns your usage) & then off... [NB obviously the difference between this & my argument for not paying a huge premium for a high end M.2 or pcie SSD *unless* you've got a specific usage that can materially benefit, is that you've already bought the ram - & i'm not certainly suggesting that anyone should buy more memory solely so that it can be used for PreFetch or SuperFetch.] ...but, unless power usage is so critical that (probably) only seconds of extra battery life would be meaningful, there's nothing inherently detrimental with having it enabled. Conclusion. So i certainly would not recommend disabling it if you're only using HDDs... ...but, as stated in the previous post, with SSDs then it's choice - & the power usage difference is likely to be tiny in meaningful terms. [Edit] Whilst this is of absolutely no use for comparing SSDs with & without SuperFetch or PreFetch enabled - http://www.hofmannc.de/en/windows-7-defragmenter-test/benchmarks.html appears to give a reasonably meaningful comparison of them with a 7200rpm HDD. Now the primary point was to compare defraggers at a given time, however it also gives comparisons of no optimisation, both with & without SuperFetch or PreFetch. [NB it's important to note that most/all decent defraggers use the PreFetch & SuperFetch results for organising data on HDDs... ...&, naturally, without any optimisation then there won't be contiguous reads for any data that's then cached.] Obviously the boot times are slightly longer with no optimisation & PreFetch enabled - but loading things in advance is part of what PreFetch does. Anyway, whilst it's a little on the limited side in terms of actually testing the defraggers imho, what this does demonstrate is that the article that had been linked to was clearly wrong about HDDs.
