On 10/11/2018 5:51 PM, Flasherly wrote:
On Thu, 11 Oct 2018 15:36:09 -0700, Mike wrote:

Windows writes the contents of the working memory to the
hard disk each time you switch to sleep mode and back again as soon as
the PC wakes up. Between 2 and 8 gigabytes of unnecessary data are
written to the hard disk per process! Since SSD hard disks are very fast
anyway, this process is not necessary, but it shortens the life span of
the hard disk enormously.

TBW is the provided SSD rating. This one, for instance, I saw today
sale in the neighborhood of $40 or mo 256G Adata XPG SX850 Sata
SSD. It has an 100TBW. (Unless it was a GIGABYTE UD Pro, I also saw,
neither being stellar in the ratings of a likes of Tom's Hardware.)

So 256 gigabytes over1000 gigabytes for a terabtye written is 256000
divided by 8 gigabytes, SSD Fresh allows, or 4000 sleep cycles. If
you put Windows to sleep once a day, then a SSD with the 100TBW will
last a theoretical 10.9589041096 years before the TBW count is
exceeded and the unit no longer is able to write.

You lost me on the math. It's not unusual for my system to sleep 10 times
a day when I'm not using it heavily.
The objective is to have it sleep without writing much of anything.
It's supposed to keep it all in ram instead of writing anything to the
disk...other than housekeeping.

One thing I don't yet grasp is write amplification. If you cache a
web page, there might be hundreds of tiny files involved.

Current bleachbit preview shows 132MB in 1298 files to be deleted.
That's about two days worth of surfing. There's also a lot of stuff
with GUID identifiers. Have no idea what that is.

I have no idea whether deleting caches and all the other stuff bleachbit
deletes is good or bad for an SSD.

Another thing I fret over is what used to be called the file allocation
table.
IF that's a fixed section of the drive, how do you wear level that?
Seems like that would be the most written section of the drive.


I don't sleep, hibernate or otherwise go green. Or at least I try not
to (my mechanicals are fixed duration in their firmware, at a bare
minimum and necessity, and I haven't yet written a recursive routine
to better, say, read from them at some count just below that firmware
upper allowance).

Although I do define the swap file to a mechanical HDD, which helps
some, I wouldn't, regardless, want to be without platters, not only
for a couple small and dedicated plattered partitions, 8G possibly,
for swap files.

I don't need much ram. Been at 4GB for years. This machine has 8GB.
I put a small swapfile on a spinner drive, but It probably won't ever
be needed.

I also overprovision SSDs by a smaller factor of
ten-percent, some recommending twenty-five, while others, apparently
exclusive to newer SDDs, are already firmware "over-provisioned".

10.9589041096 years need not be subjectively enormous, exactly, but it
is in principle still about fun time thrashing on your dime, and not
the Operating System's.

My concern is that I don't understand what's going on behind the curtain.
You can manage the stuff you know about. What you don't know can hurt you.
Timestamp and prefetch are examples I learned from ssdfresh.