On 5/16/2010 12:46 PM, maccrawj wrote:
No because just as back then I know of no benchmark that will
exhaustively read all files to determine access times. It's not going
to affect creating a new file & reading it back unless the drive is so
fragmented that 1000's of non-contiguous clusters get created by the
test. Reading all the files before & after is the only real test AFAIK
and I've not done that.
We all accept that copying 1TB of small files takes longer than 1TB in
a single file due to seeks from directory to data & back having
impact. A single 1TB could still be slow if fragmentation forces seeks
for sequential data reads of groups of clusters in butterfly-like
pattern. The smaller those groups and the more they are spread out,
the more cumulative time lost. It's only ms per seek but it does add
up with the sheer number seeks inherent with reading more than a few
KB casually.
Now as to noticing, yes. Stalker:Call of Private is a streaming
content game which loads massive amounts of data at load and good
sized chunks while roaming around the Zone. Before defragmenting I was
crashing constantly during the accesses or had massive stuttering.
Despite the games other bugs these were the result of HDD data
stalling just long enough to cause the VPU recovery watchdog to kick
in (it's not happening under stress testing with FutureMark or other
games, card is good). Valve's Steam has a built-in defrag function for
similar reasons given the massive archives that make up their games.
And on the flip side of that coin, I can't remember the number of times
over the years when after defraging a hardrive a game got borked and had
to be re-installed. Just sayin..
Now is the impact going to show enough for post defrag to be
noticeable? Yes, I've never had a machine stay the same or get worse
after a defrag and the menu popups are one of the 1st things to
improve along with seeing the HDD light on for briefer pulses.
Measurable? Maybe, with the right software.
On 5/16/2010 5:40 AM, Thane Sherrington wrote:
At 06:21 AM 16/05/2010, maccrawj wrote:
That and a serious defrag using MyDefrag.
I've never seen Defrag have any impact on performance (and I did a lot
of testing a few years back.) Do you have numbers showing any
improvement?
T
