Overclock.net banner

21 - 40 of 125 Posts

·
Registered
Joined
·
2,772 Posts
+rep for the thread

i am now thinking of doing this, my HDD takes too long, imho, to access saved games and such when im loading games and stuff like that. just dont know if im willing to reinstall XP for it...

do u have to have raid? or can i just create 2 partitions when i install an OS? the first partition being the OS 'drive' and the 2nd partition being the storage 'drive'

after realising there were more posts than i origianlly thought, i might try using one of those programs to move/create partitions to short stroke my HDD, but thats another day.

good info, i will be checking up on this thread
 

·
Registered
Joined
·
2,772 Posts
Quote:


Originally Posted by eflyguy
View Post

From the original post:
"There are benefits to short stroking any mechanical disk, regardless of whether the drives are used in any type of RAID array."

..a

ya this is what i thought,

i think im going to experiment with this, thanks for the thread!
 

·
Registered
Joined
·
2,772 Posts
im doing benchmarks for my 3200AAKS using HD tune before i short stroke it, ill use partition magic and see the differences.

the originals seems pretty bad... access time is terrible, CPU usage is like 10% and it peaks really low sometimes. i set this to be the most accurate testing, which took the longest amount of time

graph is attached

edit-
sooooooooo partition magic wont let you do anything unless you pay for it, ill try the other program now....
 

·
Registered
Joined
·
2,772 Posts
got the partitions taken care of so i have an 80gb OS drive and a 220GB storage drive. the access times are great and transfer rates are sustained above 100mb/s consistently on my OS logical drive and the slower storage drive isnt used much.

great thread! i guess people can see my results and show what this can do. i dont have a particularily fast HDD, but i notice loading times have decreased significantly.

my access time dropped about 5 ms (25%!!!) and my sustained transfer rates are awesome. creating these partitions and then defragging them did wonders for my system!

i dont know how to get HDtune to test just one partition, maybe i cant do that with the trial version. but if you look at the first 25% of the test (starting at the left), this is my "OS drive" and shows very consistent performance. the remaining 75% is where my "storage drive" is located. it is slower, but only needed from time to time.
 

·
FOCAL ARIA FTW
Joined
·
9,351 Posts
This is a myth. Just like setting number of processors in W7 to 4 to "improve performance".

Quote:
Bottom-line is that the average performance is a weighed arithmetic mean across all zones that is interesting for statistical and drive architectural purposes but has little bearing on real drive performance and absolutely nothing to do with e.g. controller performance. Even within the same model of the same manufacturer, different capacity drives will have different averages, which solely reflects the use or non-use of the innermost tracks and nothing else

A bit of reading before pulling the trigger won't hurt ... google is your friend.

Quote:
Run some applications that are disk limited like compression, game loads, encoding, PCMark HDD Suite, etc, show us what we will gain by using this method! This is the kind of ignorance I have battled when I was doing some work for Seagate many years ago when PATA drives with shorter platters were measured against SATA drives with longer platters and the conclusion was that PATA was noticeably faster than SATA (in that case it was Maximum PC).
 
  • Rep+
Reactions: 4.54billionyears

·
Registered
Joined
·
1,822 Posts
Quote:


This is a myth. Just like setting number of processors in W7 to 4 to "improve performance".

Sooo you are saying that short stroking doesn't increase a mechanical HDs performance in any of it's parameters?
 

·
FOCAL ARIA FTW
Joined
·
9,351 Posts
Quote:


Originally Posted by Old Hippie
View Post

Sooo you are saying that short stroking doesn't increase a mechanical HDs performance in any of it's parameters?


If your metric is the average between fastest and slowest and you eliminate all but the very outermost tracks, then your drive will, on average, look faster on paper. If you have too much extra unwanted space you don't need, you decrease the drive size to only use the outermost drive tracks hence reducing seek times by a large margin, which means the performance perceived where any seek is involved, is enhanced. The actual drive performance remains largely fixed but the faster performance comes from reduced access latencies with the much smaller head movement involved. Synthetic apps will in these cases largely exaggerate the benefits but they are not apps I rely on since I don't run them as my work. Simple heavy read/write daily operations will be enough to show me any benefits I might gain from such a technique. These benefits are not on par with what synthetic apps are saying.

This "tutorial" like the Tom's hardware article is misleading and gives a false impression that a hard drive "with short platters" is faster than a hard drive with "longer platters". What about the controller lads? By that logic, my example with an analysis done by maximum pc still stands :

Quote:


... when PATA drives with shorter platters were measured against SATA drives with longer platters and the conclusion was that PATA was noticeably faster than SATA (in that case it was Maximum PC).


We've stumbled across a PARADOX! The PATA controller was never faster than the SATA controller, yet the "short stroking" process on the PATA drive showed the opposite? Some people should research a bit more before believing every trick out there.

This (short stroking) is actually an extremely commonly believed online "enthusiast" myth. You'll see all sorts of strange "complex looking" explanations given by purported "computer engineering/physics experts" on fora on this topic

IOMeter due to it relying entirely on the IOps metric is one of the only benchmarks to show high gains, since it supposedly relies on data access times to generate its results. Honestly, I am really suspicious of this benchmark - its throwing out very strange and uselessly synthetic MB/s results in many of my tests. Result fluctuations which real world tests do not corroborate.
 
  • Rep+
Reactions: 4.54billionyears

·
Premium Member
Joined
·
7,161 Posts
I have been reminded that this is OCN and we search for performance gains in all systems. But I ask, is any gain in performance (even .1 of 1%) worth any amount of effort?

What I am getting at is can we qualify when the gain is worth the effort, somehow?

Don't get me wrong I do like exploring these type of modifications as they increase my overall understanding of how systems work and are a lot of fun.
 

·
Registered
Joined
·
1,822 Posts
Quote:


If you have too much extra unwanted space you don't need, you decrease the drive size to only use the outermost drive tracks hence reducing seek times by a large margin, which means the performance perceived where any seek is involved, is enhanced. The actual drive performance remains largely fixed but the faster performance comes from reduced access latencies with the much smaller head movement involved.

This is a great explanation short-stroking and by it's own admission seek speed is enhanced.

Not all drive parameters can/will be enhanced with this method but it's doing what it's supposed to do.

IMHO the only failure to short-stroking would be the expectation that it would make a magnitude of difference.

PCCstudent has hit the nail on the head....how much effort vs how much gain makes it "worth it" to you.
 

·
Registered
Joined
·
1,501 Posts
Quote:


Originally Posted by Wishmaker
View Post

If your metric is the average between fastest and slowest and you eliminate all but the very outermost tracks, then your drive will, on average, look faster on paper. If you have too much extra unwanted space you don't need, you decrease the drive size to only use the outermost drive tracks hence reducing seek times by a large margin, which means the performance perceived where any seek is involved, is enhanced. The actual drive performance remains largely fixed but the faster performance comes from reduced access latencies with the much smaller head movement involved. Synthetic apps will in these cases largely exaggerate the benefits but they are not apps I rely on since I don't run them as my work. Simple heavy read/write daily operations will be enough to show me any benefits I might gain from such a technique. These benefits are not on par with what synthetic apps are saying.

This "tutorial" like the Tom's hardware article is misleading and gives a false impression that a hard drive "with short platters" is faster than a hard drive with "longer platters". What about the controller lads? By that logic, my example with an analysis done by maximum pc still stands :

We've stumbled across a PARADOX! The PATA controller was never faster than the SATA controller, yet the "short stroking" process on the PATA drive showed the opposite? Some people should research a bit more before believing every trick out there.

This (short stroking) is actually an extremely commonly believed online "enthusiast" myth. You'll see all sorts of strange "complex looking" explanations given by purported "computer engineering/physics experts" on fora on this topic

IOMeter due to it relying entirely on the IOps metric is one of the only benchmarks to show high gains, since it supposedly relies on data access times to generate its results. Honestly, I am really suspicious of this benchmark - its throwing out very strange and uselessly synthetic MB/s results in many of my tests. Result fluctuations which real world tests do not corroborate.

Better access times are still a good result of doing this. Of course its not close to what ssd's produce but wouldn't you say this improvement does help?
 

·
FOCAL ARIA FTW
Joined
·
9,351 Posts
Quote:


Originally Posted by eflyguy
View Post

It does help. Either he's intentionally trolling, or completely missing the point.
Move on. Nothing to see here..

..a


Insulting me will not change the fact that you did not research more thoroughly about this issue. I took the time to read what you wrote, a well raised person would actually read what other people write too. I showed with clear examples and explained in a few posts short stroking and real life benefits. You saw something about short-stroking, you saw the Tom's Hardware article and you wrote this up. I give you points for the effort but not for the validity of the tutorial. You are misleading people and being cocky won't change this. You can continue to be ignorant and live in your little world. What is sad is that some people come to OCN and learn crap like this.
 

·
Premium Member
Joined
·
18,611 Posts
OK, so when I next install Windows, I can do this on my Samsung Spinpoint F3 1TB?

What sort of size partition are we talking I have to create? Can I do this in Windows installer?
 

·
FOCAL ARIA FTW
Joined
·
9,351 Posts
Quote:

Originally Posted by Old Hippie View Post
Might as well report me too cause I agree with him!
LMAO!

Short stroking is a very easy concept to understand and just because your opinion doesn't coincide with the synthetic apps results....

doesn't mean everyone agrees with you.

You'd better report this post also!


At the end of the day, if you lot want to fall of the cliff nobody is stopping you. You seem to be another user who does not read what other people post. If you use synthetic apps all day long, then yes, you will see the short stroking gain in the higher numbers when the bench finishes. Start using real world applications where the "short stroke" gain is not inflated (this is what synthetic apps do) and things are different. Not for the blind of course ...

While you are at it, make sure you tell windows how many cores you have because it will "significantly" reduce your boot up speed and make windows run faster!!!


***Another myth and eventhough it was debunked by Microsoft people still believe it. "Short-stroking is no different : myth, folktale with no tangible gain in THE REAL WORLD. If you lot bench all day then I advise you short stroke everything. Also, use IoMeter because this is the software that inflates the scores the most when you short - stroke.****
 
  • Rep+
Reactions: Novelty_Glasses

·
Premium Member
Joined
·
12,297 Posts
Please guys, don't argue - debate.

If you've got proof on either side, then post it. So far from reading this thread, I'd be on the side that short stroking does increase performance simply because the benchmarks show that it is higher, and I haven't (yet) seen any proof that these benchmarks inflate the results.

There is also no need to tell people you've reported their post; it just leads to flame wars.
 

·
Registered
Joined
·
1,822 Posts
Quote:


While you are at it, make sure you tell windows how many cores you have because it will "significantly" reduce your boot up speed and make windows run faster!!!

I have no idea what you're talking about but maybe after my cliff fall it'll probably become crystal clear.
 

·
Registered
Joined
·
1,501 Posts
Quote:


Originally Posted by Old Hippie
View Post

I have no idea what you're talking about but maybe after my cliff fall it'll probably become crystal clear.

lol he's talking about the boot options in ms config.
 

·
Registered
Joined
·
495 Posts
Well, just to add my two measly cents:
I am actually going to support Wishmaker.
Although I concede that benches show that short-stroked HDD's are faster, in my experience and understanding, it does not really make a difference. I actually tested the difference with a timer some time ago (lol, yea, I know, ghetto). I had my family rig setup with a single large partition, timed the boot up and the loading times of a couple of games (repeated it another time just for certainty) and wrote down the results.

I then used GParted to make my partitions smaller (just to the very limit of what I was using, maybe 50MB spare) and rebooted for the timings. After loading Windows and the same couple of games a couple of times, I checked the times... and they were virtually identical. I think that I even had a result where the slowest short-stroked time was slower than the fastest regular time.

Anyway, this made more or less sense to me, because if I recall, when GParted made my partition smaller, it basically took 10 secs to do it. It didn't move any files, it just placed the delimiter of how much space was allocated to the partition to a earlier spot on the HDD.

Anyway, I know that is a really sketch test (on my families Dell XPS desktop, not my sig RiG) and probably doesn't mean much to you guys, but there it is anyways.

I actually came back to this thread with another purpose in mind.... does anybody know how to modify partitions with GParted but when playing around with a RAID 0 array? I am trying to get a dual boot on my new RiG, but GParted doesn't seem to recognize the array. It doesn't even see Windows 7, it just tells me that there are 2 blank HDD's....
 

·
Premium Member
Joined
·
4,496 Posts
Discussion Starter · #40 ·
Quote:

Originally Posted by Wishmaker View Post
I showed with clear examples and explained in a few posts short stroking and real life benefits. You saw something about short-stroking, you saw the Tom's Hardware article and you wrote this up
You quoted other peoples comments and provided no links to the source.
Your first comment in this thread was "it's a myth".
You provided no data of your own.

I, on the other hand, created an array with old and slow Maxtor DiamondMax drives, shortstroked to 10% of available capacity, and benchmarked to demonstrate a significant reduction in access times, which is the primary benefit of short stroking.

I have never seen a TomsHardware article on this topic - I got most of this from a member here, as clearly stated in the original post:

Quote:

Originally Posted by eflyguy View Post
Thanks to the_beast, who has posted much of this information across multiple threads over the past few weeks. I wrote this by distilling a bunch of his posts down to a single document..

Finally, The second point to this thread is to explain to people what it is, and how to set it up, driven by the multiple posts per day asking this exact question. I'm not advocating everyone sacrifice a significant portion of their hard drive to improve performance.
..a
 
21 - 40 of 125 Posts
Top