Short Stroking Question

What is the optimum space percentage for a shortstroke partition? Is it 10% or 20%?
For example if I were using the 2TB WD Caviar Black, how much space should I allocate to the short stroke parition?

10% would be 200GB
20% would be 400GB
30% would be 600GB

I think the point of shortstroking is to ensure that the outer faster part of the drive is read, right? Then I could use the rest of the drive, the inner portion as storage space. Correct? Also what is the best software to use for shortstroking a Hard drive, or will just partitioning it be enough to get the effect I want?

The optimum is 20% to 25% IIRC.

It's simply partitioning the drive. Any partitioning software you want to use will do just fine.

The smaller you make the partition, the faster it will be. You're limiting the physical area that the head has to cover. The less area it has to cover, the quicker it can cover the whole area.

You can use the rest of the space for a storage partition, but keep in mind that any time you're accessing that partition, you're inherently preventing access to the fast partition. To get the most performance out of your short-stroke, you shouldn't use the rest of the drive. If you have free space at the end of your fast partition, remember than any data on the other partition will be forced out that far on the disk too, actually making it slower than if it were right up next to the other data.

Another option is to do a sorted defrag with JkDefrag/MyDefrag. This will allow you to put all the files in a directory physically next to each other. That means that when you're accessing WoW's files, the head will stay in one particular area. It doesn't help if you're trying to access files in both C:\Aardvark\ and C:\Zebra\, but it should keep all the files in any given dir physically close to each other to help reduce the amount of time the read/write head spends moving around.

Thanks for the advice. I've decided to get a WD Caviar Black 2TB, and create a 500GB SS partition on it. 25% should be pretty fast and give me enough room to install a large number of games.

Is there really any benefit to short-stroking a drive like that any longer? How much faster is it really going to be?

I don't think you are going to see a performance increase by short stroking the single drive. If you short stroked it, and had it in raid0 with 2+ other drives then you would probably notice it.

Get HDTune. Install it. Reboot if you're testing the system drive in place (already installed with the OS on it) - if the drive is still uninstalled/unmounted, perfect. Mount the drive and attach it, don't format it, don't partition it, leave it untouched.

Run HDTune's basic benchmark - do NOT choose the short stroking option, you need to test the entire drive's capacity. One suggestion: when HDTune is opened, click File - Options - Benchmark and then adjust the slider to the lowest setting which is "Accurate." The testing will take longer but, obviously you'll get far more accurate results; this helps cut down on the "spikes and valleys" of the graphing going up and down like a mountain vista or whatever.

When the test is done, take a look at the resulting graph, look at the part from the dead beginning to the part where the speed drops below a few points from the max, almost always it'll be the first ~15-20% of the drive depending on capacity (if you short stroke the test to that much, you'll end up with a nice nearly-flat read speed which is why you can't short stroke during this test).

Use that as the partition for the system, for the OS, the stuff you access the most; set the rest up as a secondary partition for data storage primarily, maybe some games, etc.

Here's an example:



That's the laptop drive I'm using presently, a Hitachi 250GB SATA II 7200 rpm drive in this older Toshiba Satellite. As I noted with red circle, that's the highest speed at the beginning of the drive going in about 20-25% deep (to the ~60GB point, actually about 67.5GB but I cut it a bit short to keep it even at 60GB). Now look at the same drive tested with the short stroke option set to 60GB:



Aside from that nasty ugly hideous drop near the very end, the drive gives me sustained rates above 60MB/s in that range from 0 to 60GB, and that's why my Windows partition is 60GB. The rest of the drive is the secondary partition for basic storage, etc.

Currently I still have 28GB free on my C: drive so... I'm not hurting for space, with a 2 month old Windows 7 Pro x86 installation and pretty much every application I use daily including a 17GB Ubuntu 10.04 installation on the same partition using Wubi.

And also, note the random access time - nearly 33% faster, down from ~16ms to ~12ms. It's a laptop so I'm stuck with a single drive, but I'd rather have it set up this way for a variety of reasons, including faster OS performance overall. My most used apps on a daily basis all sit on C: for very fast access; the data those programs create or access when they're in operation are all on D: for safe-keeping.

Nice nearly flat graph, not too bad for a mechanical drive, eh?

<ok, for the nitpickers I re-did the math, it's about 25% faster since 16 - 25% = ~4 = ~12 or so... I hate math, but even so, it's still faster...>

Short stroking is purely for performance. Its greatest benefit is used mainly with large raid arrays. You do not use the rest of the drive for storage.
Posted via [H] Mobile Device

Bahamut said:

....)

Click to expand...

NO wonder I got same benchmarks.. lol, I don't have PRO.

I have tried short-stroking a 7200.7 120 GB Barracuda to 30GB and then simply creating 30GB partitions on it. When measuring the difference with HD Tune, there was no real difference between the short-stroked drive and first partition of a partitioned drive.
For ordinary use, I think partitioning will be probably better.

delete.

I have some pics I`ll attach later but the latency def. drops when you short stroke a considerable amount. And the WD Blacks are def faster than the normal/older WD hard drives, and even blow the old raptors out of the water.

The Shortstroke partitioning seems to work on my new 2TB WD Caviar Black (WD2001FASS). Here is the result on CrystalDiskMark.



The formatted size of the partition is 465.75 GB (25% of formatted total capacity). The secondary partition is 1397.26 GB, but I haven't put anything on it yet.

According to Hardware Canucks, the average read speed of the 2TB Caviar Black is 115 MB/S. So that's approximately a 27% increase in average read speed on the SS partition.

What happens in real-life when you start adding data to the second partition? Can you do some testing before and after you add set up the second partition for use? I'd be curious to see how those improvements hold up over time.

Forceman said:

What happens in real-life when you start adding data to the second partition? Can you do some testing before and after you add set up the second partition for use? I'd be curious to see how those improvements hold up over time.

Click to expand...

Yeah I think I'll post another bench after I fill in the storage parition some and see if the SS partition speed still holds up. I think it would because it would still be accessing the fastest outer 25% part of the drive right?

Oh here's an HDTune benchmark. Seems that the average speed is only 136MB/S but it's still 21MB/s faster than the whole disk alone. So that's what? 18% increase? Oh and the access time is only 8.3ms on the SS partition compared to like 12ms on the full drive. That's a nice improvement in access speeds.



It's still significantly faster than the velociraptor at 110 MB/S.
It would be pretty impractical to SS a velociraptor since they don't have a lot of storage capacity in the first place.
Though with the new 600 GB Velociraptor you could SS 25% of it for a 150GB hyper faster HDD. But then you're running into storage capacity limits again.

On a typical mechanical hard drive, if you set aside the first ~20% for your OS as the system partition, and the rest as a "storage" partition or whatever, you're going to find that the actual read/write heads spend well over 90% of the time they're floating above the platters ('cause they don't touch, of course) will be floating above that ~20% of the drive. The "virtual short stroking" that we do by creating such a partition (it's not real short stroking since we do intend to use the rest of the space on the platters, aka the entire drive capacity for something) does increase the performance of the operating system itself, sometimes rather dramatically.

It might fly in the face of whatever logic you choose to abide by, The Dude notwithstanding, but it does help things quite a bit, especially with the faster access times because of the shorter travel.

No, it doesn't truly emulate or effect actual short stroking - where you'd set the ~20% on the drive and that's it, the rest of the drive is not put to use for any data access whatsoever - but it's a simple quick way to improve system and operating system performance on the hard drives that many of us continue to use.

I have the same question:

I just purchased a SAMSUNG F3 1TB.

Is short stroking it 25% (250GB)of its capacity (1TB) enough to get get performance?

Also any particular HDD utility to achice this is better (or easier) then one another , or just simply use the Windows Vista install utility that lets u partition your disk before installing the OS?