Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

GGJstudios

macrumors Westmere
May 16, 2008
44,545
943
Excessive memory pags-outs from disk = Horrible performance due to not enough physical memory. Will suck regardless of drive being full or not. Will suck more when the drive is nearly full due to fragmentation.

Drive nearly full = poor disk I/O due to writes being fragmented in order to fit in the remaining, smaller contiguous areas available on the drive.
You're missing the obvious fact that if there is no free drive space available for a page out, or for Safari to download to a cache or for the OS or other apps to use, performance will suffer, even if there is zero fragmentation. In most cases, fragmentation is not a major factor in OS X performance. Stop thinking Windows and start thinking Mac.
 

subsonix

macrumors 68040
Feb 2, 2008
3,551
79
You're missing the obvious fact that if there is no free drive space available for a page out, or for Safari to download to a cache or for the OS or other apps to use, performance will suffer, even if there is zero fragmentation.

The drive being completely full is a different matter though to be honest.
 

subsonix

macrumors 68040
Feb 2, 2008
3,551
79
Whether it's almost full or completely full, the same factors are true.

No. If it's completely full it will be impossible to fit swap files, caches and temporary storage for application (or anything else for that matter).

The factors related to decreased performance as a function of free space is are different.
 

GGJstudios

macrumors Westmere
May 16, 2008
44,545
943
No. If it's completely full it will be impossible to fit swap files, caches and temporary storage for application (or anything else for that matter).

The factors related to decreased performance as a function of free space is are different.
No, it's not different. When there is insufficient drive space available, performance can suffer, even if there is zero fragmentation present. The same scenario is true for having no space available; except performance not only suffers, but stops completely, as I said. The point is, performance can be diminished due to insufficient free drive space available, with or without fragmentation present. These are facts, not opinions.
 

Yahooligan

macrumors 6502a
Aug 7, 2011
965
114
Illinois
No, it's not different. When there is insufficient drive space available, performance can suffer, even if there is zero fragmentation present. The same scenario is true for having no space available; except performance not only suffers, but stops completely, as I said. The point is, performance can be diminished due to insufficient free drive space available, with or without fragmentation present. These are facts, not opinions.

Your opinion on this fact is, in fact, incorrect. I don't do Windows, what I discuss has nothing to do with Windows and everything to do with how data is written to disk and how VM is managed.

The fact that you think a nearly-full (Not full) disk that isn't fragmented will suffer from degraded performance that's not due to fragmentation, yet can't explain WHY you think performance will suffer if it's not fragmentation, shows that you don't actually understand the underlying systems and how they work or interact.

The reason for noticeable performance hits with nearly-full disks is due to fragmentation, whether it's file data fragmentation or VM/swap fragmentation and page in/out, it's still fragmentation. If there were no fragmentation problems then a nearly-full drive would perform similar to a nearly-empty drive. Additionally, as has been mentioned, a completely full drive causes a completely different set of problems than a nearly-full drive. And again, the fact that you don't understand this shows that you don't understand what is really going on. Sorry.
 

subsonix

macrumors 68040
Feb 2, 2008
3,551
79
No, it's not different. When there is insufficient drive space available, performance can suffer, even if there is zero fragmentation present.

And why is this?

The reason is not that there is no space available.

The same scenario is true for having no space available; except performance not only suffers, but stops completely, as I said.

It's not the same scenario.

With no space available, it just wont fit.

If it fits, albeit barely, why is it slower.

The point is, performance can be diminished due to insufficient free drive space available, with or without fragmentation present. These are facts, not opinions.

I don't think anyone is disagreeing with this.
 

GGJstudios

macrumors Westmere
May 16, 2008
44,545
943
The fact that you think a nearly-full (Not full) disk that isn't fragmented will suffer from degraded performance that's not due to fragmentation, yet can't explain WHY you think performance will suffer if it's not fragmentation, shows that you don't actually understand the underlying systems and how they work or interact.
I understand perfectly well how this works, and I've explained several times why performance will suffer without fragmentation, but it appears you either can't or won't understand. I'm not going to waste any more time trying to educate you on the facts. You'll have to do some reading and/or experimenting and you'll learn eventually that fragmentation does not need to be present for performance to suffer from diminished free drive space.
The reason for noticeable performance hits with nearly-full disks is due to fragmentation, whether it's file data fragmentation or VM/swap fragmentation and page in/out, it's still fragmentation. If there were no fragmentation problems then a nearly-full drive would perform similar to a nearly-empty drive.
That is patently false. If you cared to do some reading or testing, you would understand the facts.
I don't think anyone is disagreeing with this.
Yes, Yahooligan disagrees with it.

The fact that someone doesn't understand or disagrees with the facts doesn't make the facts any less true. If there was an apparent interest in learning the facts, I would cheerfully continue. As it appears the desire is to argue for the sake of arguing, with no regard for the facts, I'm done wasting my time here.
 

MisterMe

macrumors G4
Jul 17, 2002
10,709
69
USA
Your opinion on this fact is, in fact, incorrect. I don't do Windows, what I discuss has nothing to do with Windows and everything to do with how data is written to disk and how VM is managed.

....
Wow! Just wow. Windows HDD fragmentation--particularly in FAT-based file systems used for DOS-based Windows--displayed substantial performance degradation with a day's time even on HDDs with in excess of 50% free capacity.

I have benchmarked Macs running HFS and later HFS+ both before and after file optimization going back to System 7. In none of my 24 years as a Mac user have I ever found a measurable performance benefit above noise level provided by file optimization. I had begun to dismiss fragmentation performance hits as an urban legend until I accepted responsibility for the care and feeding of my secretary's Windows computer.

On Windows, the performance degradation of the OS is significant. The performance improvement caused by running the DEFRAG utility is nothing less than dramatic.

To claim that you are a Mac user who has no reference point in Windows is disingenuous in the extreme. Millions of Mac users run their computers for years with never a defrag. If it were necessary to defrag your Mac, then Apple would have done what Microsoft did. Apple would have shipped each computer with a defrag utility. Actually, it would added the functionality to Disk Utilities.

Among the many things that you don't understand that defrag utiliies were necessary in the Windows/DOS world due to the design failures of Microsoft file systems. Apple is not Microsoft. It does not suffer these kinds of design failures on its users.

The only way for you to see the kind of performance hit in OS X that Windows users see on their computers is to smoke some really good stuff. What are you smoking?
 

Yahooligan

macrumors 6502a
Aug 7, 2011
965
114
Illinois
Wow! Just wow. Windows HDD fragmentation--particularly in FAT-based file systems used for DOS-based Windows--displayed substantial performance degradation with a day's time even on HDDs with in excess of 50% free capacity.

I have benchmarked Macs running HFS and later HFS+ both before and after file optimization going back to System 7. In none of my 24 years as a Mac user have I ever found a measurable performance benefit above noise level provided by file optimization. I had begun to dismiss fragmentation performance hits as an urban legend until I accepted responsibility for the care and feeding of my secretary's Windows computer.

On Windows, the performance degradation of the OS is significant. The performance improvement caused by running the DEFRAG utility is nothing less than dramatic.

To claim that you are a Mac user who has no reference point in Windows is disingenuous in the extreme. Millions of Mac users run their computers for years with never a defrag. If it were necessary to defrag your Mac, then Apple would have done what Microsoft did. Apple would have shipped each computer with a defrag utility. Actually, it would added the functionality to Disk Utilities.

Among the many things that you don't understand that defrag utiliies were necessary in the Windows/DOS world due to the design failures of Microsoft file systems. Apple is not Microsoft. It does not suffer these kinds of design failures on its users.

The only way for you to see the kind of performance hit in OS X that Windows users see on their computers is to smoke some really good stuff. What are you smoking?

Clearly you didn't read the bulk of this thread and what was being discussed, nobody here, not even me, is saying the it's necessary to defrag a Mac as part of routine/normal "maintenance." I never defrag my Macs nor do the Linux systems get defragged. I also don't run my disks to near their space capacity.

Try reading the thread before going off on someone and making wild assumptions about them.
 

Lastmboy

macrumors regular
Original poster
Jan 16, 2012
125
0
In my plane of existence, the outside edge of a disc travels faster than the inside edge. Regardless, in modern drives the difference is not likely to be noticeable - even a full-stroke seek to a file is measured in single milliseconds.

OK, so now I'm curious...
As I mentioned previously, my OS drive is the internal SSD. My data drive is an thunderbolt RAID 5 array with six 2TB drives. I have 32gb of RAM installed. Therefore the issues of lack of ram, HD too full, etc. should not be a problem for me. A hardware vender told me that it's actually "bad" to defrag an SSD. Said it actually reduces their life span. I'm not sure if that's true or not, but I've never found the need to defrag an SSD. However, with the RAID 5 array... as it fills up, are each of the 6 drives writing from center toward outer edge, and are they all getting slower as they near the outer edge?? i.e. RAID array get's exponentially slower as it fills up? :rolleyes:
 

MisterMe

macrumors G4
Jul 17, 2002
10,709
69
USA
...

The reason for noticeable performance hits with nearly-full disks is due to fragmentation, whether it's file data fragmentation or VM/swap fragmentation and page in/out, it's still fragmentation. If there were no fragmentation problems then a nearly-full drive would perform similar to a nearly-empty drive. ...

Clearly you didn't read the bulk of this thread and what was being discussed, nobody here, not even me, is saying the it's necessary to defrag a Mac as part of routine/normal "maintenance." I never defrag my Macs nor do the Linux systems get defragged. I also don't run my disks to near their space capacity.

Try reading the thread before going off on someone and making wild assumptions about them.
So you are saying that defragging is not the fix for performance problems that you assert are caused by fragmentation? Got it!
 

talmy

macrumors 601
Oct 26, 2009
4,726
332
Oregon
So you are saying that defragging is not the fix for performance problems that you assert are caused by fragmentation? Got it!

All facetiousness aside, a full drive cannot be successfully defragmented, while a drive with lots of free space gets automatically defragmented to a degree and avoids fragmentation of new files by OS X. So the solution to the performance problem is to either delete files to regain free space or replace the drive with one that has a higher capacity. The solution is not defragmenting the drive.
 

MisterMe

macrumors G4
Jul 17, 2002
10,709
69
USA
All facetiousness aside, a full drive cannot be successfully defragmented, while a drive with lots of free space gets automatically defragmented to a degree and avoids fragmentation of new files by OS X. So the solution to the performance problem is to either delete files to regain free space or replace the drive with one that has a higher capacity. The solution is not defragmenting the drive.
If you go back and read Post #59, you will see that I have benchmarked the effects of defragging Mac hard drives since System 7. I have never seen a measurable improvement in doing so. That said, full optimization which is much more ambitious that a defrag can be done on a nearly full hard drive. However, it is a moot point because it makes negligible difference in performance.
 

talmy

macrumors 601
Oct 26, 2009
4,726
332
Oregon
If you go back and read Post #59, you will see that I have benchmarked the effects of defragging Mac hard drives since System 7. I have never seen a measurable improvement in doing so. That said, full optimization which is much more ambitious that a defrag can be done on a nearly full hard drive. However, it is a moot point because it makes negligible difference in performance.

I had and I fully agree!

I've never used defragging software on a Mac (or Linux/UNIX boxes for that matter) but have done effective defragging by disk cloning with SuperDuper! I don't know if any improvement was measurable but it certainly wasn't noticeable! In fact, even with Windows, where I have measured improvements, I don't bother to defrag because the time taken to defrag is greater than the time savings from doing it! And as you pointed out MSDOS was a real disaster for fragmentation problems. Nearly full drives have a problem in that to create a defragged file requires contiguous free space on the drive that may not be obtainable. In any case defragging time seems to go up exponentially with proportion of the drive being used which makes it even less useful just at the time you would seem to need it most!

My understanding is that fragmentation is only a problem when doing video work with multiple streams, and that this is best handled by having multiple hard drives rather than attempting to store all the video files on the single drive.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.