Discrepancies between defraggers?

[Megalith]

Does anyone know why there would be massive discrepancies between defragmentation software? For instance, the Windows 8 native defragger tells me that one of my drives is 0% fragmented. However, Piriform's Defraggler tells me that drive is 100% fragmented. What?

I also can't help but feel that defragmenting is a huge waste of time. Or worse, shuffling all the data could reduce the lifespan of my drives.

 

[drescherjm]

It's a total waste of time on SSDs. On hard drives it may have some benefit.

However, Piriform's Defraggler tells me that drive is 100% fragmented. What?

That is crazy. Every file fragmented does not make sense since there will be files of 1 cluster which by definition can not be fragmented.

 

[cyclone3d]

I really hate the misinformation that almost everybody spreads that defragging an SSD is pointless.

SSDs read/write faster sequentially than they do randomely.. just like a regular HDD.

Windows 7 and 8 will not defrag SSDs. & will let you do it manually, but 8 will not let you do it at all.

If an SSD is highly fragmented, you can measure the boot time, defragment it, and then measure the boot time again. From my experience, the boot time after defragging will be noticeably less than before defragging.

Programs will also load faster as well after defragging a highly fragmented SSD.

Windows also has issues sometimes when drives (SSDs and HDDs) are highly fragmented. Anything from programs hanging, to crashing, to the computer having trouble even booting.

The program I use is a free one, called MyDefrag. I only run it manually every so often. SSD life is not going to be noticeably affected by it unless you have a super old SSD that had a shorter life than newer ones.

 

[patrickdk]

It also depends on what you mean by fragmented.

Do you mean that there is a single part of the file, that is out of order.

Or do you mean, there is fragmented freespace, and reorganizing every file, to make all the freespace into a single stream, would be best.

Personally, these days, with ntfs, you should only worry about defragmenting files, unless you want to shrink your partition/disk.

 

[SirMaster]

SSDs read/write faster sequentially than they do randomely.. just like a regular HDD.

Windows 7 and 8 will not defrag SSDs. & will let you do it manually, but 8 will not let you do it at all.

But with many modern SSDs you cannot tell the SSD where to physically put sectors of data (without special tools from the manufacturer). The internal firmware that performs wear-leveling tells the OS it put the data in sector 1 for example, but secretly it put it in sector 123503 because that was the least written to secor. Windows thinks all the data is sequential, but in reality its all over the place physically on the drive due to wear-leveling.

 

[cyclone3d]

But with many modern SSDs you cannot tell the SSD where to physically put sectors of data (without special tools from the manufacturer). The internal firmware that performs wear-leveling tells the OS it put the data in sector 1 for example, but secretly it put it in sector 123503 because that was the least written to secor. Windows thinks all the data is sequential, but in reality its all over the place physically on the drive due to wear-leveling.

Maybe so.. but that doesn't change the fact that it reads faster "sequentially" than it does "randomly" in relation to how the OS sees it.

If that weren't the case, then SSD benchmarks would be totally useless as you would never be able to replicate the same test twice unless you only did one test on each new drive.

 

[devman]

Maybe so.. but that doesn't change the fact that it reads faster "sequentially" than it does "randomly" in relation to how the OS sees it.

If that weren't the case, then SSD benchmarks would be totally useless as you would never be able to replicate the same test twice unless you only did one test on each new drive.

You are correct. However, fragmentation on an SSD would have to be severe as files are not usually broken up in to non-contiguous 4KiB extents and performance picks up more rapidly from the worst case on SSDs than on HDDs. One would likely only see that form of fragmentation on a modern file system if the file system were nearly full (which would make defragging difficult).

 

[Aesma]

That is crazy. Every file fragmented does not make sense since there will be files of 1 cluster which by definition can not be fragmented.

If it's a secondary drive it could contain only or mainly big files, I've seen 100% fragmentation plenty of times.

Maybe so.. but that doesn't change the fact that it reads faster "sequentially" than it does "randomly" in relation to how the OS sees it.

If that weren't the case, then SSD benchmarks would be totally useless as you would never be able to replicate the same test twice unless you only did one test on each new drive.

Actually it took years to many websites to come up with meaningful SSD tests, at first they were indeed all over the place, not showing the effects of multiple writes.

The only way to defragment a SSD is to make a full disk image, then reset the SSD (secure erase), then restore the image. I'm talking about a logical image not a bit by bit one, else it would be fragmented too.