DiskFresh?

rezerekted

rezerekted

2[H]4U
Joined
Apr 6, 2015
Messages
3,054
Is this a good tool to use or more "snakeoil"?

DiskFresh - Refresh Hard Disk Signal

"DiskFresh is a simple yet powerful tool that can refresh your hard disk signal without changing its data by reading and writing each sector and hence making your disk more reliable for storage. It also informs you if there are any damaged/bad sectors so you know the right time to replace your disk. The best part is, unlike other tools it does all this when Windows is running and it does not interfere with the speed of your work at all."
 
Sound like it just reads and re-writes every sector. Hard drives shouldn't lose data in sectors but it can happen. I wouldn't say it's snake oil but I wouldn't say it's something that everyone would need to do. Maybe use it with very old hard drives.
 
Scandisk is free and comes with windows. This reeks of snakeyness.
 
Again with this kind of product?

Now I have to ask the inevitable question: are you actually using these programs or are you shilling them across the board for some kind of kickback? It sure seems to me like you're hopping from one useless program to the next and asking others what they think of them which smacks of shilling behavior to me, constantly bringing up stuff that no enthusiast around here would ever dare use and linking to them almost like you're hoping it attracts business to those links/applications.

Honestly man, you really have to stop doing this crap and find something better to do. :)
 
It's free, so, once again, they are not trying to sell anyone snakeoil. If you self proclaimed computer "geniuses" just answered the technical aspects of my questions we could all get along much better. If you have no technical knowledge of the subject then you should just move along.
 
IF i wanted a "fresh disk" i would just do an image backup, then a full secure erase, restore image and then fully defrag it. But honestly i dont see this needing to be done on most disks. Not to mention windows by itself can fully erase your hard drive before doing a backup image restore.

Either way it puts a lot of extra wear and tear on your hard drive.
 
This is a legitimate program and does exactly what it is supposed to do. Mostly used on certain Samsung 840 non evos that slowed down after a few months. All data eventually decays whether ssd or Hdd this just refreshes data and has a read only mode for testing ssd data fade. I use it every few months on my old 840 ssds...
 
zalazin said:
This is a legitimate program and does exactly what it is supposed to do. Mostly used on certain Samsung 840 non evos that slowed down after a few months. All data eventually decays whether ssd or Hdd this just refreshes data and has a read only mode for testing ssd data fade. I use it every few months on my old 840 ssds...
Click to expand...
is the task similar to what magician does under optimizations?
 
primetime said:
is the task similar to what magician does under optimizations?
Click to expand...
Yes, up until very recently with the firmware update that enabled non-Evo 840 drives to use the optimization, Samsung tech support was actually pointing users to Diskfresh.
 
PliotronX said:
Yes, up until very recently with the firmware update that enabled non-Evo 840 drives to use the optimization, Samsung tech support was actually pointing users to Diskfresh.
Click to expand...

Actually, the problem was with non-Pro 840s. I used DiskFresh regularly because I had an 840 Evo that definitely suffered from this slowdown. The Pro models had different chips that didn't have the slowdown problem, and all 850s are fine as far as I know.

DiskFresh is solid software that does what it claims to do. The problem is that it's now answering a question that nobody is really asking anymore.
 
All ssd suffer from this just takes longer for mlc and slc. Tlc is just shorter lived. Even hdds do, just takes a lot longer.
 
I'm guessing it can prevent bit rot on spinners, but might cause wear to solid states and interfere with their own algorythms.
Though with spinners, the defragger (even the built in one) should take care of that.
I'm not sure if this is an issue though; I don't know if you've seen my photos of a Conner Peripherals 210MB HDD with Win95 on it. The data was accessible after like 15 years.
The data integrity/media check sounds nice, but if you need that, you might also want the ZFS file system.
So, all in all, I personally wouldn't bother. In cases where I do have write once-read seldom archive scenarios, I just keep checksums and backup my backups every once in a while.
However if there's no performance hit associated with this software, and if it can also do checksums of files, I may be interested in checking it out.
 
tuggernuts said:
Actually, the problem was with non-Pro 840s. I used DiskFresh regularly because I had an 840 Evo that definitely suffered from this slowdown. The Pro models had different chips that didn't have the slowdown problem, and all 850s are fine as far as I know.

DiskFresh is solid software that does what it claims to do. The problem is that it's now answering a question that nobody is really asking anymore.
Click to expand...
Right, but whereas soon after the degradation was discovered, the 840 Evo received firmware that enabled the optimization within Samsung Magician, this functionality did not make it to the 840 Non Evo until within the last month. In that time, Samsung was telling frustrated users to use a third party software to make up for their laziness in not making the same change as it did for the 840 Evo.
 
PliotronX said:
Right, but whereas soon after the degradation was discovered, the 840 Evo received firmware that enabled the optimization within Samsung Magician, this functionality did not make it to the 840 Non Evo until within the last month. In that time, Samsung was telling frustrated users to use a third party software to make up for their laziness in not making the same change as it did for the 840 Evo.
Click to expand...

...that sounds about right for Samsung. They drug their feet every step of the way when that mess was playing out, doing everything they could to shirk responsibility. I won't buy Samsung anymore if I have the option.
 
Ok - lets get rid of some misconceptions. I have been in the magnetic business since 360 Kilobyte floppy disks. I understand and can compute low level FM MFM and GCR disk encoding. RLL is too obscure and has many levels. On ALL magnetic hard drives, data will eventually decay to unreadable - the question how long is eventually? Depends on drive manufacturer, formula for the magnetic coating, density of ferrous particles, quality of read heads...basically lots of variables.

Up until diskfresh I usually used spinrite, which alas can't be used on GPT partitioned disks, and had to be run at boot time and have exclusive use of the disk. Diskfresh avoids both issues.

First - "Scandisk is free and comes with windows" - ScanDisk is for NTFS database integrity checking. Do all the files and directories point where they should and not overlap each other. Yes it can be used for reading all blocks of a disk to check for "bad" sectors - however for a boot drive it must be done at boot time - and you can't be using the drive while it is performing that task. Most people don't have multiple partitions just one big boot drive - so good luck with that one. Diskfresh can do the testing while you are using windows.

Second - Really?? "image backup, then a full secure erase, restore image and then fully defrag it." - wow, um ok - that would take several reboots or at least two. Fully defrag would have to defrag Master File Table and pagefile and only possible at boot time.

Third - "is the task similar to what magician does under optimizations" - good question. Answer: No - Disk magician performs a TRIM operation on all unused clusters of the disk. This informs the SSD internal software that those blocks are now free and available for the SSD to use for maintenance, garbage collection, etc.

Fourth- "The problem is that it's now answering a question that nobody is really asking anymore." - Alas all too true. The 'spinners' or hard drive manufacturers could include such features themselves, in fact I think some high end enterprise models might (SAS and the like) - but if they did, their drives would last even longer and....well...you know built in obsolescence. If they don't fail, nobody will buy new ones.

Fifth - well I'm going to take the fifth.

Sixth - "Though with spinners, the defragger (even the built in one) should take care of that." - A deframenter, even an optimized one, will only pick up and move blocks of files that are - a. fragmented, and b. not in the best location on the drive - once a block of a file is in an optimum location it never gets moved by the defragmenter. So recently written files get moved, but they don't require refreshing, as they are already new. Old system files and program files and pictures and word documents are optimally placed usually once and then never get moved.

Seventh - READ only operation on an SSD should never degrade performance or reduce expected lifetime. Write operations reduce expected lifetime. Diskfresh has a read only option.

Eighth - Even on a Samsung 850 Pro, it can improve performance - My personal experience I ran a READ benchmark of my 850 Pro before ever using Diskfresh. This tool, SSD-Z, can read 100 or so blocks from areas across the drive beginning at block 0 and ending at the end of the drive. It draws a graph of the read speed for each set of blocks. For my 850 Pro, the graph was fairly choppy, some data points were maximum speed and others were maybe 75% of maximum. Then I ran a READ ONLY refresh of my entire drive. Then I ran the benchmark again and the line was almost flat near the maximum mark.
 
Tiberian said:
Again with this kind of product?

Now I have to ask the inevitable question: are you actually using these programs or are you shilling them across the board for some kind of kickback? It sure seems to me like you're hopping from one useless program to the next and asking others what they think of them which smacks of shilling behavior to me, constantly bringing up stuff that no enthusiast around here would ever dare use and linking to them almost like you're hoping it attracts business to those links/applications.

Honestly man, you really have to stop doing this crap and find something better to do. :)
Click to expand...

I just used diskfresh on all my HDDs just last week. I've had this install of Win8.1 running well for years and plan to keep it that way for a few more.
 
FYI: Samsung's SSD Magician software does not support the bulk Samsung SSDs supplied to OEMs like Dell, HP, Lenovo, and the rest, for use in their systems. There are ample user reports to indicate these OEM drives (e.g. PM841, PM851, other TLC based) suffer from the same problem of voltage drift but it seems to take longer than in the affected "retail" Samsung branded 840/850 series drives. Also these users might not get the firmware fix that changes the algorithm used by the SSD controller to refresh/rewrite 'old' data in the background. Or if there is a firmware update available, may not use it because some firmware updates warn it will result in erasure of the drive (necessitating a full backup and restore).

The only real solution, then (even as implemented by Samsung ultimately), is to use some method to periodically refresh (rewrite) this older data. Whether it comes from the SSD's internal controller doing it in the background, or Windows application, doesn't matter. Disk Fresh seems to be the only such modern utility available that is specifically designed for this purpose, other than SpinRite's "refresh surfaces" feature but SpinRite is designed for magnetic recording disks, has not been updated since 2004 among other reasons not to use it (it costs $$ while Disk Fresh is free for home/personal use).
 
Benediction said:
Ok - lets get rid of some misconceptions. I have been in the magnetic business since 360 Kilobyte floppy disks. I understand and can compute low level FM MFM and GCR disk encoding. RLL is too obscure and has many levels...

Seventh - READ only operation on an SSD should never degrade performance or reduce expected lifetime. Write operations reduce expected lifetime. Diskfresh has a read only option.
Click to expand...

I appreciate what Benediction shared. I have a bit longer experience, having been in the business since well before floppy disks. I wrote the standard Internet description of RLL back in the mid 1980's :-D

In any case, here are some things NEW to me, I just learned this week. Much of this was demonstrated by an IBM Research fellow a couple of years ago. And while he's not alone, the real world implications have not been widely publicized. I'm still digging in so what follows is NOT technically complete as far as I'm concerned. But it is a start.

1) READ only operation on an SSD absolutely can and does degrade performance. In fact, a lot of reading on one set of cells can degrade the performance of physically adjacent cells!

2) While SSD's have longer overall life expectancy than HDD's (sure, Spinners ;) )... they tend to lose data more.

3) The expected lifetime of non-refreshed SSD data is much shorter than we think. Both because it's just short, and because it's NOT random. Some cells degrade much more quickly on any given SSD.

4) The degradation pattern is actually independent of SSD technology.

5) NONE of the self-test systems I have seen address this. Because they ALL are focused on write-failure, not static-data-failure.

6) SMART does not automagically fix this (although I suspect it could, if the appropriate people would work on it. Could have a single status bit per block that records whether the block has been rewritten since the last refresh pass.)

7) The most fragile static data is data that a) has sat there the longest, b) is read a lot. This is typically the most important part of any drive: the first few sectors, holding the drive boot info and partition map.

The result of all of the above, just this week for me...

- My boot SSD in my work computer (Surface Pro) suddenly failed. A few early sectors bad. While still functional (I hibernate between sessions), a variety of important files were damaged, most likely due to bad partition map etc data. By the time I realized I had a disk problem, my computer was in bad shape. (I do have backups... but not $$$ to replace this out-of-warranty computer right now. So this IS a painful thing for me ;) )
- After creating a full disk mirror image (ddrescue), I attempted to recover. Haven't had a chance to do that, live, with an SSD, until now. Oh joy?!
- Every test said this mSATA is in terrible shape, heading for disaster. I steeled myself to attempt opening the surface and replacing the mSATA. (NOT easy)
- On a hunch, I thought I'd see if I could recover the GPT and partition table, and maybe find a way to swap in a good sector for the bad, even at sector zero.
- Used a linux GPT repair tool. To my surprise, it worked easily and well!
- To my even greater surprise (at the time), rewriting the GPT automagically "fixed" the bad sectors by causing the drive to map them out.
- Suddenly, the mSATA reports 100% perfect on SMART. It's as if there were no issues at all. It has a few mapped sectors; that's it.

FOR ME, the result of the above is I want a free tool that does exactly what DiskFresh does, but for every platform:
- Rewrites either selected or all sectors
- Easily configured to do so every once in a while (DiskFresh does it 4 times a year)

So, we have DiskFresh. What is there for Mac and Linux?

(I'm sure it's easy to write a SH script for Linux. Make a cron job with a DD script...)

I'll reply to this with a link to the IBM paper when I get home... gotta post this before my battery runs out, waiting for my oil change :) )
 
Sorry for the delays; life and health intervened.
However, you benefit from me re-searching... here's some improved references. :)

1) 2013 research, from Intel CMU and more, on **MLC** chips. This foundational paper identifies four different failure modes, and three of the four involve no writing of the data that's degraded!
Error Analysis and Retention-Aware Error Management for NAND Flash Memory
https://users.ece.cmu.edu/~omutlu/pub/flash-error-analysis-and-management_itj13.pdf

Here are the four types:

* An erase error happens when an erase operation fails to reset the cells to the erased state. This is mainly due to manufacturing process variations or defects caused by trapped electrons in the tunnel oxide after stress due to repeated P/E cycles.
* A program interference error happens when the data stored in a page changes (unintentionally) while a neighboring page is being programmed due to parasitic capacitance-coupling.
* A retention error happens when the data stored in a cell changes over time. The main reason is that the charge programmed in the floating gate may dissipate gradually through the leakage current.
* A read error happens when the data stored in a cell changes as a neighboring cell on the same string is read over and over.

2) I'll jump to the very latest paper I have found. This is less than a year old. It incorporates understanding of MLC, TLC, 3D NAND, etc. It dives deep into the read failure modes, and discusses four potential on-chip mitigation strategies. One interesting surprise: the switch to TLC has (temporarily) INcreased cell sizes! Thus, while that doesn't solve these issues, it helps a little.
Error Characterization, Mitigation, and Recovery in Flash-Memory-Based Solid-State Drives
https://arxiv.org/pdf/1706.08642.pdf

My big question at this point:
- Which SSD manufacturers use effective on-chip methods for mitigation?
- Does anyone admit the real-world retention period for data on their drives, and/or the impact of temperature and P/E Cycles-to-date?

Here's a nice diagram from the second paper, showing how errors accumulate *between* write cycles:

Depiction of SSD Error Types.PNG
 
One more post. If anyone else discovers more info, I am VERY interested.

I just did a bunch of searching and reading to see if I could find ANY vendor-specific information.

1) Not really!
2) Micron has filed a number of patents that seem to relate
3) Samsung is very quiet... yet I get the sense they are probably on top of this. (My guess based on experience: they consider this trade secret. Thus, they would not publish.)

In fact, if anything the mfg's seem to avoid talking about these issues at all.

My bottom line guess: If someone were to do an intense test... you would probably discover...

* The 1st tier manufacturers are quietly addressing these issues. However, an important question is the storage lifetime of modern SSD's and in particular under stressful conditions, which are cold-active combined with hot-storage.

* Cheapie mfg's may not do much at all, so probably easier to create scenarios that make their SSD's fail or at least visibly degrade at the SMART level

* Others are likely in between.

* Practical reality: if you buy SSD's from top mfg's probably not an issue. Embedded SSD's inside tablets etc are likely more at risk.

But that's all just guesses. I have found no real data beyond what I've posted. And nothing vendor-specific at all.
 
You must log in or register to reply here.
Back
Top