Rapid Mode still runs without Magician running Steve.
Also you're right, all SSDs have some hidden space for over provisioning. The 240Gb EVO is an example I think. It's really 256GB with 16GB hidden for OP. <<That may be wrong too.
See my avatar, I have a lot of wrong thoughts.
I think you are right on OP. As far as I know, all SSDs have built in OP, to even out wear leveling and is needed. But, I had thought rapid mode required Magician to be running, glad you straightened me out. See, you're not alone in having wrong thoughts.
This might be a stupid question:
If you enable OP, can you change your mind later and reclaim that space?
I'm thinking (wrong thoughts?) that if the OP space is there it will start 'filling up' with redirected nand cells that would be permanently allocated.
I've seen several misconceptions here.
Additional overprovisioning is not needed for TRIM to work; both the 840 Pros and the 840 EVOs are already overprovisioned anyway. Any space added to the factory overprovisioning would be better used as overhead. What is needed to enable TRIM to do its thing is some overhead, same as HDDs need it for defragmentation. While 10% over head (unused space) is the figure I see bandied about most frequently for HDDs, SDDs perform best with 20-25% overhead. Any less could cause performance reduction and increased writes which may cause reduced life. Overprovisioning is mostly just to increase life by holding cells (the SSD equivalent of HDD sectors, kinda sorta) in reserve that can take over the function of cells that have reached their maximum write capacity. There are a couple of tests that shown SSDs will last way past their projected write life with just the factory overprovisioning; the whole limited write cycles issue has been blown way out of proportion.
Magician does not have to be running for RAPID to work once it has been enabled.
Benchmarks are not an accurate representation of real world performance. Benchmarks can look as impressive as all get out, yet, in actual use, those gains may not be particularly noticeable, especially since boot times will be unaffected (RAPID doesn't kick in until after boot). The benchmarks on my EVO look fantastic but the actual perceivable performance gain was much less. Doubling the speed of an operation that normally takes only a second is going to be far less noticeable than doubling the speed of an operation that normally tales a minute.
Excellent explanation Jeannie. Thanks
Another thought burp:
I assume that when an SSD reaches a certain point, in terms of bad cells, that is will fail the SMART test and the user would be notified. This (hopefully) would give the user enough time to replace the drive. (I know this is the best-case scenario and it is equally likely the drive would just go *blink*.)
If you are over-over provisioning, does this just mask the condition and "extend the drive's useful life" by allocating a greater number of cells for replacement? Meaning: you would not know the drive is near end-of-life as soon?
Once again TVeblen thanks for this thread.
I have learned a lot. Now just to remember where all this information is when I need it.
Thanks to all for their input and expertise.
Possibly, I don't really know. In actual use, unless you abuse the SSD with excessive writes, such as defragging every day, it is extremely unlikely you will reach that situation. An SSD is far more likely to die from controller failure than from excessive writes.
Initial concerns for early SSD failure due to high writes have been proven to unfounded by actual tests. I was leery of the Sammy 840 EVOs when they first came out but user reviews have been just as good for the EVO as they were for the 840 Pros and professional endurance testing has shown the rumors of early demise due to writes have been highly exaggerated (apologies to Samuel Clemens).
Almost all drives will eventually fail. As far as depending on SMART to warn of impending drive death goes, its not uncommon for any kind of drive to fail unexpectedly without any SMART warning so, rather than worry about it, just prepare for it by maintaining frequent, regular backups. Also, it is possible for backups to fail so keeping multiple backups, both onsite and offsite, is essential. That way, if a drive fails unexpectedly, your backside...er...data is protected. And yes, maintaining multiple backups (or even just one) costs money but how much is one's data is worth? Enough to spend several thousands of dollars for data recovery that may not even work if (more like when) a drive dies?
Last edited by Lady Fitzgerald; 18 Mar 2014 at 13:35.
I don't think I meant extra OP, just that all SSDs must have some. It's the difference in the logical capacity minus what the OS reports as usable space.
Nevermind, this article on Wiki will explain (and confuse) it all. It also brings up Wear leveling and Write Amplification which hasn't been mentioned.
I have it started at GC.
Write amplification - Wikipedia, the free encyclopedia
Jack (The Bear) the Wiki would be a good thing to bookmark or copy.
Quote