Bit the bullet and bought 2 SanDisk Extreme PROs

Everyone at OC Forum advised me that the 850 PRO 1TB is a better performer, but I always go by what tilleroftheearth tells me so since he recommended the SanDisk Extreme PRO along with Mr. TomJGX, I went ahead and bought 2 SanDisk Extreme PRO 960GB SSDs from Amaazon for $430 USD each. This drilled a big whole in my pocket I hope I can sell those 850 PRO 1TBs quickly....

The reasons I did this were:

1- As recommended by the SSD Doctor tilleroftheearth

2- The SanDisk Dashboard can see the SSDs and update their firmware even when they're in RAID mode, crappy Magician can't.

3- Lost faith in Samsung after keeping us EVO users waiting for more than a year and still no fix available for the TLC crap they sell. I know it's coming next month but as we stand, there isn't any firmware to download.

4- They released a firmware for the 850 PRO that crippled the performance and bricked many 850s......Mine wasn't bricked since they were already on the updated firmware when I bought them but that makes me not trust them even more. What's worse, is they didn't release a new firmware to remedy the performance issues that the EXM02B6Q suffers from.

5- Samsung's customer service sucks!! big time! they don't honor their international warranty if you're not in the US so that's not so international of a warranty..... their customer service also takes a long time to respond. When I contacted SanDisk last night, they confirmed that the warranty is indeed valid worldwide and they replied to my other questions within a few hours of me sending the email.

6- Consistency in performance, I heard that the SanDisk Extreme PROs are the consistency in performance kings......

Can't wait till my drives arrive


I hope SanDisk releases an MLC based mSATA SSD so I can get rid of anything that says Samsung in my computers and never buying anything from this once great company that has went downhill.

 

Can someone explain to me how in normal Benchmarks, the 840 gets higher benchmarks usually than the SanDisk SSDs so obviously that means the 850 PRO would be even a bit higher, but when it comes to performance consistency, the SanDisk even the Ultra II which is slower than the Extreme PRO in the below review blows the 840 PRO out of the water?????

http://www.tweaktown.com/articles/6...tested-for-real-world-performance/index3.html

 

840 EVO fix being released this week. Congrats on your purchase, I doubt you'll notice any difference between 850 Pro and Sandisk Extreme Pro (especially if you don't push your SSD's hard), but the firmware update in RAID for the Sandisk is helpful.

 

Thanks man, as silly as it may sound, believe me, when I have my Samsung SSD in RAID, I always felt like I was living in a locked dark cave, not knowing what firmware updates if any are available, not knowing my SSD health, written TBs, etc.... this was one of the main reasons I bought the SanDisk Extreme PROs and their performance consistency that tilleroftheearth always talks about. I have lost all my admiration and respect towards Samsung after their recent hiccups.

 

hey hey hey.. Don't blame me for the whole in your pocket! You bought it ..

Congrats and hope you enjoy it!

 

I have a request in for RMA on one of my SanDisk Extreme Pro 480GB drives Hopefully I can get this taken care of quickly, I've never RMA'd to SanDisk before.

 

I ignited the fire, that's right, but you got the matchsticks / gasoline

 

At the time the 840 Pro was released there weren't many, if any, review sites testing for performance consistency. Firmware. Samsung chose to go for synthetic benchmark results, SanDisk chose performance. With review sites using synthetic benchmarks, Samsung was king. Consumers ate it up and still do. Samsung took it to the bank. Along came [H]ardOCP and AnandTech with their steady state, performance consistency, Destroyer tests showing that true high performance was SanDisk. Later the PCMark 8 you link to on TweakTown confirmed the 840 Pro was not a high performance drive. Samsung are not dummies. They saw/knew the 840 Pro really wasn't pro at all and changed for the 850 Pro. The consumer benefits.

 

Great explanation. Now let me ask you this......

I am almost gonna receive these pair of SSDs........ I am contemplating whether to RAID 0 them or not......

I keep reading mixed reviews and posts about this.... in one review says RAID 0 is only good for benchmarks but for the average and even power users, we will never reach a que depth that would make any benefit of the super fast sequential reads/writes and a Single SSD has lower access time / latency and that's what counts most......

What do you think? keep the first SSD for the OS and the second for the games + my software collection where I have all the setup files of my programs or RAID them?

Also do note, on my current setup which is two 850 PRO 1TBs.......if I install them in AHCI / single drive......let's say I want to make an image of my C: partition as a backup using Macrium Reflect......it takes 1 minute exactly....

but in RAID 0, it takes 1 minute 30 seconds or even 1 minute 40 seconds.......so where is this sequential speed benefit in the real world? I don't understand the logic.....

BTW, what I do on my lappy is:

Surf the net / forums 80% of the time
Watch movies 10% of the time
Play games 10% of the time

 

RAIDed SSDs provide almost 0 real world benefit except for benchmarks and the ability to show it as one drive.

In some cases, like you've already noticed, a single SSD is faster.

And since I'm guessing you have a laptop keep in mind its not hardware raid, its most likely firmware raid which is considered by many to be the worst type.

 

I know you love overkill

Here's a link to the SE Pro 240GB RAID0 steady state heavy usage. No 1TB RAID0 results are available. Comparing single drives the 1TB is faster (233 vs 209 steady state). It's a six drive review but two drive and single drive results are there too. You should also look at Page 4 [Benchmarks (Trace Based OS Volume) - PCMark Vantage, PCMark 7 & PCMark 8] It covers lighter usage. Which seems to be your workload most of the time. These aren't the graphs you hate to look at.
http://www.tweaktown.com/articles/6...o-240gb-six-drive-ssd-raid-report/index6.html
And from the Final Thoughts:

If I had two of these bad boys I would, definitely, try RAID0 and decide for myself.

 

oh snap!! that's exactly the question I was asking here so that's very bad as it uses more CPU to do stuff right?

 

That's contracting what you just said above bro.....you said that a single large SSD is better and now you're saying you would RAID 0 them?

PS: Does the SanDisk Dashboard do the Overprovisioning like the Samsung Magician or shall I do it while partitioning getting ready to install Windows?

 

To be clear a single 960GB beats a single 240GB in steady state. The link in the previous post has 240GB results. This link has 960GB results. Single drive. Compare the two links.
http://www.tweaktown.com/reviews/64...sd-review-sata-iii-to-the-extreme/index7.html
You can also compare a single 960GB vs RAID0 240GB. RAID0 wins in this example.

 

I see, thanks for being patient with me......BUT.....what about the fact that I am doing this on a laptop and is probably not hardware dedicated RAID and would probably steal more CPU cycles? As given in the example I mentiond before, RDx2 or RDx3 results in 1 min 33 seconds to 44 seconds while creating the Macrium Reflect Image of my barebones Windows installation while if I do it on a single SSD backing up the C: drive to the 2nd drive (D) it would take 57 seconds to 1 min max.......

what do you think? This is no benchmark, this is a real world usage scenario where RAID actually hurt the performance rather than help here

 

You have to always match the performance of the tech appropriately to the workload. If this is your heaviest workload, you have your answer: don't use RAID. Myself, I don't need further confirmation from anyone else.

I may try different parameters to test further with, but adding up to 75% more time to complete a task is nonsense just to claim fast RAID 'scores' with.

On a notebook with no hardware RAID controller the best performance is always to use multiple disks separately and appropriately for the main task they'll handle. (Assuming the notebook will not be used for RAW video editing).

Splitting the tasks logically across two, three or more drives will give the best result when increased productivity is the goal.

Two Drive Setup:
One drive for O/S + Programs. Second drive for (raw) Data, WIP (Work In Progress), Scratch/Temp files and Finished work.

Three Drive Setup:
One drive for O/S + Programs. Second drive for (raw) Data and WIP. Third drive for Scratch file (temp files) and Finished work.

Four Drive Setup:
One drive for O/S + Programs. Second drive for (raw) Data. Third drive for WIP and Scratch file (temp files). Fourth drive for storing Finished work.


Choosing the best drive for each (main) task above depending on the drive configurations in the above scenarios is the 'trick'. The easy way is to simply buy the biggest and most consistent drives and use them for all situations, and while you may spend more $$$ than you would initially want to, it also gives the best long term bang for the buck too.

 

Well the whole reason I bought these two SanDisk EXTREME PRO 960GBs is due to your recommendation and TOMJGX's, despite me having two 850 PRO 1TB SSDs I did it. Then when I saw the consistency tests online and saw how the SanDisk Extreme PRO even though gets much lower benchmarks than the 850 PRO, steady consistent performance wise, it blows it out of the water and that shocked me big time.

My current setup is and will remain as the following:

1st drive: C Partition, used to install Windows only and the used space would be 65 GB after having installed all my programs
2nd drive: D Partition, Software folder where I have all the software setup files, and where I install the games, used space: 204GB
3rd drive: 840 EVO mSATA 1TB, I place my OneDrive folder there + Videos, used space = 532GB, it is currently OPed by 10%
===========================================================
Does the SanDisk Dashboard do OP?

Does it have a function to easily set OP percentage and apply it like Samsung Magician or must I do the OP while partitioning to install Windows?

Additionally, is it true that SanDisk SSDs usually have 12.7% factory OP? So how much OP shall I apply?

space is not an issue but still I don't want to OP beyond what gives me optimal performance and durability if you know what I mean as I heard that free space DOES NOT equate to Overprovisioning (Don't know the logic behind that)

 

As I said, if I had two 960GB SE Pros I would try them in my workload in RAID0 and see my results.

@tilleroftheearth @RCB are great with the math in MBs for formatting sizes with Windows installs. It might be very useful if one of them had a thread, (maybe it already exists?) that listed drives by size 240GB formatted size, C: size, D: size, then delete C: to have unallocated space between C: and D: for expansion/contraction and end up with 30% OP. Then the same for 250GB, 256GB, 480GB, 500GB, 512GB, 960GB and 1TB.

All SSDs have OP. Level 1 in this link http://en.wikipedia.org/wiki/Write_amplification#Over-provisioning
1024GB physical capacity - 7% = 960GB logical capacity. 7% factory OP (Level 2 in above link) for the Extreme Pro family. The 12.7% we've seen comes from the conversion - Level 1 and the factory OP - Level 2. Confused yet? I am
http://www.sandisk.com/assets/docs/WP004_OverProvisioning_WhyHow_FINAL.pdf

960GB formatted to 893GB - 30% = 625GB

Not all SanDisk SSDs have factory OP.

 

You'll never be using the SSD's at steady state though, because you're a light user when it comes to the demands on the SSD. You want to be more interested in light workload performance - to be honest you'd probably have been better off with the intiail x2 850Pros, and then use them as seperate drives (not RAID). Even more shockingly I think the 850 Evos are better than the 850 Pros at light workloads, and they're really cheap too - think I remember concluding this from some reviews I saw. Get some 850 Evos! ;-)

 

Of course, he'll be using his drives in steady state. I bolded it this time. He may not fall under heavy usage most of the time.

 

Never buy anything from Samsung again! A company that has lost all meaning of quality, does not respect its international warranty because when I contacted asking why the Samsung Dubai say that if my SSD ever fails they will not help me? Samsung USA replied back saying yes the warranty is international as in you can use the SSD anywhere but you have to send it to the USA for any repair or warranty claim! people trust that an international warranty is international, not the dodgy way Samsung has it whilst SanDisk when I contacted them said they will honor my Warranty anywhere in the world no matter where I purchased the SSD from.....

let's not forget about the 850 PRO 2nd firmware update that crippled the performance by about 15% and it has been more than 1 month now and still no fix.......

I know this performance issue because I had a 256GB 850 PRO on the original firmware and it performed great but the two 1TB 850 PROs I later bought already came with the newer firmware so no way to downgrade and the performance isn't as great, this has been discussed in great detail at the OC forums / SSD section

Finally, I saw how the Macrium Reflect backup operation took 75% more time to complete in RAID 0 mode vs a single SSD so I'm not very enthusiatic about RAID to be honest especially after you guys mentioned that Software RAID sucks since Im on a laptop

 

I think that a light user hardly ever uses his SSD in steady state. Steady state occurs after the drive has been written to for a 'long sustained period', which doesn't happen very often for light users. I think I'm right about that. EDIT: although I might be confused about the definition of "steady state" after googling it just now. I always thought "steady state" was a temporary condition after a sustained period of heavy constent writes, after which point the drive can recover with rest & TRIM. My point was that light users don't push them enough to reach steady state.

Haha, yes, I know, I mainly said you should get Evos for a joke, hence the wink smiley at the end of the sentence! Although I do think they are good drives for light users, and are fast drives for light users too, so I stand by that point, but I was joking when I suggested you buy them now that you have x2 850Pros & x2 Sandisk Extreme Pros already!

 

The confusion might be that for review sites to test a new drive in steady state they have to "artificially" put it there. Hopefully, SNIA guidelines are followed - consistent, repeatable - fair. The long sustained writes put the drive into steady state for testing purposes. A light user's drive will take longer to reach steady state "naturally" compared to a heavy user. But every SSD will eventually reach steady state. Steady state = used state. Using a car analogy, let's say 20k miles would be this "used state" designation. It took me 2.5 years to get there but I did. While another driver got to 2ok in 1 year. He got there too, just quicker than me.

SSS (Solid State Storage)

http://snia.org/sites/default/files/SSS PTS Client - v1.1.pdf

http://www.hardocp.com/article/2013/06/05/sandisk_240_gb_extreme_ii_ssd_review/8#.VTddYWeTB68

 

Ah, I see, although I'm not sure how I'd determine whether my drive was in steady state or not? Steady state only influences the write speed right, so maybe that's another argument for light users getting a drive with an SLC cache, because writes to the cache are always gonna stay out of the box fast for light workloads - e.g. the Evos. Most of the writes I do will fit within that SLC cache, so would always feel fresh out of the box fast I reckon, regardless of the fact that the rest of the drive is in a lower performace steady state.

 

Consider using sysprep and moving the Users and ProgramData folders to the secondary drive. Then the active drives are being used more efficiently. It isn't RAID but they are both being accessed simultaneously.

 

Just read on how to do that........too much of a hassle to be honest....

http://www.eightforums.com/tutorials/4275-user-profiles-relocate-another-partition-disk.html

 

This may be the official definition, but I still feel like it does not match the "steady state" shown in reviews. If it is the "final state" of the SSD, why do the benchmarks, such as the PCMark 8 Storage Consistency Test, show performance drop as it enters steady state after sustained writes, but then recover and return to original levels after the drive has been given a chance to recover, as seen here? As soon as the drives are given a few idle minutes, they can recover to their original performance. It would appear that the "steady state" described in reviews does not match this definition.

 

Yes, this is the view that I had too.

 

The total time required is between 5 and 10 minutes extra, but if you're constantly reimaging and creating bad images, swapping out for newer drives and so on, I guess it is. Personally I haven't reimaged my computer in over a year and not planning on needing one for another year or two and at which time when that occurs I will switch the drives positions and balance the wear.

If people want maximum performance, efficiency and safety they should at least try all of the options available.

 

I just formatted and installed Windows 8.1 (I never clone my disks, clean fresh start FTW), configured all my user folders placing them my E: partition (the mSATA 840 EVO) as usual like pics, docs, music, videos then I did all the Windows updates

Then I did the disk cleanup to clean the Windows updates and temp files and took my first image using Marcum Reflect Home

Results:

Single 850 PRO finishes this job on 57 Seconds

Dual 850 PRO in RAID 0 finishes this job in 1 min 33 Seconds

Single SanDisk Extreme Pro finished the job in 42 Seconds

This is a very good start and mind you no drivers or anything g are installed yet and it achieved 461 MB/S Read and 487 MB/S Write in the Macrium Reflect progress bar where the single 850 Pro maxed out at 380 MB/S in that progress bar

I'm starting to like this drive and starting to more and more believe that synthetic benchmarks are meaningless compared to real life usage scenarios



Will post more benchmarks soon

 

Sysprep saved my rear when I accidentally erased the only image I had of my sig Clevo laptop. Sysprep'd my desktop, moved the drive into my laptop, reinstalled all the drivers and I was good to go again. (Although I will note that boot time increased by a few seconds...)

 

So here are the benchmarks, a bit slower (in benchmarks) than my 850 PRO 1TB but faster scores than all the reviews I've seen online probably due to my faster laptop....

Samsung 850 PRO 1TB:

AS SSD Benchmark with IRST 13.6.0.1002 (W8)



CrystalDiskMark with IRST 13.6.0.1002 (W8)




SanDisk Extreme PRO 960GB:

AS SSD Benchmark with IRST 13.6.0.1002 (W8)




CrystalDiskMark with IRST 13.6.0.1002 (W8)



But what matters at the end of the day, is performance consistency and real world performance as I have experienced myself right from the start even before installing the IRST drivers the backup/imagine process was about 20% faster on the SanDisk Extreme PRO.

 

Sorry, no time to reply to individual posts:

1) I do not consider any OP'ing the manufacturer does as OP'ing for my uses. The reason is that they OP to get (just) past their warranty period and/or to be able to claim the synthetic 'scores' they advertise. I OP because I want to use the SSD my way, at any time, without having to wait for it to finish it's background cleanup routines.

I have chosen 30% OP'ing as the best compromise between fast/more consistent performance and less user capacity for almost all my notebooks and this works well for my heavily used mobile setups (which I can only 'push' so much). On my desktop workstations, I am more inclined to go to 50% OP'ing and even 70% as on those systems, the SSD in question is simply a Scratch disk and is only used exclusively for temp files. Here, I want all the performance I can get, consistently, and sacrificing a few GB's of capacity is the easy way to do it.

How do I calculate 30% OP'ing for any specific system? First, I don't install or use any utility to do it. At the time of a clean Windows install I simply choose Advanced setup, choose the SSD I want to install to (making sure it is the only storage subsystem device connected during the install except for the USB Windows 8.1 'key') and choose Format. That number (in MB's) that is displayed is what the SSD has available to the user... I multiply that by 0.70 and I now have the capacity in MB's with a 30% OP. Of course, I do it at a deeper level than this quick example I show here now, but the extended version will have to be at a later time.

2) Steady State is reached by all SSD's in a matter of minutes. By the end of installing Windows (around 5 minutes from a USB stick) and it's updates, most SSD's performance has tanked vs. the initial 'wow' factor. If I remember correctly, a 4K R/W workload will put an SSD into SS in as little as 15 minutes. While a sequential file transfer will need about 3 hours to do the same thing.

Steady State does not matter how much or how little an SSD is used. If it is used at all, it will reach steady state as already mentioned by others.

3) When doing a quick comparison of drive capacity vs. user capacity after OP'ing, this may help:

With a 240GB 'nominal' capacity drive equaling 'Y' in the example below:

Y x 1,000,000,000 = 240,000,000,000 usable capacity in bytes.

Divide above result with 1024 / 1024 / 1024 = ~223.517 MB's actual (binary).

Multiply above result by 0.70 = ~156.46GB after OP'ing by 30%


Here are some other common capacities after 30% OP'ing:

1TB nominal = ~651.92GB
960GB nominal = ~625.84GB
500GB nominal = ~325.96GB
480GB nominal = 312.92GB


Considering that in the near future my new Windows installs will allow 200GB for my C:\Drive partition (which still allows for ~113GB of data on the D:\Drive partition on the smallest drive I'll consider using today), using anything smaller than a 480GB SSD going forward is just a form of sadomasochism if you want this install to last beyond just getting past the updates this year.

I'll post more on this when I have more time, but for now, here is a link of what I've already written on that (see my replies in that entire thread).

See:
http://forum.notebookreview.com/threads/raid0-setup-m-2-vs-hd-ssd.772705/

 

Oh, speaking about OP, I OPed my first drive (OS) which only has 50 GB of used space since I disabled hibernation file/fast boot by 50% I don't need more than 100GB for the OS drive but since I have plenty of free space, I OPed by 50%

I was waiting for you to comment on the fact that this SanDisk Extreme PRO has lower benchmarks but in the real world, it gave me much faster speed when creating an image using Macrium Reflect

For the 2nd drive where I install my games and keep my software / drivers installation folders, I OPed that by 30% only to give myself more free space if ever needed.

For the 3rd mSATA 1TB 840 EVO, I kept that @ 10% OP since it only has video files, documents, music, and pictures. Performance is not a concern there so I treat it as if it were an HDD.

 

These are my new drive icons

 

Matrix Leader6 I see you used irst 13.06.1002 instead of 12.8.0.1016? Is there a particular reason you changed your mind with the irst was best to use with win 8.1?

 

Wow man you have a great memory and great attention to detail you even remember what drivers I used before :d

As you know my friend, I test every single driver out there and I always go back to 12.8.0.1016 something about that driver is better for my chipset/motherboard it always gives me better results.......but I tried the latest 13.6.0.1002 driver 2 days ago and it did give higher 4K speeds and a 0.09% lower latency

Here's a comparison of both drivers on Windows 8 but on the 850 PRO 1TB:

850 PRO 1TB:

AS SSD Benchmark with IRST 12.8.0.1016 (W8)



AS SSD Benchmark with IRST 13.6.0.1002 (W8)




CrystalDiskMark with IRST 12.8.0.1016 (W8)



CrystalDiskMark with IRST 13.6.0.1002 (W8)



If you do decide to try it, please follow the below method:

a) Uninstall current IRST Driver then reboot

b) Install new IRST Driver then reboot TWICE allowing 30 seconds after the first reboot for the driver to properly install

c) after the 3rd reboot, take a coffee break, I mean, wait for 4 minutes till the IRST application loads

d) Go to the performance tab and disable Link Power Management

e) In Device Manager, go to Drives, right click on your SSD and go to Policies then check the box that says "Turn off windows write-cash buffer flushing" then hit ok then reboot

In my SanDisk Dashboard, they actually advice you to do that to get better scores in benchmarking. I always leave it this way though as there is no harm since it's a laptop and it has a battery so a sudden power loss will not affect any data cached but not written to be lost.

f) After the final reboot, run a TRIM command then wait for a while till everything has been loaded, then do the benchmarks. Do not run AS SSD and CrysalDiskMark right after each other to allow the drive to revive its performance.

 

I do not think I'm forgetful + rep

 

I see your point. The wording can be much clearer for PCMark 8 Storage Consistency Test. It is a simulation (Trace Based) of steady state and must be run on an empty drive.
The results don't show the Pre-Conditioning Phase. It begins at Degrade 1. This is non-stop writes from pre-conditioning through degrade to steady state worst case results. Then recovery, TRIM and GC can happen. This is not recovery to a new state. The test method is in the link below. Bolds by me.

http://www.tweaktown.com/articles/6...-tested-for-real-world-performance/index.html

http://www.tweaktown.com/reviews/7079/crucial-mx200-500gb-ssd-review/index7.html

In the above link, TweakTown used another method for determining light usage.

 

Ah, in that case the discussions around Steady State are mostly moot, because all people's SSDs will be in steady state almost from the get go (as long as it's the OS drive, a data drive might take longer to get there I guess).

The Steady State Performance often talked about in reviews must be a different type of state though - as normally this term is used to describe the temporarily decreased write performance of a drive after a sustained period of heavy writes (performance is quickly recoverable with idle time). This point just links in with discussions earlier in the thread debating the importance of buying a drive based on Steady State Performance; using the definition described in this paragraph, then it looks to me that light users do not need to be to concerned with steady state performance as their workloads would not push the drives into this kind of Steady State.

Looks like we need to use the term "Steady State" with care, as it seems to describe 2 separate conditions.


(EDIT: Bullrun, you addressed some of this in your latest post above (some of which I understand), I'd written my post before I read yours. Still there's 2 seperate usages of the term "Steady State" I think.)

 

Hopefully you'll be able to figure out what is slowing the boot time. Maybe there's a clue in event viewer.

 

I think it's the terminology used in the PCM8's simulation that leads to the confusion. Or seemingly two steady states. It doesn't use SNIA guidelines from what I could gather. It's trying to simulate steady state as the drive being hammered with writes. This is a consistency test. Pushed hard for hours on end.

The other test, light usage, OS Volume Steady State uses SNIA which can be done on a drive with an OS and 75% full in this case. Two in the same review.

 

Yes, still 2 different types of steady state.

 

It's a very confusing term to be honest........Steady State makes me think an SSD is in it's optimal condition to be tested.....but the reality is the opposite as what you are saying.....it's the worst case after a heavy workload / benchmarks

 

performance consistency



http://www.guru3d.com/articles-pages/samsung-850-pro-ssd-review,9.html













Also see this:

http://anandtech.com/bench/product/1242?vs=1252

 

This is the first benchmark I see where the SanDisk Extreme PRO destroys the 850 PRO! Surprising:

http://anandtech.com/bench/product/1242?vs=1252

 

Out of all those drives in the graph, I'd choose the 850 Pro because performance is higher than the other drives when it has recovered - my light usage wouldn't push the SSD into the degraded steady state ever.

EDIT: scratch what I said just there - the other benchmark variables aren't that favourable to the 850 Pro. Hell, I'd just buy the 850 Evo because it's cheap and fast for light workloads (probably just as fast the 850 Pro in those light usage cases).

 

wanna buy my two 1TB 850 PROs then? First one has 3 TB written to it and the second one has 6TB written to it so they're basically almost new

 

I think you're the only one who is confusing what steady state is. It is the state that is officially defined by SNIA's guidelines. And the state which an SSD falls to no matter how it is used. This state is much lower than FOB performance and even when left to 'recover' doesn't do that much to reach those highs ever again.

Btw, steady state is what synthetic benchmarks don't test at. Doing a SE before running another (synthetic) test is like having the NASA team soup up your Yaris first before pitting it against a Ferrari and then claiming that all performance has stopped in the gasoline powered arena (when they test their Yaris against a Ferrari).

Steady state is not what happens for some SSD's. It is a fact for all that are in use.

It is not a temporarily decreased write performance of a drive after a sustained period of heavy writes, nor is performance slowly or quickly recoverable with idle time.

See:
http://www.tweaktown.com/reviews/6714/samsung-850-pro-256gb-ssd-review/index6.html


If you want proof of my statements above, look at the graphs in the link. Do you see any SSD hitting the SATA3 limits? Yeah; not if their life depended on it. Very similar to a peak performing HDD's for most actually.

This is the state of current SSD performance (on a SATA interface).

 

I understood already your previous description of Steady State I think, so I won't repeat what I wrote in my previous post to you. Looking at your link you posted above, then I take it that my SSD in my laptop is always in the Recovered Phase, as I hardly place any write demands to it. So based on one of your previous posts saying that everyone's OS drive is practically in a Steady State, then I can assume that my SSD is in both Steady State & Recovered Phase. Is that a reasonable assumption?