SSD Performance Problem

Recently I have bought an SSD. To be more specific, the Sandisk Ultra 2.

I did remember what others have told me in the thread (SSD Choice Dilemma) so if you were one of the people that replied to that thread, please do not think I ignored your advice.

Now really, this SSD shouldn't be that bad- yes it's TLC, but according to reviews it is (fairly) good and that's enough for me. But the problem is that my read and write performance for small files are terrible (picture below-do tell me if it's not visible). For example, my 1KB performance is 19.5 mb/s read and 15 mb/s write in ATTO benchmark.

After contacting support (and ignoring everything but one section of Crystaldiskmark), they suggested a clean install, and now my numbers are even worse- both in ATTO and Crystaldiskmark.. In addition, they are not consistent- the scores change a lot every time I try.

I am not an expert in SSDs, but the fact that my hard drive can get over half the scores in ATTO for small file transfers leaves me wondering why the improvement isn't bigger- as I didn't spend a lot of money to get a somewhat higher performance.

In addition, when I look at what reviews have gotten, I am concerned. For example, the speeds for 1 KB transfer in ATTO in reviews are:

AnandTech: ~80 mb/s read and ~80 mb/s for write
TweakTown: 90 mb/s read, and 84 mb/s write

And this pattern remains in ATTO until about the 32 Kb range (where the speed differences isn't as large).
As for Crystal Diskmark- whereas before the clean install only the 4k write performance was bad, now all my values are not great.

So my question is, do I have a faulty drive? Or is there some limiting factor on my computer (and the Sandisk Dashboard reports that I have SATA 3 ; 6 Gb/s connections, so that's not likely to be the issue).

Thanks for reading and have a good day.

 

More than likely; in reviews, the drive is not the O/S drive - it is just attached to a secondary SATA port and tests run on it (sometimes without even formatting the drive at all...). This changes 'scores' dramatically vs. ones tested in real world usage scenarios.

Secondly, the two different utilities you're using to get those 'scores' are probably not using the same type of test data (compressible vs. non-compressible). This would also be affecting your comparisons.

Thirdly, while it may not have any effect on these synthetic scores, OP'ing should have an obvious effect for you if you choose to OP by 30% or more (depending on how responsive you are to small changes in responsiveness...).

See:
http://forum.notebookreview.com/threads/why-over-provision.760922/#post-9766709


The above is one of many posts I have made on the benefits of OP'ing vs. simple free space on an SSD.

Google my username and forum.notebookreview.com for further reading.

 

I would add that if you are testing on a new Windows installation then there could well be housekeeping going on in the background which will adversely affect the SSD test results.

Leave the computer on overnight and see if the scores are better in the morning.

John

 

Just thought I should give an update:

I secured erased my SSD and put in a clean install of windows, and now the SSD feels much faster. I haven't tried any benchmarks, but I can tell programs load faster than before (except for a few, but I'll look into that further- but overall, noticeable improvement)

Thanks for the suggestions though- it was nice to get other opinions on this issue.

And as a side note, how do you increase the over-provision (OP)?

 

Just leave some space when partitioning. You can shrink an existing partition if necessary.

Unless your workflow involves copying tons of data and then deleting them, this is usually not much of a problem under Windows.