No answer ...
Was the question not clear? Shall I rephrase?
RAID includes all the functionalities of AHCI "just like you said".
Recently TRIM has been supported under RAID levels. Native Command Queuing improves the performance of certain few SSDs. However the Samsung 840 isn't affected with or without NCQ.
Typical users would not need it, although it does not hurt to have it there.
Yes, your driver version supports TRIM under RAID. By recently I meant that at first TRIM wasn't functional under RAID.
The Samsung 840 pro is among the best & fastest available SSDs on the market today. You made the proper choice by buying the pro version.
Just take it easy on the benchmarks . You don't want to make those unnecessary writes which might affect performance on the long run.
Yes, but what about "Random Write IOPS"?
3 out of 4 performance are where they should be, but "Random Write IOPS" is off?
Do you have any explanation for this?
1 of 1 people found this helpful
Your IOPS is not off "actually it's good"
Input Output Operations Per Second "IOPS" is a bit complicated thing to explain. It depends on the drive's storage capacity, the benchmark itself that is being ran, etc... Not to mention that published numbers doesn't guarantee real world scenarios. Do note that HDDs top it out at 100 IOPS.
Also, it says up to 90K & yours was at 62K. In real world apps you won't realize any difference. Albeit, I know that's not what you want to hear, you want to know why your drive isn't topping the 90K.
Like I said this number varies according to many things but, still 60K is more than you & I will ever need. Moreover, most SSD manufacturers focus on read speeds rather than write speeds to replicate real world scenarios.
In a nutshell, your drive utilize something called toggle NAND, which is a mixture of both. Yet they focus on read speeds cause that's what we do most of the time. Hence, the read speeds are higher in IOPS
OK ... thanks a lot for the explanations!
1. TRIM and NCQ is supported with my IRST 184.108.40.2066.
2. My SSD IOPS benchmark is fine.