my Plextor m.2 is faster than my array of RAID0 Samsung EVO 250X2

[BirdDad]

I am trying to figure out why my array is not faster, boot up times are less with the Plextor and the array is pretty fast but I thought it would be faster.

 

[LurchFrinky]

It doesn't sound that impossible.
Exactly what are the speeds you are seeing?
How big is the Plextor?

The Plextor should be faster than one EVO, but the actual speeds are going to depend on the size (smaller ssds of the same family are slower), which RAID controller is being used, and how you are measuring speed.

 

[Blain]

* Are the EVOs 840's or 850's?
* Have the EVO's firmware been updated?
* Are the Plextor and EVO array being tested in the same or different machines?

 

[corkyg]

Not surprised. The M.2 has less to do to achieve the same goal.

 

[KentState]

Going from an HDD to SDD make a decent impact on boot times, but I doubt you are pushing a single SSD to it's limits. An array may have an improvement with certain read operation at the cost of double the latency.

 

[Ketchup]

The M.2 interface has more bandwith available than SATA. That's one of the main reasons for it's existence. One EVO on it's own would be very close to the top of the bandwidth available for a SATA port.

Blain mentioned a good point about the firmware. Definitely want to get past the "bug" from the original firmware if you haven't already.

 

[VirtualLarry]

RAID-0 SSDs are only improvements for certain workloads, of which I believe general desktop workloads are not. If you had to do a lot of sequential I/O, then they would be faster, but for random I/O, there's little improvement.

 

[BirdDad]

I have not updated the firmware.
Windows 8.1 spins one and a half times(the little circles at boot up) on the RAID. The EVO's are 250GBs each.
With the Plextor you never get to see a circle it loads before the circle animation starts.

 

[Railgun]

Also depends on the controller somewhat.

 

[BirdDad]

the controller is the integrated on die south bridge of the Intel x97 chipset

 

[Blain]

* Are the EVOs 840's or 850's?
* Are the Plextor and EVO array being tested in the same or different machines?

Please fill in the blanks

 

[BirdDad]

different machines.
EVO 840's both of them
The Plextor is in my Z99 chipset pc while the EVOs are in my x97 pc

 

[exar333]

This is one reason I am excited for Windows 10. I can upgrade from 7 and add native NVME boot support for these M.2 drives. I am hoping we see some real screamers later this year that can fully use the 4 PCIe lanes.

 

[BirdDad]

I updated the firmware of the 2 evo 840's lets hope that that did the trick, I won't know until I re install W8.1, I don't see why so many have negative reviews of it, it works just fine with Classic Shell.

 

[Fernando 1]

I am trying to figure out why my array is not faster, boot up times are less with the Plextor and the array is pretty fast but I thought it would be faster.

Where are the benchmarks and what means "faster"?
Furthermore you cannot compare the performance of SSDs, which are running on completely different systems.

By the way: In November 2014 I have done an intensive performance comparison test with my ASRock Z97 Extreme6 (has full support of M.2 Ultra) using the following SSD configurations:
1. Single 512 GB Samsung 840 Pro SSD running in AHCI mode,
2. 2x256 GB Samsung 840 Pro SSDs running as RAID0 and
3. Single 512 GB Samsung XP941 M.2 SSD using 4 PCI lanes and running in AHCI mode.

Result: The RAID0 array gave me by far the best benchmark scores (only exception: Sequ. 4 MB Write).
I have published the results here: http://www.win-raid.com/t602f34-SSD-Benchmark-Test-Intel-SATA-AHCI-RAID-vs-M-Ultra.html

 

[Blain]

I updated the firmware of the 2 evo 840's lets hope that that did the trick, I won't know until I re install W8.1, I don't see why so many have negative reviews of it, it works just fine with Classic Shell.

I've used StartIsBack as well as Start8.
I'm sticking with the $9.99 / 5PC StartIsBack.
I did pony up for Start8, but haven't used it long enough to do a side-by-side, pro/con comparison.
I have used Classic Shell, but not long enough to remember why I switched away from it.

 

[Dasa2]

once you get into high ssd speeds cpu\ram performance can start to play a large role in load times despite them being significantly faster than the ssd

 

[Puffnstuff]

Classic shell works great for me plus it's free. I use it on both my desktop and laptop without any issues.

 

[Fernando 1]

@ Blain and Puffnstuff:

Is this a discussion about the Win8.1 start menu?
What have tools like Classic Shell or StartIsBack to do with the AnandTech Sub-Forum "Hardware and Technology" and with the topic of this thread?

 

[Blain]

There are 3 posts by 3 different people mentioning "Classic Shell", and the first was by the originator of this thread.

 

[razel]

Chances are drivers for your RAID is adding to your boot time. You can try to log your boot to verify what's going on. By the way, forget about RAIDing SSDs for consumer uses. Also forget about sequential read/write benchmarks for consumer uses. You will rarely ever use/process data that fast that you'll notice it yourself. If you only need 100MB of data and your RAID has double the sequential of regular SATA it doesn't mean that you'll get that data twice as fast. Think of sequential speeds are maximum speeds. It's all about latency and Raided SSD for consumer use tends to have more latency.

By the way are you on SATA m2 or NVMe m2? By your posts it sounds like you are on SATA m2 or PCIE (ACHI compatible) m2. I know... it's a mess. Straight up NVMe is still elusive... expensive and possible but elusive.

Just saw this released today. Total throughput looks impressive, nearly double, but the service times aren't. Largest difference in seconds is the time it takes you to pick your nose.