SAMSUNG 950 PRO M.2 SSD

Has anybody messed with these yet?

http://www.newegg.com/Product/Product.aspx?Item=9SIA2W03JY5023&cm_re=m.2_ssd-_-20-147-467-_-Product


The geek devil on my right shoulder wants one because it's faster.

The pragmatic angle on my left shoulder wants to know just what faster is.


My current SSD does about 550 MB/sec read and 480 MB/sec write. This drive specs show 2500 MB/sec read, and 1500 MB/sec write.

That's a pretty respectable speed increase.


However, I messed around with a RAM drive (just for the heck of it) that was even WAY faster than this...however, it didn't really "feel" faster than my current SSD.


So, should I listen to the devil or the angle?

Maybe having a read here may help.

Yes - I already read that. Here is the official from the MSI forums:



I'm running a 5820K.

I have 2 GTX 970s in SLI:
- Card 1 is in PCI_E1
- Card 2 is in PCI_E5
--- I tried putting card 2 in PCI_E3, but the card in PCI_E1 had its fans running high as it was starved for air from the card in PCI_E3
--- In this configuration, both 970s run at x8 in PCI 3.0 mode.
--- I ran in the PCI_E1/PCI_E3 slots (i.e. the "Hot" config), that gave me x16 for the first card and x8 for the second card...but I couldn't see a measureable difference in speed (using Heaven benchmark) to justify the loud fan noise.

M.2 shares its lanes with the PCI_E6 slot...which is covered up by the graphics card in the PCI_E5 slot.

So - the M.2 slot is available for full x4 PCI 3.0 speed.


Has anybody tried this drive yet?

Ahh. I haven't tried an M2 drive yet so I cannot comment. I figured that read will enlighten you as to your decision. Not sure if you will notice the 2500 MB/sec read, and 1500 MB/sec write on that M2 drive since you had a Ram drive and didn't notice the difference in speed since it was faster than a typical SSD.

JrClocker, what are you using the system for?

Pragmatic Angle
- Gaming
- Everyday use
- Simulation

Geek Devil
- Dude...it's faster!
- Fast is good!


As I posted above, I did a ram-drive experiment and things didn't "feel" faster. However, I didn't have Windows installed on the ram-drive.

When I bought the motherboard, my eventual target was an M.2 SSD. Just not sure if it's time or if I should wait.

Depending on what your "Simulation" portion is, you could see improvements, but that is determined by what software you're using.

Gaming/everyday, it's not worth the money.

The simulation is mostly my own code.

I originally wrote in 32-bit code for my old pc. As my simulation space got larger, I quickly hit the 4-gig limit for 32-bit coding. I ported to 64-bit, but the simulation ran very much slower with 64-bit code versus 32-bit. (This was a few years ago...so I will have to redo this experiment).

To up the 32-bit code speed, I setup a disk buffer thread which spools simulation data to a disk file (some of the simulation results can get 10 GB or larger.) I get pretty good speed with an SSD. A few days ago I took my "old" non-LED memory and added to my (total 32 GB), and setup a 16 GB RAM drive. I set the simulation to use the RAM drive and the simulation ran much, much faster.

For the heck of it, I installed some apps onto the RAM drive to see if they ran faster. They did...but it was hard to tell...hence the "feel".

Hope that makes sense.

Maybe I'll try a 64-bit port of the simulation (will just have rewrite a portion of my graphics engine to make it work at 64-bit).

Then yes, you would benefit from the drive, but the only huge differences would be the simulation you run.

I figured it would impact my simulation...but I can get around that with a ram drive...or do the right thing and port the code over to 64-bit. (Which may be the next step as I finish with my post-processing engine).

I was hoping for some sort of decent review on the drive.

I don't mind spending the $$...it's the time to install it, port the stuff over, and have a lackluster result...and then adding yet another piece of PC hardware in my "that was interesting" pile...

You should see my fan pile...

I would port it to 64 bit first then and see how it goes.

JrClocker said:

You should see my fan pile...

Click to expand...

Buy Noctua, you won't add that to your pile.

Already did...NF-14 iPPC-2000 PWM...not all that impressed...still running experiments

JrClocker said:

Already did...NF-14 iPPC-2000 PWM...not all that impressed...still running experiments

Click to expand...

What didn't impress you?

ATMINSIDE said:

What didn't impress you?

Click to expand...

Well, I kept hearing all of this stuff about how the fans that come with the Corsair H110i GTX are junk, and that the Noctua fans are the best out there. When you stack up their published specs you get:

Fan................Mfg.......Size.....RPM....Airflow...Pressure...Noise
SPL140L............Corsair...140 mm...2400...104.6.....3.99.......40.0
NF-A14 iPPC-2000...Noctua....140 mm...2000...107.4.....4.18.......31.5


So, at first look you say that the airflow is the same, but the Noctua is not as loud as the Corsair.

Using an iPhone app, I measured the SPL for each fan at various RPMs with the phone at the same location on my desk. While the value in the app may not be 100% calibrated, all measurements where taken with the same app...and hence the delta between an "ambient baseline" and the measured sound is valid. When you plot fan noise vs rpm for both fans you get:



Umm...pretty much the same for both fans.

The Noctua is only quieter because it spins at lower RPM. I was cool with that because the spec sheet says it has the same performance as the Corsair at a lower RPM.

I am still taking some data, but what I can tell you is that the Noctua at 2000 RPM performs worse than the corsair at 2400 RPM.

How did I come by this? After completing various temperature profiling runs (I posted another thread about radiator fan intake versus exhaust...I'll update in a bit), I turn the fans on 100% to cool down the radiator to get back to a baseline. With the Corsair, the cooling was about 7 to 8 minutes. With the Noctua, the cooling is 13 to 15 minutes.

So, how can 2 fans with the same specs at max RPM perform so differently? Only 1 of 4 choices:

1. Corsair understated the specs
2. Noctua overstated the spec
3. I got awesome performing Corsair fans
4. I got dud performing Noctua fans

So...I am not impressed.

I'll leave the Noctua in my case for only 1 reason:

- While the noise for a given RPM is about the same SPL for both fans, the spectral content of the noise is different. The Noctua noise is "more pleasant"...in other words, it's not as much of a "wretching" sound as the Corsair (does that make sense?)

So what I'm gathering is that you got a fan with a better noise profile, is 9dBa quieter, and moves the same amount of air..... but that isn't enough?

Also, find a way to measure airflow. You'll be shocked at how much better the Noctua holds to spec than the Corsair.

No.

I have a fan that is quieter...yes.
Better noise profile...yes.
Moves the same amount of air...NO.

The noise is lower only because it spins at a lower RPM.

My data indicates that the Noctua moves less air than the Corsair...even though the specs say they should be equal.

JrClocker said:

No.

I have a fan that is quieter...yes.
Better noise profile...yes.
Moves the same amount of air...NO.

The noise is lower only because it spins at a lower RPM.

My data indicates that the Noctua moves less air than the Corsair...even though the specs say they should be equal.

Click to expand...

How does your data show that it moves less air?
Sound level =/= airflow

If I cap the Corsair fan speed to 2000 RPM (the same as the Noctua), it takes 13 to 15 minutes to cool off my radiator...just like the Noctua.

In other words, at the same RPM, both fans have close to the same SPL and move close to the same amount of air (by looking at cool down times.)

The Cosair can go higher in RPM for more airflow, at the cost of noise.

The Noctua fan has a more "pleasant" noise profile...and it's a more solidly designed fan.

I was trying to make an "apples to apples" comparison.

At the end of the day, I'll leave the Noctua fans on my radiator because the fan noise is more "pleasant".

Does that make more sense?

Wow, that sounds like the most inaccurate form of air flow testing I've ever heard about...
All it takes is a slight change in ambient, and/or case, temperature to skew this time.

ATMINSIDE said:

Wow, that sounds like the most inaccurate form of air flow testing I've ever heard about...
All it takes is a slight change in ambient, and/or case, temperature to skew this time.

Click to expand...

Um...no.

Re-read what I was measuring...the cool down time of the radiator.

It's not a precision airflow measurement...but a relative one.

You have an engineering degree...the thermal mass of the radiator plus coolant is large...add in the case and it gets larger.

All things being equal, a change in ambient will not skew the data hard...notice I said 13 to 15 minutes...versus 7 to 8 minutes? The "soak" time of each run is 30 minutes...so the case temperature is pretty constant.

I wasn't quoting seconds here - stop treating me like I'm stupid.