M.2 NVMe SSD for OS

[phusis]

Hi, folks --

 

My current and newly (partially) updated HTPC has Windows 10 Pro 64-bit 2004-build stored on a Samsung 840 EVO 128GB SSD, but I'm contemplating the upgrade(?) to an M.2 SSD (2280) NVMe standard for my OS (not music storage), a standard that the new Gigabyte Z390 D ATX motherboard in my HTPC supports. The rationale behind this possible change isn't necessarily speed per se (although overall performance will likely see a significant boost), but rather that it may have a positive effect on sound quality.

 

What are your thoughts and experiences here with M.2 NVMe SSD's for OS storage; do they wring out more sonic potential compared to a SATA SSD like the one I have, and if so how is this attained mostly effectively? Much has been written about RAM types and sizes and what suits audio reproduction the best, but is there a general guideline in regards to M.2 NVMe SSD's and what to go after for best sound quality? I'm circling the A-Data XPG SX8200 Pro series, 256 or 512GB, also because of their low power consumption compared to alternative brands, but Intel Optane variants seem popular around here - albeit at high prices - and with fairly low capacity to boot. Is higher capacity values and sound quality mutually exclusive, even? Please chime in.

 

/Mikael 

[phusis]

Oh, well - a bit of read-up here on those M.2 NVMe SSD's indicates low latency and low power consumption to certainly be among the vital factors in attaining a desired sonic outcome - whether it's mostly a subjective assessment or not, with some assessments in particular that gain traction. Capacity value, high or low, appears irrelevant. Intel's Optane variants have run to the lead in named categories, at the highest prices of course - go figure. 

 

My take will likely be for a relatively cheap M.2 NVMe variant spec-tested successfully in real-world conditions, namely Kingston's A2000 (500GB). It sports quite nice latency and power consumption numbers, and yet: how dreadfully incognito in regards to hifi-sensibilities, right? We'll see how that fares. 

[phusis]

@seeteeyou --

 

Thank you for the detailed info on the Intel Optane item. Although not outrageously priced (except perhaps in relation to capacity..) it's still fairly expensive. Latency numbers in particular are impressive, but if they're 'Random' measurements the Kingston A2000 isn't far off, nor is the stated power consumption by comparison. 

 

Still, what's the significance as it applies to perceived sound; is it a matter of specs in particular categories like latency and power usage, and/or more a matter of hardware details and the specific components used? It seems to me, and I may become unpopular claiming this, that there's a rather blind(?) adherence to the Intel Optane's (or the Apacer ECC RAM's for that matter, though consensus based on actual listening impressions seems more broadly applied here), simply because they've been favorably tested (compared to what?) by a select few individuals that for some reason have become the über-maestro's as sonic evaluators for others to follow. Or, is it truly a matter of broad consensus based on individual experiences, compared to a variety of others brands, that the Optane is the preferred choice of an M.2 NVMe SSD?

[phusis]

On 9/10/2020 at 12:49 PM, seeteeyou said:

Not so sure about that "a select few individuals" part, even the world famous SGM Extreme got Optane 900P installed

 

https://audiophilestyle.com/ca/reviews/reality-quest
-going-to-extremes-with-the-taiko-audio-sgm-extreme-part-2-of-5-r908/

 

(with all due respect to poster @ray-dude's awesome build) Exactly my point; is that necessarily telling of the Optane's sonic strength compared to cheaper alternatives, also with impressive latency, power consumption and read/write speeds? If the Optane is still in a slight(?) lead in regards to latency and power consumption specs, how much does it matter as it pertains to actually perceived, audible impressions? It seems to me the Optane has become the default go-to item as boot memory, following the lead of a few other individuals, that as of late - or ever, really? - hasn't undergone testing against cheaper M.2 NVMe devices, devices that I gather would still make a worthwhile difference compared to SATA SSD's here?

As poster @lmitche wrote:
 

Quote

Tuning the OS to play music is all about reducing non-music related activity. Your newly issued optane drive is the fastest drive in the world. Guess what, if your computer spends less time waiting for a Windows log file to write to completion, the more time it has to make music.  You are hearing the benefits of this.  

 

Why would the Optane be exclusive to these traits?

 

Quote

 

For the lowest latency out there, just go for something with Samsung B-Die and then try RAM disk instead of SSD

 

http://jplay.eu/forum/index.php?/topic/2658-ramdisk-with-windows-server-20162019-configuration-guide/

 

16GB of RAM should be more than enough

 

https://twitter.com/mohrpheus78/status/1298967049289506821

 

Thanks for the heads-up on this. I'll keep it in mind for future (Windows Server) consideration, but as is isn't an option being that I use the consumer variant of Windows 10 (Pro 64-bit) and therefore takes up too much space.  

 

Quote

IMHO if you're still having any doubt about SGM Extreme and others, no one will stop you to buy whole bunch of drives and compare all of them on your own. That's the only to find out what really works best for you.

 

It's not doubt over the SGM Extreme as such (I can only gather it must be a sonic bliss as a server-source device), but sure; the only way I can ever become the wiser on my inquiry or otherwise get some first-person perspective into the matter is to roll up my sleeves, pony up the dough and start comparing a bunch M.2 NVMe devices. I'd just appreciate some insight into whether others has done similarly and learn of their views to begin with, but going by your evasive answer it would seem it's a limited field of experience.  

 

My Kingston A2000 M.2 NVMe SSD (500GB) should arrive tomorrow, and I would expect the upcoming weekend to have some first impressions ready. We'll see..

 

Thanks for your insights so far. 

[phusis]

10 hours ago, lmitche said:

Two reasons:

 

1) The Optane drives are not made from silicon rather an amorphous material that was invented by the same fellow that invented the Nickel Metal Hydride batteries used in hybrid cars. His name is Stanford R. Ovshinsky.

 

2) Optane drives random write memory speed and latency is faster than anything else out there. Our music players are constantly spitting out data writes in small block sizes that Optane excels at.

 

So, "1)" leads to "2)," or how does '1)' have significance in itself from the perspective of sound?

 

Random write memory speeds for the Kingston A2000 is favorably tested:
 

Quote

The performance of this SSD is unprecedented; not in a sense of pure speeds, as there are plenty of PCI-E 3.0 drives that are faster in sequential tests, but in the way that it defies underwhelming sequential speeds to sprint past its rivals in more true-to-life file-transfer tests. We’ve tested plenty of SSDs that wowed in CrystalDiskMark only to crumble in our own tests, but never one that essentially does it the other way around.

https://www.expertreviews.co.uk/kingston/1411219/kingston-a2000-review

 

10 hours ago, lmitche said:

 

I recently tested new silicon nvme drives from Samsung and Intel and Optane beats them hands down.

 

Larry

 

I might add: is that a fact? What's entailed more specifically in "beats them hands down"?

[phusis]

22 hours ago, lmitche said:

More transparency, space and a denser image are the result of switching to an Optane drive for the OS and metadata storage for music players. Both generate numerous disk writes during playback.

 

It is the noise (RFI/EMI) generated by the silicon SSDs during write processes that is the problem here, not bit errors.

 

As I quoted from you earlier:

 

Quote

Tuning the OS to play music is all about reducing non-music related activity. Your newly issued optane drive is the fastest drive in the world. Guess what, if your computer spends less time waiting for a Windows log file to write to completion, the more time it has to make music.  You are hearing the benefits of this.  

 

So, in the meantime another sonic influence is brought into the mix (or..?) with:

 

Quote

The Optane drives are not made from silicon rather an amorphous material that was invented by the same fellow that invented the Nickel Metal Hydride batteries used in hybrid cars. His name is Stanford R. Ovshinsky

 

.. meaning this elusive "amorphous material," invented by a guy it's supposedly important to know of, is another or the primary reason for this?:
 

Quote

It is the noise (RFI/EMI) generated by the silicon SSDs during write processes that is the problem here, not bit errors.

 

"Bit errors" (jitter?), certainly as a formulation, was never on my mind in this particular context, but rather what's expressed with noise generated by (higher) power consumption, latency, and random read/write speed (for which there appears to be a range of measuring models, some more "real-world" than others) - maybe this is what you mean by "bit errors" in the first place. In these fields at least reviews indicate other and cheaper alternatives than the Optane's are now comparable. 

 

 

 

@seeteeyou

 

In light of the popular use of the Optane's my aim was/is to hone in on the essential parameters that supposedly accounts for their sonic pedigree, and you'd think these're objective in nature given the Optane's almost unanimous implementation in "reaching for the sky" music servers. If perceived sound quality is predominantly a subjective matter, then why wouldn't more use NVMe devices from Samsung, ADATA, Kingston and others, if there was indeed a larger experimental field to compare them with the Optane's? 

 

And btw.: I'm not "knocking" the Intel optane's. I'm just interested in the reasons for their sonic prowess, and whether someone has investigated into this more thoroughly. It all started out with the possible sonic traits implementing M.2 NVMe SSD's for OS storage, and it turned into "the Optane's." I'm just wondering why..

[phusis]

On 9/16/2020 at 2:35 PM, lmitche said:

Hi Phisus,

 

I'll let that one slip..

 

On 9/16/2020 at 2:35 PM, lmitche said:

 

Maybe it's time for you to run the experiment yourself.  32gb Optane drives are available on Amazon for $76 this morning. Please let us know what you observe.

 

May wait with the Optane for OS duties for now, but one day should try it out (possibly with Audio Linux(?) and not what I use now, Win10 Pro 64-bit).

 

But observe I did in the best empirical fashion I could muster; first impressions are in with the Kingston A2000 as boot device for my Win10, JRiver MC26 (soon MC27) and a few other programs, and compared to the Samsung EVO840 SATA SSD boot drive (120GB) it's a slightly darker, calmer and possibly (subjectively) a bit better resolved sounding as well. Background seems quieter perhaps, but the thing is I find the overall balance to tip just a wee bit too much to a, again, dark-ish sonic imprinting, robbing the sound of some liveliness. It would certainly require a readjustment of the main speaker placement and tweaking some values in my digital cross-over (as is oftentimes needed when new components/software are introduced), but I'm thinking differently right now:

I'm leaning towards staying with the old Samsung boot drive for now, and maybe try out the Kingston M.2 drive as storage device for my music library instead of my build-in Seagate 1.5GB HDD/NAS, and while I'm at it uncouple the SATA-loaded HDD. 

What's the thought on using M.2 NVMe SSD's as music storage? 

 

On 9/16/2020 at 2:35 PM, lmitche said:

 

Your questions here have been answered in good faith. We are empiricists here, basing our judgements on observations made with our ears. At times we can only conjecture about why something sounds best.

 

I posed my questions with good intentions as well, inquisitive they may have seemed. I can certainly understand the complicated nature in giving reasons for why something sounds the way it does.  

/Mikael

[phusis]

Thanks for your replies, @lmitche and @StreamFidelity. 

 

About a week ago it seems my old Samsung EVO840 SATA boot drive (C:) had checked out, so to speak; some "read error" occurred when I powered up my server (maybe linked to my enabling X.M.P. in BIOS?, that I've since disabled again), and being several re-boot attempts failed I headed over to BIOS and changed the boot drive to the Kingston A2000 M.2 unit, which is what I'm running still, and after some tweaks it's not for the worse - sonically speaking or otherwise. I then disconnecting two SATA drives, namely named EVO840 and a Seagate HDD, while also disabling 3 out 4 SATA ports in BIOS - leaving only one enabled that's connecting my Asus Blu-ray drive.

 

I've then spent some days continuing to fiddle with BIOS settings disabling the likes of Intel "Turbo boost" options, Speed Shift, C-States, audio controller, EIST and others. Hyper Threading it seems is not supported by my i3 9100 CPU, and I disabled C-States because I experienced music stuttering after (and only after) some hours of playback, suspecting that the CPU went into an idling state. It's a bit of a trial and error process navigating with these BIOS tweaks trying to find out what they're good for, and if they have any positive (or negative) effect on sound quality.

 

All being told and while still in the process of tweaking I find the current level of sound quality has advanced into a more "analogue," stable and coherent presentation. 

[phusis]

For what it's worth: 

 

Today found out that my Samsung EVO 840 SATA SSD was alive and well (about a week ago received a "read error" notification when trying to boot via this drive following a series of boot sequence changes between that and the Kingston M.2, and after several boot attempts failed assumed it was dead), and being that I initially preferred the Samsung over the Kingston as boot drive then continued BIOS tweaking with the former as C: drive. 

 

A friend of mine and I made the following observations by ear in the tweaking process via my budget HTPC: X.M.P. disabled proved the better sonic option (i.e.: 2133MHz), and we then set our minds/ears trying to figure out if the different options in 'Memory Enhancement Settings' made a sonic difference - they do. Shifting between each single setting from 'Relax OC' to 'High Density' (we passed on DDR 4500+) we honed in on exactly those two "outer" settings as the best sounding ones; 'Relax OC' sported a musically very inspiring and whole presentation albeit a wee bit too rounded, with 'High Density' being, surprisingly, of a similar-ish sonic cloth though noticeably more insightful, distinct and resolved. More monitor in the best sense of the word with a lovely sonic balance. 'High Performance' and 'High Frequency' settings sounded thinner and more overtly detail-oriented, while 'Normal' - though quite pleasing sounding - appeared rather dull overall. 

 

It's amazing how easily audible these different settings were to our ears. Theory had to go by the wayside (i.e.: theory-based assumptions into sonic implications), and listening ruled.