Intel Optane 900P SSD

While restricted to desktop systems (for now), these are true game changers to the storage industry.

Even the 280GB model (tested) today is worthy to be installed in any DT workstation for the O/S and Programs and even some DATA too - the more they're filled/used - the faster they are.

For the remainder of your DATA? A 2TB+ Samsung 960 is okay for now. It will even feel faster than when it was used as an O/S drive too....

I can finally visualize my next platforms taking shape...

They will be built around Optane tech and will only serve to make all notebook 'workstations' feel more inadequate than ever, once again.

See:
https://www.anandtech.com/show/11953/the-intel-optane-ssd-900p-review

See:
http://www.tomshardware.com/reviews/intel-optane-ssd-900p-3d-xpoint,5292.html

 

http://wccftech.com/intel-optane-ssd-900p-launch/

 

This is what I was waiting for from Optane. Intel promised this and it delivered. What am I talking about?

"Optane 900P also delivers an entire system boost".

Now to get this into the hands of mere mortals and see what it does for our own workflows.

This is 2.75x over what a Samsung 960 Pro delivers. (What we've been stuck with for too, too many years now of Samsung's thermally throttling (at least in notebooks) junk).

From the link in the quote below:

The 'caching' that is done from inside a CPU to the external DRAM and finally the storage medium is finally catching up with what modern CPU+RAM platforms require: MORE DATA, FASTER.

People can moan and complain about 5% or less gains from Intel over the last decade or so. But the path they were on during all that time is ready to pay off handsomely. Limited to DT's for now, but this is gen 1 day 3... can't wait to see this in a notebook that I would want to buy and use around clients (i.e. ThinkPAD).

 

Just in case people think Intel is sitting on Optane tech and twiddling their thumbs...

Intel just doesn't introduce new tech and see which ones stick - they know the bottlenecks of their processors and are actively working on minimizing the effect on users workflows.

While the following requires programmers to learn new code - M/B makers to add new hardware/logic - the improvements promised will truly take the 3D Xpoint tech to the '1000x' greater than nand that Intel first hinted at when it teased us 2 years ago with Xpoint tech.


See:
https://software.intel.com/en-us/ar...programming-with-persistent-memory-from-intel

For my friends that have a ton of time on their hands:

hmscott

See:

 

LOOL
Maybe bro @hmscott have seen it before

 

For the average person, a gamer, doing browsing and video watching in between, maybe doing some school work editing / etc, even a PCIE x4 SSD costing 2x the cost for the same sized storage SATA SSD is overkill.

Now Optane introduces another option with 2.5x the cost of PCIE x4, and the market for that need drops another order of magnitude

Optane, to pay for itself, is going to require very specific needs and testing to prove it will pay off if used.

Maybe for you it will be what you've been looking for since SSD's promised you better workflow performance, and failed?

Too bad the new Optane's throughput is only 2.5GB/sec vs 3.2GB/sec with PCIE x4 NVme SSD's, maybe Intel needs another generational PCIE throughput increase, or a special higher speed interface to design toward?

 

I think you're missing the bigger picture. This is effectively gen 1 day 6 of Optane SSD's.

Their current form isn't even close to where Intel is envisioning them for optimal leverage in a compute platform. What I read from that? SATA and NVMe is effectively dead for Intel (at least, long term). The next leap in platforms isn't going to be even more cores on a cpu... it will be the enabling of (fast) storage to be used like RAM is today. Xpoint allows for that. PCIe never will.

Even on the horribly inefficient PCIe protocol, XPoint delivers in spades. The average person doesn't care, granted. But even they will be able to reap the benefits of this tech if/when it is delivered and matures. Benefits like instant on. Benefits like greater battery life. It may even come to pass that certain platforms may be cheaper to build because of XPoint (after all; they can replace both the RAM and the storage components in an optimal setup and if not fully replace those parts; at least partially...).

At the MSRP prices listed: Optane SSD's will have paid for themselves for every DT user that even remotely values their time. Almost regardless of actual workload (note; if they value their time; their workload is probably not posting to fb all day...).

SSD's still promise performance that they can't deliver - at least not sustained... But with OP'ing; they haven't failed me totally (not since I OP'd the first Intel SSD I tried that on 5 or 6 years ago...). But what they don't do is increase my productivity greatly, given the 'scores' they can post vs. HDD's (even over a mechanical storage subsystem that is also properly and optimally setup). Why do I use them then? Mainly because the cost difference is within my budget and any increase in actual performance is effectively a bonus (yeah; there are a few places where they shine; pdf editing, database manipulations, Scratch/Temp disks @ 50% OP'ing - but those few examples do not make up a large portion of my workflows/workloads, overall). There are other reasons I use them too (i.e. my competitors use them; I have too) but HUGE jumps in productivity? No, not there yet (or ever).

Optane promises to change that, drastically. If not today; soon (enough). Disruptive change that will change the face of compute forever.

Don't concentrate too much on the raw 'scores' when you want to compare a new product. In actual workloads, even compromised by the PCIe bus today, there is nothing that comes close to Optane - look at post 3 again.

Think of it like enjoying up to 240FPS in your favorite game with the currently best gpu available and a new gpu card comes out that offers 150FPS. You'd scoff, immediately. Yet; if you delve a little deeper (or; just used the new component in your actual workflows for a few days...); you may find that the 240FPS gpu is actually a maximum and fluctuates from 120FPS to even a low of 20FPS minimums - while the new card gives 150FPS - sustained performance that barely fluctuates. This is what Optane is to NVMe based nand drives today.

Intel doesn't have to develop a new PCIe increase - they've already written it off (as they should - look at the slides again above). What they already have and are working towards is the RAM interface. No storage drivers involved. More direct access to the CPU registers. Effectively HUGE amounts of RAM vs. DIMM - and all this at lower prices than previous components too.

This will rewrite the time honoured equation of CPU + RAM = Work done
to: CPU + Persistent Storage + RAM = Work DONE

A cursory look at this new tech and the products now offered may give some here reason to write this off as another Intel money grab. Giving very little and charging a lot for it.

The truth is a little more complicated than that. Moving forward on Optane without making money isn't doing anyone favors - yeah; not even us mere consumers. Taking the time to put the required systems in place - at the O/S and programming levels - is more important than spewing out products without a plan.

I don't claim to know Intel's plans any more than anyone else. But what they do show us makes me confident that Intel is taking care of me (i.e. their customers) because they're taking care of themselves - as they should.

After all, I'm not going to invent, market and monetize a new type of memory... all the while keeping investors, customers and everyone in between happy - and also while not bankrupting the company to do so.

But I will be willing to buy all that I can from everything I've seen so far.

 

Well then, you are also saying not to buy what is being shipped today? As it's running now and benchmarking now, it's not the true promise of Optane as Intel is working toward.

That's what I was saying too

 

Nah; I didn't say that.

Baby steps, baby steps.

What is supposed to be shipping today (which I can't source, currently, btw...) is still leaps and bounds above what we had before Oct 27, 2017. In price, in performance and in usability in the types of systems we all have running today.

Like I keep saying; directly compare a new component/platform in the exact same workflow/workloads you'd use in your current platform and it* will speak for itself. *it = 'is it worth it'.

Given what I researched so far; I can't see how Optane SSD's (even at a 'puny' 280GB size) would be considered 'not worthy' in my DT workflows.

Ideally, I would have 1 for my O/S + Program drive, I would have another few for my Scratch/Temp disks - and I could even have one or two for my WIP (work in progress) drives. But now, the situation has changed. We're not talking one drive - we're talking a handful or two of them ($$$$$) - and, I'd want the 480GB models throughout... - That puts the storage subsystem at a price level of the rest of the platform, or higher. Now multiply that by the number of platforms I'd need that on (a few dozen). That is why I don't buy blindly or on 'faith'.

If/when I'm able to work on such a setup as described above, I'll know if doubling the cost of the platform is worth doing for my workloads/workflows. But right now? Even a single 480GB Optane SSD driving the O/S + Programs seems to offer enough extra oomph for a mere ~$600 above what any PCIe x4 SSD can today to be called 'worth it' to me, personally (I value 'snap', very highly).

When they're available from my favorite sources; I'll be able to answer those questions for myself.

Right now; patience is all that is required on my part.

But to ignore the potential increase in productivity offered by any current product is not in any way shape or form what I said or meant.

 

Well then, when you get your Optane units in house, please be sure and post benchmark and work-flow results so we can all gauge whether paying 5x as much for the same sized storage is worth it for our use.

 

I see what you did there.

But I don't do that.

(see what I did)?

 

Ha, you're talking yourself out of getting new Optane hardware, yup - I saw that

 

Not even close!

 

Optane is a niche of a niche product, not worth the costs, and it's not going anywhere just like the previous iteration as Optane cache.

Not until Optane storage size, cost, and performance matches competing products will it be mainstream, and that's going to be a long time.

Whether you buy it or aren't going to buy it, test it or aren't going to test it, whether you are going to allow yourself to be pinned down to either or any state of being, it's not really worth spending time discussing it with you right now, is it?

 

I see you're still missing the point of Optane, now and in the near future.

Optane is a niche product like oxygen is to the internal combustion engine.

I am praising what Optane promises and also what it seems to deliver today/now (yeah; even the tiny 16GB/32GB Optane Memory is fantastic).

How in the world can I be as unoptimistic as you?

Oh yeah; you have a thing against Intel.

 

No, it has nothing to do with Intel, it has to do with performance / value / usefulness. It's still in development, it's not ready for prime time, as shown with the progression of implementation, with future elements still not realized that will allow the performance promised - yet not delivered - to arrive.

Intel isn't the only player...

Micron's Intel Optane competitor will ship later this year
Micron's QuantX will ship later this year, which will then be adopted by storage makers.
MAR 27, 2017 11:28 AM PT
https://www.pcworld.com/article/318...-intel-optane-competitor-later-this-year.html

"Internal benchmarks showed the Optane SSD up to eight times faster in random read and write than conventional SSDs. Intel initially hyped the 3D Xpoint technology being 1,000 times faster than conventional NAND flash, but the company later clarified that applies only at the cell level, not to SSD media like Optane."

From 2015...

3D XPoint™ Technology Revolutionizes Storage Memory


Intel 3D XPoint Technology

 

For future use, will DRAM be cheaper than today and in great size to match Optane? If not, will there be programmatic implementations that demand Optane for persistent memory?

I found this interesting...not necessarily to encourage purchasing "Optane" today for general use, but maybe for development, in preparation for roll out into production - someday.

CppCon 2017: Tomasz Kapela “C++ and Persistent Memory Technologies, Like Intel's 3D-XPoint”

Some of the instructions and techniques mentioned indicate useful debugging ideas for why Optane installations might not be helping or performing as expected.

 

Here you go, next time you need to render your movie or 1.1 billion particles in your unreleased game...

Intel's crazy-fast Optane SSD 900P finally hits the street
Finally, consumers (with fat wallets) can buy into Optane's blistering speed.
https://www.pcworld.com/article/3235284/storage/intel-optane-ssd-900p.html

"Got a render job for your movie or game involving 1.1 billion particles? Then you may just want Intel's new Optane SSD 900P for the job. The long-awaited drive series, launched Friday, was able to cut down the rendering of a 7-second scene from 17.4 hours to 6.3 hours, Intel said."

 

And let AMD instead increase cores for mainstream and Hedt? Ain’t happening, bruh

 

What does posting what we already know (and reposting what I've already shown...) have to do with your position that you think Intel products are less than useful to you?

Everything Intel has said is true. It just isn't to your satisfaction, is all.

Since Intel and Micron jointly developed this tech; they're not competitors - more like partners, imo.

Of course there will be teething problems on day 6 of a new tech. I think even you can realize that.

In the final analysis your conclusion is just plain wrong.

When you think a product you haven't tested isn't ready for prime time, your argument falls apart really fast, btw. Especially when you also show it to be a ~2.75x increase over a fast PCIe nand based SSD too.

See:
http://www.pcgamer.com/rumor-intel-may-release-3d-xpoint-system-memory-in-2018/

That 'someday' is coming up on us really fast. 2018 may or may not be the year we see/hear about it. But you can be sure that Intel/Micron isn't going to let a decade of R&D go away with a whimper.

Ignore the forest at your own risk. Be careful out there; those trees are made out of wood.

 

AMD already has more cores than Intel does today. Doesn't shift the performance outcome though for 99.9% of users.

If/when more cores = more performance for mainstream users? I have no doubt Intel is working on that as we speak.

Today; over 8 cores gives you increasingly smaller return for your $$$$'s in almost any workstation class workload - and in those workflows where more cores do make a difference - that segment of the market is hardly significant to Intel today. By the time that changes (programmers; get to work), Optane Dimms (and I hope from other manufacturers too...) will be here ready to make those extra cores work as hard as possible. Whether Intel or AMD gives us the most dense cores/productivity at that time? Who cares?

I know I'll still be buying the system(s) that give me what I need (stability/reliability/dependability/productivity) at that time too.

 

Read carefully, bruh.... Mainstream and HEDT Would Intel release 7960x as highest core count chips for consumers <if> Amd pushed out 18 cores Threadripper earlier this year? Come on bruh

 

Note that I edited my post above.

We don't know if Intel would have released anything if AMD had kept napping.

But that doesn't materially change what I wrote though.

 

Correct has to be correct

 

Log is Wood

 

Lol...

"The log, from Blammo".

 

See:
https://www.anandtech.com/show/11930/intel-optane-ssd-dc-p4800x-750gb-handson-review/6


So the Intel Optane 280GB 900p SSD matches the newest Intel Optane DC P4800X 750GB SSD in the Aerospike Certification Tool (ACT).

At better than a 10x improvement over an Intel SSD DC P3608 (with single controller/800GB, non-raid config) in that specific load. Not bad for two years work Intel.

Ack! I wish they would release faster, faster.


Give Optane the right type of workload and it will pay for itself in days...

 

I personally don't feel any excitement at all. I mean, look at the thing! It's fracking massive, with giant heatsink on top, and very low density. No way it'll come as m.2 module of reasonable capacity anytime soon - my best bet would be yet another m.3 monstrocity. NAND prices ain't falling, while RAM prices just climbed a new record; the industry is one big disappointment, it's steadily moving back with this insane pricing - not forward.

 

That is one way of looking at it, granted. And for yourself, you may be 100% right.

I don't move/buy tech just because I can; they need to prove themselves in 'my' workflows - not the other way around.

How I see it is this is tremendous progress since a consumer Optane SSD was introduced 12 days ago...

This doesn't just enhance a specific workflow: it affects the whole O/S is ways we can't experience yet (because: availability). The 280GB is enough for my DT workflows - as no OP'ing is required or even remotely needed from everything we've seen so far - at least for an O/S drive and a Scratch Disk drive. The 750GB version would be better, of course. But it doesn't offer any more performance - just simple capacity. Capacity that I could put to use, yeah, but as I don't see one at the corner tech outlet; I'm not rushing to get one (even to 'try').

What I do see is how Optane Dimms will cause even more disruptive platform changes. Even just to come to this forum in the not too distant future.

I'm excited, but controlled. The product, the press releases and the commitment evident from Intel towards Optane is not enough to do something 'stupid', yet.

When I can pick one up locally and test at my leisure, then I may become uncontrollable. Maybe.

 

Nothing will happen for notebooks, before Samsung push out Z-SSD's with new Z-NAND, the power consumtion go down and harder competition. For now is Optane ssd's more for desktops.

 

It will happen for notebooks. Not throttle happy NVMe M.2 implementations though; the real deal.

Optane DIMMs. Can't wait!

 

Yeah, I know, but not now


FYI.
Intel Optane SSD DC P4800X Now Available in 750GB Version for roughly 3000 Euro-guru3d.com

@tilleroftheearth How many of them is your target? I'm sure you can find a ok deal if you buy 10 of them

 

Not now, but possibly next year (earlier the better, of course).

10 is a good target to do internal testing with. I don't like 'okay' deals. I always go for Whoa! deals when I upgrade my entire operation.

 

REVIEW
Intel SSD 900P review: Optane finally starts to live up to the hype-pcworld.com

 

Thanks for the link!

It's interesting to see the 900P in more mundane workloads where it still excels for the most part when price and overall performance is equally important.

The point that the almost year old Optane Memory was a 'snooze fest' shows how much they are underestimating (at the time and even now) what Optane holds in store in the future.

With mere days of consumer Optane based SSD's being introduced, comparing them to older/mature/fully optimized traditional 'nand based SSD's - it is expected that they won't sweep the performance categories (after all; the PCIe connection is still the limiting factor holding them all back).

Optane DIMM's and the platform(s) that support them is where I'm looking forward to now. Optane is 'proven' enough (to me) at this point. Let's get to where the designers actually envisioned already!

PCIe is a poor connection to the rest of the system for such a medium.

With DIMM's and their direct connection to the CPU is where this battle will really be won.

If the above quote can be taken at face value, Optane is already 4x faster than anything they had tested before (at least for the BM runs they were prepared to do).

They're also missing the boat on expecting Optane to match DRAM speed. It won't. Especially not on PCIe connections.

What Optane DIMM's will offer is access to (persistent) data that is closer to DRAM speeds than to HDD/SSD speeds (whether SATA/PCIe - doesn't matter). That is where the performance of any platform based on OD's will explode.

Effectively giving consumers > 64GB memory and possibly 'instant on' capabilities of the O/S when they're finally tuned to this disruptive tech.

Not holding my breath to see this anytime soon on a notebook platform, for example. But see it we will. (There is nowhere else to go in the SATA/PCIe world of storage - even if they seem better is some/most comparisons today).

 

Samung Going After Intel Optane with Z-NAND SZ985 Storage Unit
"Z-NAND seems to be tweaked Samsung flash technology. In essence, the claim is Samsung is going back to the bare roots, by using SLC NAND (1 bit/cell). That SLC written NAND is paired with a blazingly fast NVMe controller - and all customised controller to speed things up."

 

I read that a few days ago when I thought I also posted it here? Hmmm....

Either way; not much news nor competition to Optane SSD's (and even further away from competing with Optane Dimms' sometime next year...).

Not only is the nand slow (yeah; even SLC - which it isn't***) - the PCIe/NVMe interface/controller is too slow to make much difference in everyday responsiveness/productivity like Optane SSD already give.


*** If this was SLC nand, Samsung would be shouting this from the rooftops. This isn't.

See:
http://www.samsung.com/us/labs/pdfs/collateral/Samsung_Z-NAND_Technology_Brief_v5.pdf

See the Samsung pdf above? Not one mention of SLC nand. This is Samsung 3D Nand with a new controller.


The performance improvements seem good on (Samsungs) papers... What is even more important is how easily those goals can be achieved in a live, production system. Especially as Optane seems to be closer to double the productivity leader (at least double...), not to mention available in bigger capacities too.

For the companies that bet on nand based computing for the near/medium term; this is a good dead end product.

For the companies that bet on merging RAM and Storage - Optane is still the best long term bet with the highest (possible) rewards.

 

The main point or a better words <Higher competition>. Intel can't demand whatever they want for Optane anymore, if Samsung's Z-NAND drives work well and can hold reasonable prices. To the best for us all. Optane will be for the very few with the prices they demand for it now. It's welcome news

 

imo, Samsung's products are overpriced and under deliver. I don't see how or why they would give their consumers a deal on Z-N products when they can easily charge as much (and in their minds; probably more) than Intel does currently.

See:
https://www.tweaktown.com/reviews/8...280gb-480gb-aic-nvme-pcie-review/index11.html

Of course, when tweaktown actually tests Z-N from Samsung we'll have a better idea of how far they're behind Intel (almost a year already and the counter hasn't stopped yet - and won't - until Z-N is delivered in a purchasable product...).

 

Yeah, I know Samsung NVMe is overpriced. But for me was 950 Pro the only options for NVMe ssd’s. And Samsung is still there.

See the Toshiba NVMe ssd's overheating and then dying in Dell's newer Awbook’s. Even better cooling can’t save the trash.

The hope is that Samsung can staggering Intel’s nasty price policy for Optane. Even with somewhat better performance... Intel can’t escape from the reality when the competition finally will start.

And we need seeing Optane in laptops... Aka form factor, lower wattage, decent sizes and more normal prices. We ain’t here yet.

 

We do need to see Optane in notebooks! But from what I'm seeing, it won't happen for a while.

At least not while 16GB Optane chips are the biggest available... I'm in no rush to upgrade right now. If they were actually available, I would be.

Intel is taking it slow and steady; just like I want them to. When and if Optane production has a paradigm shift and allows M.2 products to become available at the capacity and power levels that will make sense in a notebook/mobile system, Intel will release them.

As you say; they won't wait/sit on that tech if they have it available today. The would be competitors are knocking on their door...


And tech is moving too slow for me too - and I'm aging faster with each passing minute - but since I can't invent this stuff for my own use, I choose to enjoy what I'm able to have today.

 

Intel Optane SSD 900P in the test: The benchmark emperor is hardly faster in everyday life-computerbase.de

Conclusion
"The benchmarks speak for themselves: The Intel Optane SSD 900P delivers the lowest latencies and is vastly superior to any conventional SSD in 4K Random Read. Sequentially, data is transferred at high speed but not at full speed. Overall, the performance is enough to put in many benchmarks to the top. In some tests, the distance to the previously fastest SSD in the test field is huge. If continuously high continuous power is required, the Optane 900P plays in a different league. The typical break-in of the IOPS with NAND Flash-based SSDs is not present in the 3D XPoint memory, the advantage correspondingly huge."

The great potential often remains unused.
"Thus, the Intel SSD 900P meets the high expectations in this regard. And yet disillusionment with regard to everyday use is widespread. In the field tests, the Optane SSD can not retrieve its potential and provides barely noticeable advantages over other fast PCIe SSDs. Applications will not start or install faster, with the rest of the system being able to brake. At least when unpacking and copying with parallel read / write load, it sets new record times."

 

I very much hope the massive heatsink on 900P is a marketing gimmick, not obligatory to keep the thing within reasonable temps.

 

Thanks for the translation.

btw, anybody buying these drives for 'everyday' life has too much disposable $$$$.

These drives are meant to be used and used hard. Nothing else like them at this moment.

I had a chance to briefly play with a client's Optane Memory (16GB) based platform powering a 4TB 5400 RPM HDD as the O/S, Program and DATA drive. In a few mere seconds I had convinced myself that it was as fast - at least in the ballpark - if not faster in some aspects, of any pure SSD system I've setup personally (multiple SATA III drives or even desktop, multiple, M.2 storage subsystems - with proper/massive air cooling).​

A pure Optane SSD will be on a different level and something that I would easily and willing pay for (simply for the increased 'snappiness' in simply navigating the O/S). The increased performance in actual workloads for paid work would be icing on the cake - not to mention paying for the small increase in $$/GB penalty totally a non-issue for me or anyone with a paid 'work', workload.

 

No gimmick. Nand/Optane cells don't like (excess) heat - just like most electronics too.

Glad that Intel simply built them as they needed to be built and not try to cheap out to make them $1 cheaper but infinitely less useful for day in/day out, continuous workloads.

Samsung is also putting (more) real heat sinks on it's newest OEM M.2 drives - even if the platforms/chassis themselves is where the real cooling design work needs to be done, imo.

Too bad throttling is such a big issue with M.2 drives - so much promise, yet so much crap delivered...

See:
http://www.tomshardware.com/reviews/samsung-pm981-980-nvme-ssd,5323-2.html

And the above is just 'easy-peasy' sequential transfers. The more important 'sustained' mixed random read/write workloads will offer just a fraction of what SATA III promises, let alone what PCIe actually delivers. Sigh...

 

See:
http://www.tomshardware.com/reviews/intel-optane-ssd-900p-3d-xpoint,5292-3.html

RAIDing 2 Optane drives (or; how to neuter the performance advantage Optane offers for a desktop workstation user).

What most of the above quotes are bluntly pointing out is that we desperately need an O/S or at the very least; a new set of drivers for current O/S's that more fully take advantage of the flash based storage we have available - not to mention what it would do for Optane's impact in every aspect and range of computing workloads (from 'light' to 'extreme').

Unless working with large video/sound files on a daily basis, or at least for a large part of your workflows each day/week/month, RAID0 is a useless exercise in the pursuit of more performance for most desktop workloads (just like I discovered around 2003/2004 already and started using the fastest available drives (WD Raptors).

I am eagerly waiting for Tomshardware to optimally tune their test system for Optane and all other SSD's to show what they're really capable of today (things like disabling the PCI Express power saving features, for one example).

But mostly; I'm waiting for an SSD/X-Point based version of Win10x64Pro. Can't come fast enough.

The above is blatantly obvious when comparing (for responsiveness - not overall productivity) a SB based platform to anything current with identical SSD's, RAM (capacity) and O/S builds (i.e. clean install).

Come on Intel, AMD, MS and anyone else who purports to value high performance and a more fluid user experience - turn up the heat already!

 

THIS! And almost none of today's laptops have 2x 2.5" slots for old spinning Hdd's.

 

While I don't really agree with all the suggestions in the following guide from Intel, at least they do point out that testing in 'your' real world workloads is the metric that truly matters.

See:
https://www.intel.com/content/dam/s...orage/IntelOptaneSSD_900P_SeriesEvalGuide.PDF


What is interesting is the setting changes required to get the most repeatable performance from Optane devices. Note that this isn't the maximum performance, nor the highest performance with the lowest power impact - just the most repeatable platform configuration possible.

While they do indicate the importance of their NVMe driver for their products vs. the generic driver available in Win10 (yeah; Win10 highly recommended...)... they stop short of addressing the current O/S's pitfalls when it comes to optimally handling flash/X-Point based storage.

(Silly PDF... doesn't allow copying of text... see last paragraph on page 6 about the importance of real world testing).

Another item that needs to be pointed out: Intel recommends leaving the drive on at least 3 hours before attempting to characterize it's performance. In my experience with just/even 'normal' SSD's since ~2009; the second attempt to do a clean install and setup (drivers, updates, programs and DATA) on the same, brand new SSD, is always a much more consistent affair than the first run which always makes me thing it will (almost) fail at any time because it is so laggy/unresponsive vs. any time afterwards (when it is in a 'well-used' state).


For myself, there is no point in 'tuning' the system just to see what is the best a component can do if I'm not willing to leave it in this 'tuned' state to run my actual workloads though...

I want to take advantage of all the features of a new platform, including C states (for one).

This is why I simply do a clean install of an O/S and use the storage component I am testing in the exact same configuration as if I had bought it for that use during the test (and not as a second drive, with the O/S on another drive... - not with a secure erase performed before each test - and not with anything else different from what I'm doing already (and have been doing for the past decade... more or less...).

This allows me to see the real and total progress (or not) made by all the cumulative changes made to all aspects of the platform I'm considering - and therefore give me a real basis to make a purchase decision that won't be riddled by "but's" of why it doesn't live up to the testing platform when/if I actually go live with it.

The whole is always more than the sum of the parts in computer platforms, audio (i.e. 'music') systems and any other area (eg. home theatre setups, high end photography, etc...) where the components don't matter as much as the interaction and synergy (or lack thereof) between those same components.

In each round of testing for my platform upgrades I do occasionally test for changes in the O/S as suggested by the PDF attached. So far, the real world gains vs. the real world trade offs have not been worth it.

Optane SSD's may be what changes that for me.

 

Intel Mentions 900P 960GB and1.5TB - Tomshardware.com @ole!!!

 

though i am only use it in my everyday life there are some benefits. and tbh its also for my own knowledge because without getting it, we dont really know how and what software will benefit from the higher 4k at QD1, only by getting it and compare then we'll know which software is well optimized or not and coded properly.

i got plenty of things to put this to the test but one think i know for sure will benefit A LOT from traditional NVME SSD is windows index file search.. you ever try to search file on HDD and its hella slow, replace that with SSD and became much faster, however after turning on index search specific word in text files within an entire drive/folder, speed reduce to that of a HDD once again while using a SSD.

optane will vastly improve in these area and reduce wait time with it's 4k QD1 read. other than that, like i mentioned previously, overall snappiness and to find out which software gets the boost!