Samsung 950 Pro PCIe SSD: Not Performing as Advertised

Had high hopes for the Samsung 950 Pro PCIe SSD after reading the advertised transfer rates:



Unfortunately it seems to fall rather short of these expectations in 4KB Random write:

Hopefully this is something that might be fixed with firmware update. At this point, however, seems better to wait a bit longer for future PCIe SSD models from Mushkin or other manufacturers.

 

For me, the interesting part of the first table is how the power consumption of the PCIe x 4 interface drives compares to the SATA. The extra potential speed comes at a price and it will be interesting to see if any of the notebook manufacturers make specific provision for dissipating the extra heat.

John

 

maybe that new powersaving algorithm they implemented in the 950 pro in order to prevent the heat throttling seen in the sm951 is causing the lower performance values...lets call it artificial power throttling to avoid thermal throttling *lol*

Sent from my Nexus 5 using Tapatalk

 

The truth about these M.2 drives is that NVMe wasn't created for them...

It was created for the X-Point Intel technology that will begin showing up in the next few months (if we're lucky).

NVMe on anything other than an Intel drive today is just a check box for the spec-hungry, nothing else. Issues with lower than SATA 'scores', out of whack power requirements and built in throttling issues (to not only the SSD but to the rest of the components in a mobile system too...) are the signs that these in-between beta's should continue to be ignored by most users. And almost all mobile users, period.

The final nail in the coffin for these useless examples of storage subsystem components in late 2015 is the capacities they are being offered in. 256GB and 512GB (nominal) capacity... and, more expensive too?

Anyone can see that these are toys for people with more $$$ than brains.

Having the latest does not always mean having the greatest when there is a new 'standard' to help confuse people with.

If we were offered a 2TB+ M.2 SSD w/NVMe that cost less (per GB), stayed cooler, sipped less power and was as performance oriented as the SanDisk Extreme Pro 960GB (in actual, real world usage) then we could look forward to these drives - as the NVMe spec's could almost guarantee an increase in responsiveness from that gold standard of a workhorse SSD.

But as it is, we are given the 950 Pro and all it can do is excel (only in the 256GB capacity) at short/bursty synthetic 'scores' that mean squat in actual workloads.

Again; we've seen this all before - OCZ + SF based duracrap that limited the performance to preserve the integrity of the nand - at the expense of the promised performance that was never to be reached even once... Yawn. Been there, done that.

Now, I'm waiting for the real reason NVMe was created.

Come on Optane, where are you?

 

The difference between advertised and measured speed is due to queue depth and differences in testing methodology. The advertised speed is at QD32, while the measured speeds are at QD1,2 and 4.

Tomshardware has some nice plots of transfer speed vs QD
http://www.tomshardware.com/reviews/samsung-950-pro-ssd,4313-3.html

You can see that the 950 Pros hit close to their rated random writes once you go about QD2. Judging by the plots it seems like Anandtech's data is heavily weighted towards QD 1 (240 MBs = 60K IOPS).

You'll also notice that all of Anandtech's numbers are quite a bit lower than Tomshardware too, which indicates differences in methodology.

 

We can analyze and agree on why there is a difference between advertised and the many ways of measuring the speeds... but that is partly my point. The advertised speeds don't matter on a workstation workload at a QD higher than 4 or 8, at the most. And matter less when comparing benchmark utility 'scores' with no correlation to the real world workout a storage subsystem gets.

Real world workflows rarely mimic the synthetic (and simple for an SSD's controller to adapt to...) utilities used to test them 'like the manufacturers do'.

Just like swimming is the best exercise to become better at swimming; testing a component in the actual workload it is supposed to give superior results in should be the standard testing method. Not a simple third party utility that can be programmed into the firmware to be 'told' what scores to report. Sigh...

With SATA being a half duplex communication protocol and NVMe being full duplex, that should give us much improved (real world) performance in and of itself (more than 2x in real world usage). There is no NVMe solution (except maybe Intel's 750 series which for mobile use is not exactly a good match) that approaches that goal. Another indication of the failure of M.2 NVMe drives to deliver what they theoretically should be able to at this time.

All in all, any non-Intel SSD w/NVMe capabilities is essentially a proprietary and one off example of that standard. That alone is enough for me to disregard it for my systems today. Couple that fact with the performance deficit, increased power and non-existent (in any off the shelf notebook) cooling requirements and the ridiculously low capacities being offered today... and you can see that not even the manufacturers believe in M.2 NVMe half as much as consumers want to...

 

The Intel 750 suffers from the same bathtub curve as every other SSD out there when it comes to mixed read/write scenarios. As far as I'm concerned, it's just as big of a failure as every other NVMe SSD out there today.

 

No doubt. But as the initiator/designer of NVMe, Intel is the only one to really know it in depth and can use that to build the best NVMe hybrid (short of X-Point) drive today.

Bottom line, in real world usage, Intel's 750 is still the most desirable option, if the platform can support it properly (basically; a desktop).

 

Intel made it...but Samsung was the first to implement NVME in drives with the Samsung XS1715.

The Intel 750 for the longest had terrible boot times with NVME. Don't you remember your best bud Maximus666, wait, he's Matrix Leader here, complaining about how the 750 is the worst drive in existence because of that issue?

Yes, they patched it recently, but according to reviews the 950 Pro beats it out of the water when it comes to booting.

Also, those worrying about Heat on the 950 Pro:

http://www.pcper.com/reviews/Storag...-PCIe-SSD-Review/Thermal-Throttling-Conclusio

"When Samsung announced the 950 PRO, several of you commented on potential thermal throttling due to heat generated in such a small package during heavy use. The following image represents a worst case scenario, with the 950 PRO being sequentially written with zero airflow across it:

As you can see, you would have to write nearly 150GB at over 1.5GB/sec to get a 950 PRO to warm up enough to throttle, and when it does, the throttling is very minor, dropping to only 1.2GB/sec intermittently. The slightest airflow prevents this from happening at all, and even if there was zero airflow, the chances of maxing a 950 PRO out on writes for that long of a burst is extremely unlikely in even the most demanding consumer usage scenario."

Basically, give it adequate air flow, and no throttling.

 

thats what i was expecting, just by looking at the raw power consumption numbers of the 950 pro compared with the sm951 glad to see it in hard data as well!

 

Jeez these PCIe SSDs pretty much double the power consumption of standard SATA/mSATA SSDs.

 

well what did u expect with such high throughput crammed into such small space as an M.2 stick? -

 

First means nothing without understanding how to implement the 'thing' properly. And, I have as much faith in Samsung of implementing NVMe properly (vs. Intel) as I do of RAPID actually increasing my productivity.

Boot times? Yawn. That is a 'score' that should die already (as long as it isn't in the minutes range, of course...).

Using/testing this M.2 drive on a desktop M/B is not what I would call 'real world' - especially if expecting to get these kind of results in a mobile platform.


See:
http://www.legitreviews.com/samsung-ssd-950-pro-512gb-nvme-pcie-ssd-review_174096/3

Doesn't look like the heat/throttling/power issue is 'fixed' to me.


And real world compared to previous 'tech' (HDD) real world is an eye-opener. 1000x less latency sounds astounding... orders of magnitude faster... amazing. But nothing at the bleeding edge of SSD's (SATA + NVMe) jumps out as much as simply moving from HDD to anything current. But that doesn't mean we get those promised astounding and amazing results though...

See:
http://www.pcper.com/reviews/Storag...-PCIe-SSD-Review/Sequential-Performance-HDTac


Look at the PCPer File Copy Test - File Creation graph and the PCPer File Copy Test - File Copy graph.

While it takes a HDD 6.4x longer to create 10x 1GB files than the 'fastest M.2 SSD available', it is still a fairly acceptable absolute time (51.2 Seconds). Similarly, it takes a spinning disk (yeah; a VRaptor) 11x longer to copy 100, 100MB files vs. the 950 M.2.

At the other end of the spectrum, it takes a spinning disk 10.2 seconds to create 1000, 1MB files. Our 'star' M.2 SSD does it in almost half the time; 5.5 seconds. Copying files, the VRaptor is at 18.5 seconds vs. 3.4 seconds for the 950 M.2 - again, for 1000, 1MB files.

I would need to create 1 million 1MB files to save about 4 hours and 12 minutes with the old tech, VRaptor.

In my world, these marketing BS numbers (from all manufacturers) do not mean much. The above simple math does.

Actual storage subsystem use doing common tasks. Not theoretical musings and statistical mangling of facts to try to move products.

And... keep in mind that although I compared the VRaptor to the 950 M.2, the 950 was not always the 'winner' in these real world tests either.

(And granted, of course, the VRaptor isn't capable of being used on a mobile system either).


Would I choose a HDD based platform today? No, even with all the above and $$$$ taken into consideration. But that is because I can and do create a lot of data - especially when I'm not on-site and have access to my multiple workstations - and get paid to do it.

The point here though is that people need to see that when in absolute numbers mere seconds separate the previous gen to the current/bleeding edge generations... throttling is not something to blithely ignore.

It kills any performance advantage you thought you may have had.

And, in a mobile setup (and even certain desktop setups too; think 'NUC' or bigger), it will most likely affect not only your storage subsystem, but also your gpu and your cpu performance too. Giving a 'too hot' component multiple fail strikes against it.

Just like performance cars that are affected by heat soak even with the best cooling systems possible; computer system components that ignore the cumulative effects of excess heat will bring the overall platform down to less than last gen spec's.

Let alone less than competitors better executed products that may not scream 'look at me!!!' as much as the subpar products with marketing $$$$$$$$$$ behind them to make them move as fast as possible and keep the consumers in the dark about any real performance improvements - promised and delivered (or not).

As almost all the online rags claim; the 950 M.2 is the best SSD they've tested so far. But that means squat. Especially in absolute terms.

I don't care about an 8 second car that can hit those speeds on the last Tuesday of April after a light rain fall two hours before dawn... I want a real, obvious and sustained improvements over what we have available today (>~1GB SATAIII SSD's...).

That is what I don't see yet from anything M.2... but on the horizon Optane is inching closer and closer.

And if the improvements it offers are proven true, M.2 and everything below it will be forgotten soon enough.

 

I suspect that in order to reach such high levels of peak performance, these PCIe SSDs have faster and more complex (read: power hungry) controllers than SATA SSDs do.

 

Matrix Leader's Intel 750 took like over a minute to boot from what i remember.

I feel like you just get a hard-on from hatin on Samsung. You'll disagree with any of these reviews as long as they are positive towards them.

How is the heating issue not taken care of? The thermal throttling is there to prevent damage to the SSD. Are you saying you shouldn't have adequate airflow in a system? Cuz even with minor airflow, the thermal throttling doesn't kick in at those heavy writes. Also, no normal user, even pc enthusiast would use the drive that intensly that the drive would heat up that quick.

The fact of the matter is that the 950 Pro will be the best consumer level SSD out there once it comes out, head to head with the Intel 750, and no amount of hate wishing from you will change that.

 

Doesn't matter how long the Intel 750 initially took to boot up at launch; the issue has been addressed since then.

Yeah; the reviews for the 950 M.2 are all positive, even with hard data that shows the exact opposite of 'performance'. That is the issue - those kind of reviews I can do without - and so should everyone else; as I tried to explain in my post above (which you conveniently ignore, btw).

Stop talking in circles. In a mobile setup, there is no/very little airflow where the storage subsystem components go. Using an 'state of the art' SSD for 62 to 100 seconds at a time is not what I call progress. Especially when it also heats up the rest of the system to 95C...

The useless synthetic 'scores' already point to the 950 M.2 as the best SSD so far - YAWN - but as I've already pointed out; they have very little bearing on real world performance from drives we are already using.

Couple that with the idiotic throttling/overheating issues (again; reminds me of SF/OCZ garbage with DuraCrap technology that prevented the user from using the drive at advertised speeds), low capacities and idiosyncrasies such as this***... and you can see that my hatred is not for Samsung per se, but rather the low level brainwashing marketing BS that even the 'review' sites spew forth as 'information' lately.

You go on loving Samsung unconditionally - I'd rather speak the truth as I see it.

And once again; keep your personal remarks about me, to yourself.



***

 

sooo should i get the 950 pro in my next build or what? what about putting two in raid 0?

 

If it's a desktop, go ahead.

 

raid, never. just doesnt bring it in terms of small file size performance. its just for bragging rights in benchmarks

wait for intel optane with xpoint memory, THAT will finally be a serious performance boost over existing top tier sata ssds

thats also what im gonna do, btw

Sent from my Nexus 5 using Tapatalk

 

yeah but i'm saying if i had to get a new ssd, would it be a good option? lots of multi tasking and video editing here. right now i'm using a basic low end beginner sandisk ssd. also what helps games and applications load up fast? those are clearly not just sequential

 

See post#16.

Unless you can give significant cooling to your notebook's M.2 slot, the performance is only theoretical.

That is why I said to go ahead in a desktop platform.

 

app loading, as well as gaming and os boot is all in the small file size performance (most of all 4 KB area) and not the sequential performance. sadly, sequential is the no. 1 selling point of ssd manufacturers, that only helps when u transfer HUGE amounts of LARGE files such as iso files, movies, archives, etc back and forth between superfast huge ssds, cuz that would be the only scenario being able to make use of such bandwidths. go with large external hdds and ure already out of luck cuz they cant keep up with ssd sequential speeds. so then ure stuck with basically getting a mortgage for your house for getting ssds sporting several terabytes of flash to make proper use to a large storage combined with super high sequential speeds.

another thing is the queue length of small file size performance. IF and WHEN manufacturers throw around iops (input output operations per second) numbers, theyre always in the upper echelons of queue lengths. to be more precise: server and enterprise workloads of the likes at google, ibm, facebook, etc. need those kinda queue lengths, cuz they actually HAVE the need of such humongous data throughput on a constant basis. consumer workloads are more in the area of like 1-2 command queues in the 4KB fike size area, whereas servers and enterprise workloads profit from 8+ upwards, to like 32.

therein lies the problem: the huge iops numbers manufacturers keep throwing around are based on...can u guess it? exactly! high command queue numbers! once u go down to consumer focused QDs of 1-2, u suddenly only see differences of like 10-20MB/s between sata and m.2 pcie ssds instead of 1GB+/s differences in the totally unimportant sequential areas or high QD iops numbers

again, THIS is going to be the big breakthrough of xpoint memory based flash! with 4kb small file size performance increasing 10 or even 100-fold u can expect insane ACTUAL REAL LIFE performance boosts in application/game loading and boot times and not just silly artificial benchmark numbers that are just good for bragging rights and emptying ur wallet ...

tl;dr: keep your budget sandisk ssd and wait for intel optane / xpoint ssds, cuz u wont notice any real life performance difference when swapping it out for a current m.2 pcie ssd, unless u transfer large files in the scale of houndreds of gigabytes on an every day basis back and forth between different disks.

Sent from my Nexus 5 using Tapatalk

 

The test in the video below shows the 950 Pro reaching 98.7c then throttling performance, peak transfer speed drops from 700+MB/sec to 400+MB/sec (33%), while the SM951 keeps going at full speed and reaches 110c while continuing to transfer at 700+MB/sec.

The SM951 finishes the 205+GB copy in much less time,
4:35 vs 6:08 = 1 minute 33 seconds (33%) faster.

The 950 Pro throttles itself at 98c and it does this by throttling from full speed to 66% of full speed.

Samsung 950 Pro vs Samsung SM951 Thermal Comparison


That was an open frame test on a motherboard in a large computer case. In the more confined space of a laptop, the throttling upon reaching 98c could be much sooner, and those confined spaces hold heat - there is no airflow - so a constant state of hitting that 98c throttle could keep happening - throttling even shorter transfers.

The throttling started at around 50GB, but in a less ventilated spot the throttling temp of 98c could be reached sooner. And, once reached without ventilation the heat will be retained longer.

Here is the article the video came from:

Product Review: Samsung 950 Pro 512GB M.2 Drive
Written on October 27, 2015 by Matt Bach
https://www.pugetsystems.com/labs/a...-2-Drive-725/#FileCopyTemperature&Performance

And, apparently Samsung has publicly stated they throttle the 950 Pro:

"On the other hand, Samsung publicly advertises the throttling capability of this drive so if you are willing to see some performance throttling it should be completely safe to use it without a heatsink if you wish."

I also found a great graph that shows the performance drop/rise in a sawtooth pattern once the 950 Pro reaches 98c, I will add it here if I can find it again.

Doesn't 98c seem like a *really high* temperature to throttle? I would think more like 68c would be safer for long term use.

The SM951 temp drops after completing the transfer, the 950 Pro isn't done with the xfer yet so the temp stays up at 98c, that isn't 950 Pro endurance, that is 950 Pro thermal throttled slowness...



The graph does bring up another point, the idle temps. It's hard to tell in the short start of the video, but it looked like the temps started at around 70c, which the graph also shows at the tail end for the SM951 - it has had time to cool while waiting for the 950 Pro to finish the xfer.

So, watch for your 950 Pro temps at idle and during long copy's, duplicates, and builds, and please let us know what you are seeing at the extreme's and how it is real world transfer speeds during normal usage

 

This is scary, do you think the fact that my Clevo will have thermal pads on the 950 PROs will eliminate this high temp issue/throttling? Do you recommend me to tell HIDevolution to swap the 950 PROs to SM951 instead?

@tilleroftheearth What do you think about this bro?

 

From what we've seen, it doesn't matter which M.2 SSD is installed in a non-custom notebook chassis. Throttle, they will.

Unless the thermal pads are copper and they also solidly attach to the chassis (which should be made of a quality metal...), I don't see how any M.2 SSD won't throttle in an off the shelf notebook chassis (even a chassis like a Clevo).

See posts #16 and #21, if you haven't already.

 

It's really a choice of performance or temperature.

Both are about the same performance un-throttled, the 950 looks better/worse in some areas, so that's a toss up.

A 12c throttle-point difference - the SM951 throttles at 110 or near to it from the graphs, and the 950 Pro at 98c.

If you don't think you will hit the 98c throttle point in your application, then the 950 Pro is newer, not sure what else is a differentiator - price? - warranty? - availability?

If you don't care about the heat, and want the sustained speed for large transfers, or if the application keeps the temps high all the time and you need full speed when you access, then the SM951 would be the choice.

This was bound to happen, as the form factor is small, there seems to be no attention to cooling the bare chips, and no air flow directed over any of the M.2 slots I have seen.

All of this is due to not enough attention being paid to the problem. There are solutions, but nothing has been implemented yet.

If you need the space, buy now, if you can wait, wait for early 2016 as there is another wave of new V-nand coming, higher density as well - 1TB in the M.2 2280 size.

Maybe the new M.2's will come with little tiny fans on little tiny coolers

 

I posted this question on OC Forums and one d00d said that unless I am copying multiple GBs of data back and forth to the SSD, I shouldn't have a problem

my workflow is surfing the web 70% of the time, watching movies 20% of the time, and maybe play a game 10% of the time.


What do you think?

 

Reply from Sean Webster (Editor) regarding this:

 

When someone gets the new 950 Pro, please run a CrystalDiskMark 3.0.4 1000MB test against it.

Monitor the temps with something that logs temps against throughput to get a good graph, or if you aren't set up for that you can use CrystalDiskInfo to see what the beginning idle temp is, peak temp during run, and temp right as the test finishes.

That CrystalDiskMark test ran M.2 SATA III drives to their peak temps without waiting for a long transfer, I imagine the can be true for the M.2 PCIE x4 NVMe Pro 950 / SM951.

That would be the first test I would run

The next test would be to do an image backup of the boot drive to an image file on an external drive or another internal partition.

Copying that image across from one partition to another on the same RAID0 volume is also another thermal trigger operation. Reading / Writing to the same RAID0 volume should heat up all the SSD's in the RAID0 - and if those SSD's are close together they will heat each other up too, and heat up the other components near them - like the CPU/GPU - reducing thermal overhead - reducing OC potential.

@Matrix Leader - you could let the vendor know of your concerns, and if you end up having a problem could you swap after the sale, say in 15-30 days, that way you are covered either way

 

Thanks!, that's the review I was looking for, I love that Sawtooth graph pattern, the 950 Pro throttling, cooling, speeding up, heating up, and then throttling, and cooling, etc.



In that test the throttling started at 110GB, but in the other test the throttling started at 50GB. The reason that is is because the transfer size has nothing to do with the throttling, it's the temperature, 98c.

If your environment is such that you are heating up the surrounding area of the M.2, like the slot is near the CPU / GPU / cooling elements, then heavy system loads - gaming - will bring up the temperature floor to the point where you may throttle after short transfers.

We really won't know until people get them into the new laptops and start using them, and it will be different in every M.2 layout.

 

You won't like what I think.

But since you asked, any computer component that operates at a higher temperature than the component it is replacing is never what I consider 'good practice'. In this case, an SSD is using more power, generating more heat (that in most systems can't be expelled efficiently) and operating at a higher average temp than the proverbial HDD it is replacing. Not only can it not give the performance you are paying for - but it will probably affect other aspects of the system negatively too. Heat is a sign of bad design since 1970...

With your stated workflow, you may actually never hit those throttled storage subsystem lows... but at that point, all I can ask is why? Why buy the supposed state of the art when all you'll have is a few weeks of bragging rights?

What I would do is buy with the explicit understanding that the system as spec'd and shipped is in testing mode until it passes your approval. And; your approval should be based on/compared with an SATA3 SSD of 1TB or higher vs. the M.2 offering.

If you think at that point that the extra $$$ were worth spending (with possibly less capacity?), then at least you'll have done your due diligence.

But if you're just believing Samsung and the general online buzz that M.2 is 'here'. I believe you'll be sorely disappointed (and if not immediately, in the near future...).

 

But what about those thermal pads that Clevo supply to be placed on the M.2 SSDs? How effective are they and to what extent would they bring the temps down? by how many degrees?

 

The temps from those thermal pads most be cooled from some airflow or the heat will stand still in those pads. The same also apply with those extra copper heatsink you pay for to cool down cpu/gpu. Some tested those extra copper heatsink without a notebook cooler. The temperature will go slower up but after awhile the temperature is the same as without the copper heatsink. Buy a notebook cooler if you don't already have one. Cooler Master U3 and mod with some 12v fans + 12v adapter.

 

what about if i'm doing video editing and reading and writing on the same hard drive? those are obivously bigger video files....

 

Yeah, and.... ???

 

Hard Drive or SSD?

 

i meant storage drive. so ssd. i was wondering if upgrading to a pcie ssd would make a difference.

 

in benchmarks, sure, in real world usage, unless you are doing heavy large data copying from one drive to another, then not one bit. The 4K speeds are about a 5% different between the 2 you can't even tell. 4K Read/Write is where you would perceive most of the performance as a normal or even power user. The rest of the benchmarks (sequential especially) are for marketing or bragging rights. They do come into play under a certain work flow, but that is a rare one for the average person.

 

This is something to keep in mind not just for PCIe drives, but all SSDs. A lot of fanfare is made about Extreme Pros or SM951s and whatnot, but for the average user, the difference will be small to nonexistent.

The average person will not notice a difference in their casual/gaming daily use, whether they be using a SanDisk Extreme Pro or a budget ADATA SP550. But they WILL notice the (much larger) hole in their bank account.

 

Word, I cannot notice a difference between my wife's Crucial MX100 512GB vs the Samsung 850 PRO / SanDisk Extreme PRO I had in my Alienware 18. I now realize, an SSD is an SSD, once you cross that barrier of the slow HDDs, they're all blazing fast. But hey, sometimes we want bragging rights and benchmarks

I hate it how companies put on the box of the SSD = 550 MB Read / 500 MB Write speeds!! yeah right!!! you wish! LOL

 

From what I can tell, at least in my P770DM, those thermal pads will touch nothing, therefore heat will just stay in the pad. If they DO touch something, it might make contact with the bottom panel, which is made of plastic, and won't be very good at dissipating heat either.

Well, some people with a very specific workflow probably would notice the difference. But the average user won't. I've had a whole bunch of different SSDs from a whole bunch of different companies, and the only time I could tell a difference is when I was running benchmarks. Other than that, I have no idea which laptop has which drives, because performance is the same.

 

Oh darn, so they're just a gimmick, take the heat of the SSD and keep it in the thermal pad, still bad I think we need fans for those monsters.

 

For all its faults, the ASUS ROG G751 does it (almost) right. It uses thermal pads as a TIM between the drives and the (all-metal) bracket that connects it to the laptop. So the PCIe drives you put in that particular machine will actually be cooled to a decent degree.

 

yes, i was about to say that in some cases it does make sense using those heat pads, especially with metal chassis or plastic ones that have copper plate inlays on the inside that would at least dissipate the heat over a way larger area and thus help with controlling the temps.

Sent from my Nexus 5 using Tapatalk

 

Well, you can reach those speeds - if you copy a single multi-gigabyte file from one SSD to another.

That rarely happens in real life, though. Especially in notebooks where it's standard to only have a single drive.

 

I use a little micro-fan connected with thermal tape on the inside of my drive bay. Blows directly onto the drive's controller. Stops it from thermal throttling. My drive runs like holy he..! I have a raid 0 on my lap also - can't touch the nvme drive. Without the fan the drive is useless for me. About the part of the review you are showing in the initial post: If you are expecting the drive to excel in every department, no it will not. But for latency, high queue depths, sequentials and partly random read/write, then yes it IS great. I think the ssd market should really start taking on new grade classifications. Not only consumer or enterprise. If YOU in this case are expecting an enterprise pci-e drive, then clearly neither the SM951 nor the 950 Pro ARE. In case you are thinking they are two separate drives, I can tell you I am using the official Samsung 950 Pro driver for my Sm951 nvme.

 

Is this a desktop or notebook build? If it is a notebook, would you mind some pics?

 

This is on a notebook. Clevo p751zm. Just head over to the official "Batman" forum and find my posts there. Several pictures.

Sent from my HTC One M8s using Tapatalk

 

How about some links so we don't need to hunt them down?

And, do the pictures show the micro-fans? What make / model of micro-fan are they, and where did you source them? Links?

Thanks

 

http://forum.notebookreview.com/threads/official-clevo-p75xzm-batman-sager-np9752-owner´s-lounge-welcome-to-the-batcave.767105/page-831#post-10092396
The cable outside of the chassis is used mostly because I didn't want to void my warranty. If you want to have the cable on the inside of the chassis, you can turn one of the internal usb3 ports 180 degrees so that it faces inwards(If you can solder).
I tried doing another all internal solution as you can see, but for some reason couldn't fire up the fan when connecting to internal port with pico connectors. This way I can connect/disconnect fast, and I can keep my schedule, which is to keep my computers only 1 year before selling. Depending on how savvy you are with electronic circuits one of these solutions should be possible.
I bought my fans on ebay Bought 2 in case one craps out....