>Browsing Amazon for a new M.2 SSD
>Finally found one I like
>1400 MB/s read and 1000 MB/s write speed, fucking great
>Uses 4x PCI-e 2.0 lanes
>Check my z97 motherboard specifications
>z97 chipset provides 8x PCI-e 2.0 lanes in addition to the 16x PCI-e 3.0 lanes from the CPU
>The M.2 socket is fucking limited to 2x PCI-e lanes
>Capped at 1 GB/s
Why the fuck z97 mobos manufacturers impose such a retarded limitation in the M.2 socket? Realistically, how hard it could have been for the M.2 socket to use x4 PCI-e 2.0 lanes instead of x2 or make it share lanes with the 16x PCI-e 3.0 slot? (and at the same time making it compatible with new 4x PCI-e 3.0 NVMe M.2s, two birds one stone)
There is literally a 1% performance hit when running a GPU on 8x PCI-e 3.0 lanes instead of 16x.
I just can't find any reasons to justify such a stupid decision, specially when the z97 has plenty of lanes that could have been used by the M.2 socket.
because intel wants to jew you into a memelake cpu
>>52970974
There are only so many PCIe lanes, maybe all others are populated?
>>52970974
why would you care
it's not the only limitation anyway
we don't even know your OS or the shitty filesystem it uses
>>52970984
Where in the OP do you see anything about a fucking GPU?
>>52970974
You are fucking retarded OP. SSD's get their speed from the IOPS, not the transfer speed. Anything over 400MB/s is decent.
>>52970974
Just get a PCI-E adapter card and put it in one of your free slots.
>>52971205
>needing disk speed above gigabit ethernet speeds
why even bother with SSD's
>>52970996
This so much, damn.
>>52971095
I searched the manual and online resources to find out if any other internal component, wifi or bluetooth modules use the 6x remaining pci-e 2.0 lanes. Nothing. By the way things are, it seems those 2.0 lanes just sit in the mobo doing jack shit.
>>52971137
Not that guy but when you've got an M.2 SSD taking up pcie space you've got less lanes remaining for gpus
>>52971128
Windows 7
>>52971137
How extra retarded you can get illiterate snowflake?
>or make it share lanes with the 16x PCI-e 3.0 slot?
>There is literally a 1% performance hit when running a GPU on 8x PCI-e 3.0 lanes instead of 16x.
>>52971205
We are talking about the maximum speed the bandwidth cap allows, holy kek /g/ really is full of idiots like you pretending to know their shit
>>52971219
Good idea, sadly the m.2 slot is in the rear of the mobo
>>52971137
See: >>52971257
>>52971221
SSD are getting much much faster than the originally famous 850 pro anon. The 950 pro has a speed of 2,500 MB/s read and 1,500 MB/swrite
>>52971490
multiple TeraBytes/s is insane
>>52970974
It because most Z97 boards use the other PCIe 2.0 lanes for other peripherals (NIC, SATA controller etc).
It doesn't matter anyway because M.2 Express is bloody overkill for non-professional-tier workloads.
SATA-600 SSD devices are more than enough for majority of the users out there.
Protip: NVe is the spiritual successor to SCSI.
>>52971550
Originally I thought this and searched in vain the manual and e-resources to find out what peripherals use the remaining x6 Pci-e 2.0 .
Could you share a link detailing what you just posted? If this truly is the case, the decision seems less retarded although the ideal choice was making the m.2 socket share bandwidth with the 16x Pci-e 3.0 slot.
The only M.2 SSD that uses 2x Pci-e 2.0 is the m6e from a pleb tier brand "plextor"
>>52970974
What kind of retard would actually buy a PCI-e SSD?
>>52971219
While using an adapter you are still restricted to Pcie2.0 x2 speeds, so there is no point in that.
>>52971787
Not him.
A PCIe SSD and a M.2 SSD that uses PCIe lanes are two different things.
Also if you have the free socket and the money, why not? Unlike Sata III, PCIe SSDs (and M.2 PCIe SSDs) are not limited to 600MB/s and the one in OP's picture has triple the speed of a 2.5" Sata III SSD.
>>52971787
Prosumers that do real-world work that involves moving GiBs of data on a frequent basis.
PCIe SSD are bloody awesome for that, otherwise they are overkill for anything else.
>>52971861
Because you'd get twice as much standard SSD storage for the same price
>>52971881
Pretty sure there are only a few specifc workloads that need storage that fast though
>>52971788
>>52971861
Pic. related is a PCIe SSD compared to two M.2 SSDs that uses PCIe lanes.
Forgot to mention M.2 SSDs that use Sata III also exist.
>not using based x99
>>52971923
>x99
>overpriced chipset directed to goodgoy "enthusiast" sheepie
>used in overpriced mobos employing overpriced cpus with useless extra cores
>based
>>52972031
so what are you using, baby?
I have the same 'issue'.
I have a Samsung SM951 M.2, which can run upto 4x.
But my Asus Hero VII Z97 motherboard only supports up to 2x.
I still get 750MB/s read and write though, which is enough.
>>52972078
I am using your mom
she gave me a nice overclock
Give her my regards
>>52970974
>Why the fuck z97 mobos manufacturers impose such a retarded limitation in the M.2 socket?
Not all of them did, you probably cheaped out on a motherboard.
>>52970974
m.2 is shit and should stay in laptop/tablets only
workstations mobos should use u.2 ports instead, which let you use 2.5" drives that don't get thermally throttles at the earliest opportunity.
yes, m.2 -> u.2 adapter exist, but they're not very mechanically robust and look like shit.
>>52971922
> PCIe x2 connector
> B keying?
no thank, m8
>>52972116
For that mobo a wiser choice was the Kingston M.2 in OP since the 951 is OEM and has no official support from Samsung, also using a M.2 4x PCIe 3.0 SSD (4GB/s max) in a M.2 2x PCIe 2.0 socket (1GB/s max) is a massive bottleneck compared to a M.2 4x PCIe 2.0 SSD (2GB/s max) used in a 2x PCIe 2.0 socket, 1/4 vs 1/2 speed.
Still 750MB is blazing fast and I am sure it covers most of your needs.
>>52972225
>m.2 is shit
>workstations mobos should use u.2 ports instead
You got it backwards
Not a single u.2 peripheral under a recognized quality brand is being produced and M.2 is becoming a desktop standard. The fact that z170 mobos bring support for x4 Pci express gen3 only cements this.
U.2 is shit and went through a quick, painless death. RIP U.2 nobody is gonna miss you.
>>52972519
huh?
entire enterprise sector is moving from PCIe card to u.2 for higher capacities.
Intel's DC P3xxx series and high-end consumer 750 series are both offered in u.2 and not m.2 for obvious reasons.
m.2 cards can't dissipate more than a few watts of heat, and high-end SSDs can easily sustain 20-25W (at 100s of kIOPS) if adequately cooled.
>>52972116
>Samsung SM951 M.2 on a M.2 PCI-E Generation 2 x2
Sweet jesus you are a special kind of retard
The prime example of a consumer whore who buys the newest and shiniest shit promoted by marketeers without doing any kind of research beforehand
>>52970974
ITT Housefires, 2016 edition
>>52970984
>I just can't find any reasons to justify such a stupid decision
>>52972116
>>52972403
>>52972859
Most manufacturers did an horrible job labeling m.2 slots, using explanations that only generated confusion, approaching marketing deception territory. Take Asus for example: to this day I have yet to see a M.2 SSD using PCI Express x2 (Gen 2.0??) reaching speeds near 10Gb/s. Isn't 8Gb/s the speed limit of PCI Express x2 (2.0)? Where does the extra 2Gb/s come from?
>>52970974
Why are you even worried about a Z97? That's old fucking shit bro. We are on Z170. Get your shit right and stop bitching about your Gen1 tech. Also, your faggot porn will save just fine on that old shit you've got.
>>52973233
Also note that there is no single M.2 PCI Express x2 Gen 2.0 SSD that reaches the full 8Gb/s or 1GB/s. The 512 GB Plextor m6e (priciest model) only reaches up to 770 MB/s read and 625 MB/s write, also it is the only M.2 brand that uses PCI Express x2 Gen 2.0 as >>52971675 said. IMHO not worth it.
>>52973314
>Z97
>old shit
Yeah, I am sure the millions of Z97 mobos that got sold completely dropped of earth after the arrival of Z170, and that none of the z97 are considering upgrading to a M.2 SSD.
It's thanks to enabling morons like you that companies fuck us in the ass whenever they want. At least the thread creates awareness about the issue for potential buyers so they don't waste their money.
>>52973589
z97 owners*
>>52972519 Intel's 750 has u.2. It will become a standard. u2 just has a tiny board to plug into m.2 slots.
>>52973730
>u.2
>become a standard
A standard meme of course
Tell me, why do M2 SSDs dont come with heatsinks?
>>52970984
Here, this should explain things a bit more
>>52973730
reading is for idiots too dumb to win at video games.
>>52973876
Isn't it time to finally implement 10GBit Ethernet?
I just recently started wondering why I'm not using a NAS yet and why NASs are so slow since fucking 15 years. Because that's like the timespan we've got Giggerbit nearly everywhere.
Why am I the only one wanting a NAS that's equally fast as S-ATA?
Or is S-ATA already able to be maxed out via Gigabit Ethernet?
>>52970974
Why? Because these new pcie x4 cards are quite new. There was not anything like it when z97 came along. The blazing fast m.2 drives came out right at the end of z97. x99 has a few x4 m.2 slots, however.
>>52973969
I always thought NAS was slow because of the HDD's. I wouldn't mind really if my NAS had a 3.1 or 3.0 usb that's alot of speed potential, but I would imagine it wouldn't be any faster than the storage medium.
>>52973876
The graph shows the Pci Express lanes from the chipset and CPU operate independently from each other, as already stated. Also it shows 8 x PCI Express 2.0 lanes coming from chipset.
Take this mobo for example: its the Asus z97 mini-itx. It comes with integrated wifi and bluetooth that nowhere in the manual or official site list as using PCI Express lanes. We already know the M.2 uses 2 x PCI Express 2.0 lanes, so my question is: What peripherals or internals use the remaining 6 x?
>>52974001
I see.. so the faster tech is pretty new. Makes a lot of sense, also puts the whole issue into perspective.
>>52974198
independent* from each other
fucking typos
>>52970974
>he didn't do research
I specifically bought a motherboard with m.2 x4 support.
Why didn't you?
>>52971881
>Prosumers
nice buzzword
>>52974284
Nice speeds
I wasn't going for a full ATX mobo at the time anon. The only mobo that supported m.2 x4 in my form factor was the ASUS MAXIMUS VII IMPACT and it trades too many features for the m.2 support.
>>52974198
sata-e takes some lanes, it has to share
Intel only officially supported their storage technology on a single x2 at a time.
chipset USB 3.0 ports take some lanes
>>52974376
I have that board, what features does it trade exactly? Kinda pissed they traded the eSATA port from the VI into a PS/2, dumb as shit.
>>52974376
This is one of the things that prevented me from buying a mITX build when I built my current machine.
It was disappointing, but there's no way I was going to invest in skyshit.
I have given serious thought to designing my own case and having a fab shop make it. Why aren't there any "Small" ATX cases.
>>52974436
skyshit has more lanes available to keep shit like this from occurring. Its one of the many upgrades from the limited z97. But keep on memeing it up.
>>52970974
haha wow dude my computer has 40 pcie3 lanes.
>>52974399
Also I guess to add to that, I just got my m.2 drive just the other day. Turns out it's AHCI though.
Seems to me like NVME vs AHCI makes a bigger difference than x2 vs x4. Wrong? Probably going to keep this until they fix the heat issue though.
>>52974564
same here. and I dont even plan to buy a nvme drive.
Waiting for 3dxpoint it's gonna btfo nvme.
>>52974624
doubt your old board supports nvme anyways. It isn't plug and play and needs to be hard coded into the bios.
>>52974675
The Impact VII? Yeah it does, I already checked. Is Z97 really considered old now?
>>52974624
>Seems to me like NVME vs AHCI makes a bigger difference than x2 vs x4. Wrong?
Yes and no.
In theory, and in the future, NVMe is much faster.
Especially with many small files (the stated speeds are best case scenario, they drop dramatically on random read/writes).
But software needs to be specifically written to take full advantage.
And hardly anything is yet.
>>52974765
Each drive must be added manually.
>>52974399
The z97I-Plus reaches slightly higher stable overclocks overall, also at the moment of purchase the price difference was 80 bucks and I needed the DVI port.
You have an excellent mobo anon, if I could go back in time no doubt I would buy the VII Impact.
>>52974436
Yeah, I guess with reduced size come reduced features. But I could not resist the portability the Mini-Itx cases offered
>Why aren't there any "Small" ATX cases.
Have you looked at micro-atx? Maybe there is a mobo that could fit your needs.
If only mobo manufacturers introduced mainstream support for the Mini-DTX format. What are they waiting for? It would be a dream come true.
>Mini-itx sized mobo with TWO, yes, TWO 16x PCI Express 3 for a small form factor sli-build or double PCI Express peripheral
I would pay premium price for this.
>>52974381
Good to know. Thanks for explaining
>>52974779
I don't follow.
>>52974776
So the higher IOPS don't help the random r/w? I thought NVME helps with latency/IOPS which is what makes everything feel snappy.
>>52974799
>z97I-Plus reaches slightly higher stable overclocks overall
Really? That makes no sense to me as the Impact is an RoG board advertised for it. I thought I remember looking up it has better chokes, mosfets, etc
The irony is OC is the reason I got it over the Plus and the Z87. I think it's overpriced for what it is, still not a bad board though. Only reason why I bought a m.2. drive for it is so I can justify the price difference.
>>52974808
>I thought NVME helps with latency/IOPS which is what makes everything feel snappy.
Yes, that's what I said wasn't it?
NVMe can deliver much more IOPS than AHCI.
>>52974869
Ok, just thew me off when you said they drop, like maybe the difference isn't so great because of the limitation on the NAND itself or something
>>52971332
Hang yourself Maylasia faggot.
>>52974920
AHCI drives are bottle-necked by the host-controller interface.
That's why all AHCI drives kinda perform the same.
NVMe removes that bottleneck.
>falling for the M.2 meme
It's for laptops and tablets, not for desktops.
Desktop motherboards only carry them in case you want to re-use your old laptop drives.
>>52974799
>Mini-itx with two PCIe3 slots
Fuck me I would pay mad cash for a mobo like that. Why the fags at Asus, Asrock or Gigabyte haven't produced one of this yet?
We need to contact one of them or at least make a petition in change dot org maybe if it gathers enough firms something could happen
>>52975019
mITX only has one PCIe and those U.2 drives don't even fit in my case
>>52975059
AsRock made a mITX LGA2011-3 board. They're probably your best bet.
>>52974971
Care to explain or you are trolling
>>52975019
>Being this dumb.
>>52975150
trolling
>>52975019
>Not falling for the M.2 meme
>>52975150
What is there to explain?
AHCI was never designed for SSD's.
Of course it was going to cause a bottleneck.
What is the appeal of M.2 on a desktop machine?
>>52975381
Less bulk
Less cables
Access to both sata and pci lanes
>>52975381
Currently the fastest SSD on the market uses the M.2 slot (950 pro). M.2 SSDs that use PCI-e aren't bottlenecked by Sata III speed.
Also if your board only supports 4 sata ports the extra storage could be useful, otherwise its just for fags to showoff.
If you aren't doing crazy video editing and heavy productivity tasks, even a Sata III M.2 suffices most needs.
>>52972146
>mom jokes
Back to COD, kiddo
>>52971205
>pay 8 times for capacity
>only get 3x performance
this is why SSDs are a meme
>>52975644
He means instead of a normal PCIe card.
Sure, like >>52975093 pointed out on mITX board you might not have a free PCIe slot.
But other than that there is no reason to use pick M.2 over a normal PCIe card.
>>52975381
The real pros are showcased when using a mini itx board, where space is very limited.
As >>52975533 said: less bulk and cables, more space for a decent-high-enthusiast graphics card. Using M.2 in full atx where Sata III have comparable speed is questionable... at best.
TL;DR extra speed less space used
>>52972031
>I am retarded: the post
>>52971787
because SATA SSDs are slow as fuck.
>>52971881
>prosumer
"I make dank montage parodies and vape promos"
>>52975960
500/500 on the read/write is slow? You wouldn't even notice much difference going up from there
>>52971205
>>52971787
>>52972116
>>52972403
>The human eye can't see beyond 500MB/s
>>52975885
by that logic, high-end GPUs, CPUs, RAM, etc. would be memes, too
>>52970974
You just have shitty mobo. Its not the chipset problem.
>>52976061
Kek
>>52976061
>>52976148
It's just stupid to not get 1TB of already fast storage, vs 500gbs of slightly faster storage, unless you're of the 1% of the 1% that needs storage that fast for your workload
>>52973233
PCIe 1.0 and 2.0 use 8b/10b encoding, so you lose 20% to overhead. Thus the raw bandwidth is 5GT/s*2 lanes = 10Gb/s, which is 8Gb/s usable.
>>52976212
I don't understand why anyone would want a 1TB SSD. I have literally only 130GB used on my SSD. What are you going to do with 1TB? Watch chinese cartoons? Kek HDDs are pretty good at sequential reading. I'm getting 150MB/s from my 3TB HDD.
>>52976321
vidya ofc
>>52976447
pretty much
also power computing or basement dweller gaming if your parents rich and you have lots of neetbucks to spare
>>52976321
My reaction image folder takes 9 minutes to fully load in a HDD.
And is 290GB big.
I want to stuff load quickly as possible.
I could organize it but why do it when i can just pay for faster loading times?
do what i did and get the Asrock z97 extreme6/9
dual m.2 ultra which are uncapped.
but just a FYI. the performance benefin compared to sata6 is no existant in day to day usage, even game load time isnt increased.
i traded my intel 750 pcie for a 850pro sata6 disk, which btw was faster in normal use.
