Bigbeard1986:

It is no surprise intel is an Israeli company...explains all of their conniving, back stabbing, money grabbing actions.

Um, wtf dude? Not only did Intel begin in Mountain View, California.....USA.....and is a USA company....but that is exceptionally, unjustifiably racist. Wow.

Emille:

If you walk into a dealership and take a model they have on the floor and a new model comes out 3 months later, whose fault is it that you didn't ask?

Your analogy would make more sense if it was more like You buy a brand new car, just released, and know in the future a new high performance tire+wheel will be released. New tire+wheel gets released You go to the dealership because you want this new tire+wheel, and find out it won't work on your vehicle, because they purposefully changed the way it attaches to your vehicle, and they are instead exclusive to a higher end model of the vehicle. It's not that the tire+wheel wouldn't work on your vehicle and be a benefit to it, but simply the company going out of its way to make sure they get more money out of you if you want it that badly. AKA, the company screwing over their customers for their bottom line and saying "Welp, you we something that 100% works with your vehicle, but too bad, screw you, you can't have it unless you give us even more money then the upgraded tire+wheel itself."

Emille:

Lol how ignorant. Even my z170 board from over 2 years ago is currently running 2 m.2 drives at full 3.0 4x speed as well as my graphics card at full 3.0 x16 speed.

So you're saying you ran your graphics card at its peak, and your m.2 all at the exact same time, and they were running at beyond what the chipset is capable of? And, you're also saying that Intel is lying about their products capabilities? Because, you know, underselling your product totally creates sales right? Mmmhmm, gotta love getting what isn't physically possible! Oh yeah! I mean you do realize that the z170 uses the DMI 3.0 link, correct? Something that is capable, at fully speed, of only 4x PCI-Express speeds? AKA, using two m.2 drives to their fullest, at the same exact time, would net you a total speed of 4x PCI-Express speed (or less depending on what other parts of your computer are also transmitting over that link) shared between the two m.2 lanes? The only way this wouldn't affect your m.2 drives would be if your drives did not need more then 2 PCI-Express lanes bandwidths each. These are facts. So if you come back with "but no, you're wrong, i have the system, i know what i'm talking about, blah blah blah blah". Well, good luck showing your intelligence while trying to refute facts about a chipset and trying to tell people what isn't possible, is possible. Lets all believe in the tooth fairy while we are at it.

Aura89:

Your analogy would make more sense if it was more like You buy a brand new car, just released, and know in the future a new high performance tire+wheel will be released. New tire+wheel gets released You go to the dealership because you want this new tire+wheel, and find out it won't work on your vehicle, because they purposefully changed the way it attaches to your vehicle, and they are instead exclusive to a higher end model of the vehicle. It's not that the tire+wheel wouldn't work on your vehicle and be a benefit to it, but simply the company going out of its way to make sure they get more money out of you if you want it that badly. AKA, the company screwing over their customers for their bottom line and saying "Welp, you we something that 100% works with your vehicle, but too bad, screw you, you can't have it unless you give us even more money then the upgraded tire+wheel itself."

That is an even worse analogy, when you buy new tyres for you car, you check to make sure the ones you are buying are compatible. It is not the fault of the car company if you assume tyres from a newer model would be immediately compatible with your car if you don't even check. I have never once bought a motherboard and assumed it would work with the next iteration. Even if it did, it would be absolutely useless to me as I have self respect and research product releases and upcoming products, anyone who buys an 8700k AND an icelake cpu deserves to be ripped off. Anyone who bought a 6700k and then wanted to buy a 7700k and complained about the chipset change...when they are the same damn cpu almost....deserves to be ripped off. Anyone who buys product releases that close together is insane and anyone who expects a motherboard to be compatible further apart, say 2 years between big product cycles, clearly has no understanding of how this works, in order to make a future cpu 100% compatible with a chipset that is out today, would have to seriously compromise the design just to please a hundredth of a percent of customers who reuse motherboards, or even desire to, rather than buying a new, better board with the latest features. Then you would find yourself in obsolete AMD chipset hell. There are comments online when the 8700k came out, people crying because they bought a 7700k 4 months earlier.....news of a 6 core coffeelake impending were rumours loooooooooonnnnggg before that and news still months prior. If people are so ignorant as to slap down $600-1000 before even considering researching what is coming next, and how soon so they can find out if a price cut is coming etc, then that is consumer ignorance, just like people who were surprised and angry at the 8 core coffeelake cpu being exclusive to the z390 chipset when news of that fact was plastered all over the internet and guru3d for us reading here, months before coffelake came out. I'm buying a new car soon, you'd better believe I'll be finding out how early next year the 2018 model will be out and if there is likely to be a discount on the current model, so I can compare the features to the savings to determine value. Anyone who doesn't do this when buying only has themselves to blame.

Aura89:

Your analogy would make more sense if it was more like You buy a brand new car, just released, and know in the future a new high performance tire+wheel will be released. New tire+wheel gets released You go to the dealership because you want this new tire+wheel, and find out it won't work on your vehicle, because they purposefully changed the way it attaches to your vehicle, and they are instead exclusive to a higher end model of the vehicle. It's not that the tire+wheel wouldn't work on your vehicle and be a benefit to it, but simply the company going out of its way to make sure they get more money out of you if you want it that badly. AKA, the company screwing over their customers for their bottom line and saying "Welp, you we something that 100% works with your vehicle, but too bad, screw you, you can't have it unless you give us even more money then the upgraded tire+wheel itself." So you're saying you ran your graphics card at its peak, and your m.2 all at the exact same time, and they were running at beyond what the chipset is capable of? And, you're also saying that Intel is lying about their products capabilities? Because, you know, underselling your product totally creates sales right? Mmmhmm, gotta love getting what isn't physically possible! Oh yeah! I mean you do realize that the z170 uses the DMI 3.0 link, correct? Something that is capable, at fully speed, of only 4x PCI-Express speeds? AKA, using two m.2 drives to their fullest, at the same exact time, would net you a total speed of 4x PCI-Express speed (or less depending on what other parts of your computer are also transmitting over that link) shared between the two m.2 lanes? The only way this wouldn't affect your m.2 drives would be if your drives did not need more then 2 PCI-Express lanes bandwidths each. These are facts. So if you come back with "but no, you're wrong, i have the system, i know what i'm talking about, blah blah blah blah". Well, good luck showing your intelligence while trying to refute facts about a chipset and trying to tell people what isn't possible, is possible. Lets all believe in the tooth fairy while we are at it.

They are all running at full speed, both drives are running at pci 3.0 x4 speed, and the gpu at 3.0 x16 speed. It's all right there in the screenshots and even the motherboard manual says that both drives run at x4 speed. But hey, what goes Gigabyte know? Am I right? http://i65.tinypic.com/2w4dy5e.jpg As you can see below, only the physical pci x4 slot is disabled when using an m.2 drives, and the m.2 slots do not split bandwidth ( as you can see here and in the picture above ) and as you can see in this picture there is no disclaimer on the x16 slot, it does not suffer reduced lanes as a result of 2x m.2 drives and x4 speed either. 32gb is full speed pci 3.0 x4, and it has 32gb per drive. http://i64.tinypic.com/p3rsp.jpg In the same way that plugging a second gpu in would limit the pci lanes to x8/x8 ( reducing it from x16, and gpu-z would report the pci speed correctly as x8 speed in that case ) if the motherboard was only capable of sharing pci lanes both m.2 drives would be LIMITED ( do you know what that means? ) they would not all be reporting as 3.0 x4 speed and they would not be benching full 3.0 x4 speed results which I have ALREADY posted in addition to proof that the GPU is running at 3.0 x16 also and crystal info/mark would report them as x2 speed, samsung magician also reports both drives as x4 as well but for some reason that program is miniscule at 4k so I can't really show it. They would not be circumstantially running at x4 speed. They would be limited. I guess Samsung, Gigabyte, GPU-Z and Crystal Mark/Crystal Disk are all wrong then, 2 argumentative forum users online clearly know better. You and the guy who claimed they would run at x1 speed, both without any evidence, just an AMD chip on their shoulder. Stop lying already.

Emille:

As you can see below, only the physical pci x4 slot is disabled when using an m.2 drives, and the m.2 slots do not split bandwidth ( as you can see here and in the picture above ) and as you can see in this picture there is no disclaimer on the x16 slot, it does not suffer reduced lanes as a result of 2x m.2 drives and x4 speed either. 32gb is full speed pci 3.0 x4, and it has 32gb per drive.

I just skimmed all these posts but you seem to be very argumentative about your setup. I will make it clear. Your setup is limited to 16 CPU pcie lanes and 4x Gen 3 via dmi link. You are not getting full speed to all host devices at the same time. Period. This is a technical limitation of your setup. Due to the fact under no load, the GPU will only use a limited amount of lanes. That means the M.2 drives are able to use the max lanes supported. Try a game with 100% load, then run both M.2 benchmarks at the same time if you are adamant in arguing about facts. You will not get full speed on those drives. And no, i don't care about what some program polls regarding their interface link speed.

I think we know in this thread, well one doesn't, that Intel is being a colossal ass in this instance! This Z390 chipset is a farce! Don't get me wrong a chipset/feature update is good normally. I mean Z170 over Z97 was a nice strong platform update and overdue imo Even Z170 to Z270 was reasonable and allowed for backward compatibility with 6000 series CPUs! Z270 to Z370 is sketchy at best and Z370 to Z390 is a joke! Did Intel seriously not have time to implement a couple more features that Z390 will have that Z370 won't! Seems like Z390 is mean't to be what was really mean't to be released instead of Z370 but intel said hey we got nowt right now so er here have this! Like I said before I only bought the Apex for XOC so its not a massive deal to me, but I think it would be piss poor if Z370 didn't support the 8 core coming out as there is seems to be literally no reason for it not to!

Agent-A01:

I just skimmed all these posts but you seem to be very argumentative about your setup. I will make it clear. Your setup is limited to 16 CPU pcie lanes and 4x Gen 3 via dmi link. You are not getting full speed to all host devices at the same time. Period. This is a technical limitation of your setup. Due to the fact under no load, the GPU will only use a limited amount of lanes. That means the M.2 drives are able to use the max lanes supported. Try a game with 100% load, then run both M.2 benchmarks at the same time if you are adamant in arguing about facts. You will not get full speed on those drives. And no, i don't care about what some program polls regarding their interface link speed.

The cpu has 16 lanes, and the chipset has 20 lanes, jesus christ, it's not that hard.

Emille:

Lol how ignorant. Even my z170 board from over 2 years ago is currently running 2 m.2 drives at full 3.0 4x speed as well as my graphics card at full 3.0 x16 speed. A 950 pro 512gb and a 2tb 960 pro. All z370 boards with 3 m.2 drives run them at full speed. It's so funny to see people with stone age board features making claims about new boards features that are false. Tell me what software you want me to verify the pci speeds with and I'll post a screenshot with them all in one, just for you my stone age friend. Here is it, I couldn't have 2 instances of crystal disk info open so I have GPUz, crystal disk info and a crystal disk mark bencmark of the 960 pro all in one, thanks for playing champ. http://i68.tinypic.com/29pvhnm.jpg I'll be getting a third m.2 drive with my icelake build when the samsung 980 series comes out. it will also be running at full speed. This reminds me of AM3, when amd owners said that dd2 support lingering for years and a lack of ddr3 support was a blessing to customer, thanks to the 'longevity' of that ageing platform it had dated ram, pci, usb and everything else. Technology moves forwards, with or without you, feel free to sit it out but don't lie to yourself that cutting edge z370/390 boards run m.2 drives at x1 speed when it just false.

Lolololol Are some people delusional or what? What are you showing me? Results of a single 960 pro saturating PCIE 30 x4 link,where is 950 pro in your benchmark,you are claiming that you can run both 950 and 960 pro at the same time without choking them on meagerly 1x4 link ????? Son let me inform you each of them will report separately they run on PCIE x4 but they share that link with each other and everything else in your system.Meaning that your audio,sata ports,lan port,usb ... together with your 2 SSD's share single let me repeat SINGLE 1X4 PCIE 3.0 link.You can make chipset with 100 or heck even 1000 PCIE lanes if you want,but what maters is what is speed between that hypothetical 1000 lane chipset and CPU,because CPU needs to process that data not chipset,CPU is the one that is requesting data not chipset.

Emille:

The cpu has 16 lanes, and the chipset has 20 lanes, jesus christ, it's not that hard.

Clearly you are lacking the understanding and technical capabilities of a DMI link. 20x lanes that get 'squished' to a total of 4x Gen 3 lanes. DMI link is a connection between PCH -- CPU. I'll make a simple analogy for you. There is a 20 lane highway that condenses to 4x lanes to a city. The limit is 4 lanes. Yeah, there are 20 lanes for cars to be on, but they have to slow down to fit in the 4 lanes near the city. Jesus christ, it's not that hard to understand. 🙄

kruno:

Lolololol Are some people delusional or what? .

Classic sign of somebody that is 'always right'. Nobody will one up them.

kruno:

Lolololol Are some people delusional or what? What are you showing me? Results of a single 960 pro saturating PCIE 30 x4 link,where is 950 pro in your benchmark,you are claiming that you can run both 950 and 960 pro at the same time without choking them on meagerly 1x4 link ????? Son let me inform you each of them will report separately they run on PCIE x4 but they share that link with each other and everything else in your system.Meaning that your audio,sata ports,lan port,usb ... together with your 2 SSD's share single let me repeat SINGLE 1X4 PCIE 3.0 link.You can make chipset with 100 or heck even 1000 PCIE lanes if you want,but what maters is what is speed between that hypothetical 1000 lane chipset and CPU,because CPU needs to process that data not chipset,CPU is the one that is requesting data not chipset.

If you are suggesting that muliple pci devices have dynamic pci lane bandwidth limits....that's not how it works. They don't fight for bandwidth. Each device has an allocated number of lanes. They don't go from 1 to 4 depending on load. Eithet they are running a 1, 2, or 4. They don't alter bandwidth limits on the fly. Do GPUs have UP to 8/16 lanes and then drop down depending on other pco device usage.....that has literally never happened. That's not how it works. I supposed the pci devices battle each other for bandwidth right? Rather than being limited to a st number or lanes. What a stupid suggestion.

Emille:

If you are suggesting that muliple pci devices have dynamic pci lane bandwidth limits....that's not how it works.

Lol. Yes it is how it works. I'll quote from the PCIe datasheet, you know, the people that designed PCIe spec. Because we know you'll argue against and keep digging that hole of yours.

7. Dynamic LINK Bandwidth Management Dynamic LINK Re-Configuration is an optional normative functionality that supports changing the LINK Bandwidth by changing the RATE series or High-Speed Gears identically on MODULES on both sides of the LINK, and/or changing the number of Configured LANEs per SUB-LINK. A LINK Width change is permitted to be asymmetric (i.e. a MODULE is permitted to be configured 10 with different number of TX-LANEs and RX-LANEs). Software is notified of any RATE series or LINK speed and LINK width changes through the Link Bandwidth Notification Mechanism as described in PCI Express Specification. Functions that implement Link Bandwidth Notification Capability must report all bandwidth changes as Link Autonomous Bandwidth change. The Link Bandwidth Management Status bit in 15 the Link Status register must be set when Link retraining has completed following a write of 1b to the Retrain Link bit

You should have known this as you can easily see that GPU driver requests a lower power state and it electrically sends commands to host system to use less lanes with GPU z bandwidth test.

Emille:

If you are suggesting that muliple pci devices have dynamic pci lane bandwidth limits....that's not how it works. They don't fight for bandwidth. Each device has an allocated number of lanes. They don't go from 1 to 4 depending on load. Eithet they are running a 1, 2, or 4. They don't alter bandwidth limits on the fly. Do GPUs have UP to 8/16 lanes and then drop down depending on other pco device usage.....that has literally never happened. That's not how it works. I supposed the pci devices battle each other for bandwidth right? Rather than being limited to a st number or lanes. What a stupid suggestion.

Lol to not repeat the post just look under your post at post of Agent-A01 for answer 🙂:)

you guys keep buying and intel will happily deliver a chipset for every new cpu, coming soon a ZX chipset category to have both cpu AND memory oc! also XZ+ to have compatibility with non-intel nvme as boot drives!

Knowing intel, they won't do it.. From what i saw initially 8core was bound to z390 because of "different" power management that's "missing" on z370. That said, same shit as z87 vs z97 all over again.. I could plug-in broadwell on my mobo, I even flashed IME for it using z97 flasher and it worked, but for that it bugged my current haswell pcie bclk, maybe it would have worked different by broadwell, but I flashed back since I didn't have the intention to switch to 5775c.. That said z370 might work with forced IME but at own risk.

Well while i do not like the need of yet another chipset for 2 extra cores well we knew about 390 before the lunch of z370 no? Backwards compatibility itwill also be nice say your z270 board breaks down after warranty is over having the option to replace it with a z390 and keep your 7700 as long as it covers is also something i would love to see even if there are features disabled because the cpu does not support em

Emille:

I guess Samsung, Gigabyte, GPU-Z and Crystal Mark/Crystal Disk are all wrong then, 2 argumentative forum users online clearly know better. You and the guy who claimed they would run at x1 speed, both without any evidence, just an AMD chip on their shoulder. Stop lying already.

And you did not seem to have a good childhood, my mother taught me not to accuse someone of lying before I can provide a proof that he (or she) did. Please behave yourself. You still haven't posted the "parallel results" (benching every component, GFX and both m.2 SSDs) at the same time, but argue about others to provide evidence?! I googled it for you in about 1 minute to have an example of throttling: http://www.legitreviews.com/nvidia-evga-geforce-gtx-1060-video-card-review_184301/11 or picture link: http://www.legitreviews.com/wp-content/uploads/2016/07/fanstop-gtx1060.jpg See "bus interface". Compare ... @Agent-A01: *thumbs up* 386SX

I do not think he was lying to be honest , i think he genuine though that every m2 drive he has can run at the same time full speed with out problems , although the way he came out calling people cavemen and liers made him he has to defend his claims at all cost i would have jumped in but the guys did a great job already i had nothing useful to add on this matter!

Venix:

I do not think he was lying to be honest , i think he genuine though that every m2 drive he has can run at the same time full speed with out problems , although the way he came out calling people cavemen and liers made him he has to defend his claims at all cost i would have jumped in but the guys did a great job already i had nothing useful to add on this matter!

Yeah,i do not think either that he was lying,you see in his answers that he is not let's say very informed.Nobody of us knows everything that is why we have this nice little place:):):);) to ask questions and to find out a bit more about technology.