Two between the PCIe slots (there are screws to open up panels in the metal plate covering the lower part of the board), and one on the back. The one on the back is the one that disables 2 SATA ports if used, as noted in the final page of the article.
Am now wondering if this is the same "reviewer" responsible for the semi-recent Supermicro review. If so, hardly surprised. And regardless.. a 3700X? Seriously? You pick up a behemoth and test it, how? With a 3700X.. well done; again, if it's indeed the same person. Why not a 3400G? I mean it's AM4 and everything, right? (and if it is the same person, chances are that as with the previous article, i'd find even more glaring issues to name, except this time i can't even bother to read it all; stopped at '3700X')
This site does amazing hard drive reviews; very good CPU reviews; exceptional editorials; but needs some serious love in its mobo review department. It's honestly a shame, stands out compared to the rest. Respectfully, your average lurker.
"semi-recent Supermicro review" If we mean the same thing (C9Z390-PGW), then that review was February 1st. We have very different meanings of "semi-recent" then. And it was also Gavin Bonshor who did that review.
If you can't even bother to read the reviewer name in top of article and compare it to another article, and can't read past a single word in this article. What are you even doing here if can't finish simple things.
What would you suggest then smart guy? A 3800x that hits the same clock rates +/- 25 MHz? A 3900x that nobody can get ahold of and will hit the SAME clock ceiling?
"As it currently stands, the GIGABYTE X570 Aorus Xtreme includes the best power delivery of any X570 motherboard we have seen so far in our testing." _____________________________________________________________
Uhhh .... yeah. It thumped the competition, hands down. Nice job, GB.
Unless they're including a right angle adapter I think the side pointed 24-pin ATX connector is risky idea. If you're running the cable on the same side of the mobo-tray as the board it's not an issue, but most high end cases today route the cables behind the board with holes just outside. 24 pin cables are really stiff if the wires are bundled together; I had a board with a 90 power plug a about a decade ago (was either LGA1366 or S939) of years ago and the all in a single bundle cable from the PSU wouldn't make the bend; I ended up having to run a big loop into the drive bay area and ended up with more of the cable sticking out than for a conventional top down plugin. Even with that the cable was still very difficult to bend into place to insert and was putting a lot of tension onto the board trying to lift it up from the tray afterward. Individually sleeved wire designs might be OK if you've got a long length from the plug to the last point they're zip-tied together, but a standard fat bundle is asking for trouble.
A 90* plug doesn't get you anything unless you're routing the cable on the same side as the board (which isn't normal these days outside of SFF), or can make a tight 90* bend as the cable comes out of the routing hole. Unless you have a really flexible cable you're not going to be able to do that. Instead you end up having to make a 270* loop (up, then forward into where drive bays used to be, and then down and back to the board), so you still end up with a big loop.
With a big loop a 180* can be done without putting any bending stress on the plug. A 270* either needs more cable to match the same bending radius or will have to be tighter and puts more stress on the board as a result. With the one system I had this sort of setup in there wasn't enough slack in the cable to do a loop with enough slack that it wasn't trying to bend/twist the board up. When I plugged the cable in before the board was screwed down it was flexing the board up when I tried screwing it down on the edge with the socket. With the board screwed down first, it was very difficult to get the plug to the socket partly because of the tray meaning I could only grip the cable from one side and partly because the cable didn't want to be bent tightly enough to go in. On the whole it was among the most frustrating build steps I ever did and the stiffness of the cable meant that it completely failed at the notional goal of keeping the wires out of the way that's normally behind 90* edge plugs.
My initial thought was that a rigid 90* adapter that extended out to the cable management hole would avoid the problem by removing the need to tightly bend the cable to fit. Thinking a bit more, that probably wouldn't be enough because making a tight bend behind the board would be just as difficult; you'd either need a 180* piece so the cable could stay flat on the backside of the board, or a short extension with all loose wires to make it work.
IMO we need a better solution for all the connectors that exist on motherboards. For example, those USB3 connectors. How many times have I bent a pin trying to plug one in when it accidentally gets pulled out? More than I'd care to admit. I mean hell, at least put a snap/latch on it similar to what most SATA cables have. Ideally, we've had 1 cable running from the case to the motherboard, and 1 cable running from the PSU to the motherboard. Both connectors would had the little snap or latch or whatever you call it, and both would be right angled so that they can easily be hidden from view for a nice clean look.
USB-C does use a smaller and more robust connector than USB 3.0 (you can see one on this board near the diagnostic code display) that appears to take its design inspiration from PCIe.
A single cable from the PSU to the mobo would run into one size fits all problems and end up huge, ex the difference between the needs of an SFF system using a 4 pin CPU header and a high end work station/gaming board using 2x8pin CPU headers and a PCIe header (to give extra power for multiple GPUs).
What could be done easily enough would be to gut the 24 pin cable by making about half of its wires optional; even if not followed up by a new smaller plug/socket a few years later it would remove a lot of the headaches from the worst connector on the mobo. This could be done safely because the original 20 pin connector dates back to when the CPU ran on 3.3v, everything else ran on 5V, and hardly anything needed 12V; vs today when 5V is used almost exclusively for USB, 3.3V for odds and ends (eg 10 of the 25/75W a PCIe card can draw from the mobo is 3.3v not 12v), while everything else runs 12V to component specific voltage regulators.
The reason nothing's happened is more or less the same as why the mess of jumper style headers for the front panel has never been replaced by a standard block style connection. The PC industry as a whole no longer cares about desktops enough to expend the effort needed for a major new standardization round. Big OEMs can and do address the issues via proprietary components scoring spare part lockin as a bonus; while for everyone else (eg the people who make parts for customer built systems/boutique vendors) the upfront time spent and short term costs from needing to bundle legacy/modern adapters for a few years is too high to try and push something on their own. Residual trauma from the effort spent on the failed BTX standard some years back was probably an issue back when desktops were still important enough of a market segment to get serious engineering effort in standard modernization as well.
I just have to chime in and agree with changing the entire layout. Look what oem's can do when they aren't tied to the ancient atx power supply and standard pin layout. Look at the power supply used on an imac pro. That's how it should be done. These giant cables and connectors are really unnecessary.
I like the look of the board and passive X570 cooling, but am dissapointed at the lack of expansion slots. No USB 3 header? Really? Just a gen 2 that cant be used on the vast majority of cases, and even if it can it will onyl feed a singular port? No PCIe x1 slots for, say, a USB 3 header card to make up for the lack of internal headers?
Granted, this is a subjective problem, not many people use more then 1-2 slots, but for the price, I would want way mroe expansion for future upgrades. Think USB 3 headers, replacement NIC or sound cards in case of on board failure, NVMe cards for RAID arrays and better cooling, ece.
"Users can expand on this with a single USB 3.1 G2 Type-C header which provides a single port, a USB 3.1 G1 Type-A header for two additional ports, and a single USB 2.0 header which offers users two additional ports". It says right there - USB 3.1 G1 Type-A, formerly known as USB 3.0..
so true. these days, they add a 1:00 dollar RGB light and call it " GAMING " and all of a sudden it cost 3 to 10 times more, and these kids just pay it. at most these MOBO are worth 150.00 to 300.00 dollars . can't wait to see RGB toilet paper for gamers
For me, I firmly believe you have to pay for what you get, and a better “price limit” on a motherboard might be the price of the “consumer” CPU it matches: in this case, the $749 price [announced] for the Ryzen 9 3950X.
Somehow, I doubt you would be willing to spring for one of those, either...
Of course, just because YOU aren’t into this particular [*sniff*] “consumer” motherboard, many enthusiasts ARE.
And enthusiasts often drive many diverse markets, precisely because they WILL spend the money to fulfill their wish lists, unlike skinflints, who want everything for next to nothing !!!
Human from the future here: watch out for the next year.
Also, this board is $700 because it has 12 USB ports on the rear panel and 10G networking. Those things would eat up the (extremely limited) number of PCIe lanes on AM4. The fastest single core processors available today are on AM4. You could always go to Threadripper Pro or Xeon or something but you lose a *lot* of single threaded performance. This board makes a lot of sense. The other AM4 flagships: not so much.
Interesting Materiel for sure. But...a $700 motherboard???? Not for me. I would really like to see reviews of some "real people" boards. I tend to be more about using lower priced low to mid-range hardware and then using overclocking to get more out of it. For example I've got a R7-1700 that performs like an 1800x (which was a much more expensive proposition at the time). Maybe you know your audience better than I do, but I don't know anyone buying $700 motherboards for home use.
On new egg there are actually x570 boards for as low as $149 and 12 of them for under $200 as of the time I'm writing this. So, if I spend $500 less, maybe I'll just get the 4.2Ghz and not the 4.3 squeezed out of the board in this review? Yup, I'm OK with that.
Would consider X570 Aorus Elite or Pro if they had warranty. But they built a system in Russia when Gigabyte is not responsible for anything. You can only rely on retailer. But if retailer is gone, you go do that yourself with your own problems. Asus and MSI do have after-sale service.
There is a small error (typo, maybe) in Paragraph 6 [of 8] of the Conclusion:
“... the biggest [VDroop] over compensation coming at our max overclock of 4.3 GHz at 1.375 V; the maximum load voltage observed was 1.380 V which is a 0.05 V [should read 0.005 V] increase over the set value.”
So VDroop is indeed tightly contolled, but within 0.005 V, not just 0.05 V !!!
I think this is a good thing, and the conclusion should state it correctly !
GIGABYTE X570 motherboards based on the AMD X570 Chipset provide full support for 3rd Gen AMD Ryzen™ Processors. The all-new design is a testament to GIGABYTE's dedication to design quality. GIGABYTE X570 motherboards offer a rich list of features such as support for PCIe 4.0 and USB Type-C https://www.myloweslife.us/
The new, advanced power and thermal design enables users to unleash the performance on AMD Ryzen™ 3000-series Processors, making the GIGABYTE X570 motherboards perfect for users looking to build the best AMD platform gaming system. GIGABYTE X570 motherboards maximize your PC's potential with AMD StoreMI technology https://www.upsers.club/
First of all, thanks for the great review! Well-written and very informative. I just have one question: Where is the last 4x lanes from the chipset usable from?
I understand that the x16 and the x8 expansion slots and the M2A slot uses (20 total) lanes from the CPU and the last x4 expansion slot together with the M2B and M2C slots uses lanes from the chipset at 4x lanes each.
But if I understand correctly, that's only 12 of 16 lanes so far for this platform & motherboard (4x for expansion slot 3, 4x each for the 2nd and 3rd M.2 slots). Where are the other 4 lanes? It can't be the SATA ports, right? Because the chipset already supports 4x SATA ports natively.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
42 Comments
Back to Article
Kurosaki - Tuesday, September 24, 2019 - link
Yes, but where do the M2 ssd's go?! :DKWottrich - Tuesday, September 24, 2019 - link
Two between the PCIe slots (there are screws to open up panels in the metal plate covering the lower part of the board), and one on the back. The one on the back is the one that disables 2 SATA ports if used, as noted in the final page of the article.Aenra - Tuesday, September 24, 2019 - link
Am now wondering if this is the same "reviewer" responsible for the semi-recent Supermicro review.If so, hardly surprised. And regardless.. a 3700X? Seriously? You pick up a behemoth and test it, how? With a 3700X.. well done; again, if it's indeed the same person. Why not a 3400G? I mean it's AM4 and everything, right?
(and if it is the same person, chances are that as with the previous article, i'd find even more glaring issues to name, except this time i can't even bother to read it all; stopped at '3700X')
This site does amazing hard drive reviews; very good CPU reviews; exceptional editorials; but needs some serious love in its mobo review department. It's honestly a shame, stands out compared to the rest.
Respectfully, your average lurker.
shabby - Tuesday, September 24, 2019 - link
What would be the point in testing a 12/16 core? Any x570 board can run one at 4.3ghz+, the board isn't the bottleneck it's the cpu.Death666Angel - Tuesday, September 24, 2019 - link
"semi-recent Supermicro review" If we mean the same thing (C9Z390-PGW), then that review was February 1st. We have very different meanings of "semi-recent" then. And it was also Gavin Bonshor who did that review.imaheadcase - Tuesday, September 24, 2019 - link
If you can't even bother to read the reviewer name in top of article and compare it to another article, and can't read past a single word in this article. What are you even doing here if can't finish simple things.TheinsanegamerN - Tuesday, September 24, 2019 - link
What would you suggest then smart guy? A 3800x that hits the same clock rates +/- 25 MHz? A 3900x that nobody can get ahold of and will hit the SAME clock ceiling?Go back to lurking.
SSTANIC - Tuesday, September 24, 2019 - link
They have been testing it with a 3950X, but its launch and NDA (Sep 30th) have been postponed to November..Smell This - Tuesday, September 24, 2019 - link
"As it currently stands, the GIGABYTE X570 Aorus Xtreme includes the best power delivery of any X570 motherboard we have seen so far in our testing."_____________________________________________________________
Uhhh .... yeah. It thumped the competition, hands down. Nice job, GB.
DanNeely - Tuesday, September 24, 2019 - link
Unless they're including a right angle adapter I think the side pointed 24-pin ATX connector is risky idea. If you're running the cable on the same side of the mobo-tray as the board it's not an issue, but most high end cases today route the cables behind the board with holes just outside. 24 pin cables are really stiff if the wires are bundled together; I had a board with a 90 power plug a about a decade ago (was either LGA1366 or S939) of years ago and the all in a single bundle cable from the PSU wouldn't make the bend; I ended up having to run a big loop into the drive bay area and ended up with more of the cable sticking out than for a conventional top down plugin. Even with that the cable was still very difficult to bend into place to insert and was putting a lot of tension onto the board trying to lift it up from the tray afterward. Individually sleeved wire designs might be OK if you've got a long length from the plug to the last point they're zip-tied together, but a standard fat bundle is asking for trouble.Smell This - Tuesday, September 24, 2019 - link
Maybe __ but I'm not sure I get your point.A conventional top down plugin has to bend 180-degrees to route under the tray as opposed to a 90-degree 'bend' ??
DanNeely - Tuesday, September 24, 2019 - link
A 90* plug doesn't get you anything unless you're routing the cable on the same side as the board (which isn't normal these days outside of SFF), or can make a tight 90* bend as the cable comes out of the routing hole. Unless you have a really flexible cable you're not going to be able to do that. Instead you end up having to make a 270* loop (up, then forward into where drive bays used to be, and then down and back to the board), so you still end up with a big loop.With a big loop a 180* can be done without putting any bending stress on the plug. A 270* either needs more cable to match the same bending radius or will have to be tighter and puts more stress on the board as a result. With the one system I had this sort of setup in there wasn't enough slack in the cable to do a loop with enough slack that it wasn't trying to bend/twist the board up. When I plugged the cable in before the board was screwed down it was flexing the board up when I tried screwing it down on the edge with the socket. With the board screwed down first, it was very difficult to get the plug to the socket partly because of the tray meaning I could only grip the cable from one side and partly because the cable didn't want to be bent tightly enough to go in. On the whole it was among the most frustrating build steps I ever did and the stiffness of the cable meant that it completely failed at the notional goal of keeping the wires out of the way that's normally behind 90* edge plugs.
My initial thought was that a rigid 90* adapter that extended out to the cable management hole would avoid the problem by removing the need to tightly bend the cable to fit. Thinking a bit more, that probably wouldn't be enough because making a tight bend behind the board would be just as difficult; you'd either need a 180* piece so the cable could stay flat on the backside of the board, or a short extension with all loose wires to make it work.
Ratman6161 - Tuesday, September 24, 2019 - link
How about this:https://www.newegg.com/cooler-master-cma-cemb00xxb...
Ratman6161 - Tuesday, September 24, 2019 - link
https://www.google.com/search?q=ATX+24+Pin+90%C2%B...MamiyaOtaru - Wednesday, October 9, 2019 - link
cool, connect that to the side plug and come at it from behind /seek2121 - Tuesday, September 24, 2019 - link
IMO we need a better solution for all the connectors that exist on motherboards. For example, those USB3 connectors. How many times have I bent a pin trying to plug one in when it accidentally gets pulled out? More than I'd care to admit. I mean hell, at least put a snap/latch on it similar to what most SATA cables have. Ideally, we've had 1 cable running from the case to the motherboard, and 1 cable running from the PSU to the motherboard. Both connectors would had the little snap or latch or whatever you call it, and both would be right angled so that they can easily be hidden from view for a nice clean look.DanNeely - Wednesday, September 25, 2019 - link
USB-C does use a smaller and more robust connector than USB 3.0 (you can see one on this board near the diagnostic code display) that appears to take its design inspiration from PCIe.A single cable from the PSU to the mobo would run into one size fits all problems and end up huge, ex the difference between the needs of an SFF system using a 4 pin CPU header and a high end work station/gaming board using 2x8pin CPU headers and a PCIe header (to give extra power for multiple GPUs).
What could be done easily enough would be to gut the 24 pin cable by making about half of its wires optional; even if not followed up by a new smaller plug/socket a few years later it would remove a lot of the headaches from the worst connector on the mobo. This could be done safely because the original 20 pin connector dates back to when the CPU ran on 3.3v, everything else ran on 5V, and hardly anything needed 12V; vs today when 5V is used almost exclusively for USB, 3.3V for odds and ends (eg 10 of the 25/75W a PCIe card can draw from the mobo is 3.3v not 12v), while everything else runs 12V to component specific voltage regulators.
The reason nothing's happened is more or less the same as why the mess of jumper style headers for the front panel has never been replaced by a standard block style connection. The PC industry as a whole no longer cares about desktops enough to expend the effort needed for a major new standardization round. Big OEMs can and do address the issues via proprietary components scoring spare part lockin as a bonus; while for everyone else (eg the people who make parts for customer built systems/boutique vendors) the upfront time spent and short term costs from needing to bundle legacy/modern adapters for a few years is too high to try and push something on their own. Residual trauma from the effort spent on the failed BTX standard some years back was probably an issue back when desktops were still important enough of a market segment to get serious engineering effort in standard modernization as well.
Dug - Monday, October 7, 2019 - link
I just have to chime in and agree with changing the entire layout. Look what oem's can do when they aren't tied to the ancient atx power supply and standard pin layout. Look at the power supply used on an imac pro. That's how it should be done. These giant cables and connectors are really unnecessary.4everalone - Tuesday, September 24, 2019 - link
I wish MB makers would start providing SFP+ ports instead of 10GBASE-T ports. That way we at-least have the option of running fiber/copper.TheinsanegamerN - Tuesday, September 24, 2019 - link
I like the look of the board and passive X570 cooling, but am dissapointed at the lack of expansion slots. No USB 3 header? Really? Just a gen 2 that cant be used on the vast majority of cases, and even if it can it will onyl feed a singular port? No PCIe x1 slots for, say, a USB 3 header card to make up for the lack of internal headers?Granted, this is a subjective problem, not many people use more then 1-2 slots, but for the price, I would want way mroe expansion for future upgrades. Think USB 3 headers, replacement NIC or sound cards in case of on board failure, NVMe cards for RAID arrays and better cooling, ece.
SSTANIC - Tuesday, September 24, 2019 - link
"Users can expand on this with a single USB 3.1 G2 Type-C header which provides a single port, a USB 3.1 G1 Type-A header for two additional ports, and a single USB 2.0 header which offers users two additional ports". It says right there - USB 3.1 G1 Type-A, formerly known as USB 3.0..uplink - Tuesday, September 24, 2019 - link
A great motherboard indeed, that died after 5 days of installation of the system...Andy Chow - Tuesday, September 24, 2019 - link
I don't get paying $700 for a motherboard. Consumer motherboards over $300 don't make sense. And yes, this is a consumer motherboard.TheinsanegamerN - Tuesday, September 24, 2019 - link
There's a sucker born every minute.Total Meltdowner - Wednesday, September 25, 2019 - link
This.rtoledo2002 - Tuesday, October 1, 2019 - link
so true. these days, they add a 1:00 dollar RGB light and call it " GAMING " and all of a sudden it cost 3 to 10 times more, and these kids just pay it. at most these MOBO are worth 150.00 to 300.00 dollars . can't wait to see RGB toilet paper for gamersDr.Neale - Saturday, September 28, 2019 - link
For you, they don’t make sense.For me, I firmly believe you have to pay for what you get, and a better “price limit” on a motherboard might be the price of the “consumer” CPU it matches: in this case, the $749 price [announced] for the Ryzen 9 3950X.
Somehow, I doubt you would be willing to spring for one of those, either...
Of course, just because YOU aren’t into this particular [*sniff*] “consumer” motherboard, many enthusiasts ARE.
And enthusiasts often drive many diverse markets, precisely because they WILL spend the money to fulfill their wish lists, unlike skinflints, who want everything for next to nothing !!!
Not much profit in selling to skinflints !
willis936 - Sunday, August 1, 2021 - link
Human from the future here: watch out for the next year.Also, this board is $700 because it has 12 USB ports on the rear panel and 10G networking. Those things would eat up the (extremely limited) number of PCIe lanes on AM4. The fastest single core processors available today are on AM4. You could always go to Threadripper Pro or Xeon or something but you lose a *lot* of single threaded performance. This board makes a lot of sense. The other AM4 flagships: not so much.
Ratman6161 - Tuesday, September 24, 2019 - link
Interesting Materiel for sure. But...a $700 motherboard???? Not for me. I would really like to see reviews of some "real people" boards. I tend to be more about using lower priced low to mid-range hardware and then using overclocking to get more out of it. For example I've got a R7-1700 that performs like an 1800x (which was a much more expensive proposition at the time). Maybe you know your audience better than I do, but I don't know anyone buying $700 motherboards for home use.On new egg there are actually x570 boards for as low as $149 and 12 of them for under $200 as of the time I'm writing this. So, if I spend $500 less, maybe I'll just get the 4.2Ghz and not the 4.3 squeezed out of the board in this review? Yup, I'm OK with that.
Threska - Tuesday, September 24, 2019 - link
Maybe buy a fanned motherboard and water-cool the whole thing?regsEx - Tuesday, September 24, 2019 - link
Would consider X570 Aorus Elite or Pro if they had warranty. But they built a system in Russia when Gigabyte is not responsible for anything. You can only rely on retailer. But if retailer is gone, you go do that yourself with your own problems.Asus and MSI do have after-sale service.
wilsonkf - Tuesday, September 24, 2019 - link
Every auto with PBO: 181W, 70 degree C.4200Mhz: 185W, 60 degree C.
This looks interesting.
Myrandex - Tuesday, September 24, 2019 - link
Looks like an interesting board, thanks for the review!I have to say I'm not sure what you meant by unformed in "provide a subtle, but unformed look." though
azrael- - Wednesday, September 25, 2019 - link
"Fanless AM4"? That's no feat.It should be "Fanless X570". Although I still struggle with the reason this should be a feat as well...
trishanatech - Wednesday, September 25, 2019 - link
usefulDr.Neale - Saturday, September 28, 2019 - link
There is a small error (typo, maybe) in Paragraph 6 [of 8] of the Conclusion:“... the biggest [VDroop] over compensation coming at our max overclock of 4.3 GHz at 1.375 V; the maximum load voltage observed was 1.380 V which is a 0.05 V [should read 0.005 V] increase over the set value.”
So VDroop is indeed tightly contolled, but within 0.005 V, not just 0.05 V !!!
I think this is a good thing, and the conclusion should state it correctly !
Dr.Neale - Saturday, September 28, 2019 - link
Sorry for MY typo, spelling “controlled” as “contolled”. Wish I could edit it...Gastec - Saturday, September 28, 2019 - link
GIGABYTE X570 Aorus Xtreme: $700MSI MEG X570 Godlike: $700
ASUS ROG Crosshair VIII Formula: $700
Tell me that's not collusion and price-fixing.
jessica92 - Monday, September 30, 2019 - link
GIGABYTE X570 motherboards based on the AMD X570 Chipset provide full support for 3rd Gen AMD Ryzen™ Processors. The all-new design is a testament to GIGABYTE's dedication to design quality. GIGABYTE X570 motherboards offer a rich list of features such as support for PCIe 4.0 and USB Type-C https://www.myloweslife.us/Gastec - Tuesday, October 1, 2019 - link
I wish I had a ban hammer, the likes of Mjölnir, to smite you all into oblivion.Anderson6 - Thursday, October 10, 2019 - link
The new, advanced power and thermal design enables users to unleash the performance on AMD Ryzen™ 3000-series Processors, making the GIGABYTE X570 motherboards perfect for users looking to build the best AMD platform gaming system. GIGABYTE X570 motherboards maximize your PC's potential with AMD StoreMI technology https://www.upsers.club/RyanTech - Saturday, November 2, 2019 - link
First of all, thanks for the great review! Well-written and very informative. I just have one question: Where is the last 4x lanes from the chipset usable from?I understand that the x16 and the x8 expansion slots and the M2A slot uses (20 total) lanes from the CPU and the last x4 expansion slot together with the M2B and M2C slots uses lanes from the chipset at 4x lanes each.
But if I understand correctly, that's only 12 of 16 lanes so far for this platform & motherboard (4x for expansion slot 3, 4x each for the 2nd and 3rd M.2 slots). Where are the other 4 lanes? It can't be the SATA ports, right? Because the chipset already supports 4x SATA ports natively.