Your alternative is to buy a significant number of boards (hundreds and up) and use them over a significant period (years) in order to have any hope of a reliability statistic based on a remotely relevant data set. Are you? Because otherwise when it comes to "quality" I agree, you DO base your buying decisions exclusively on online comments like this.
Just out of interest, banter aside, I've been an incidental fan of Asus mobos for many decades in that, whilst I consider all the major brands, I always seem to end up with an Asus. It's not intentional, it's just perhaps that their selection of feature sets to market perhaps always seems to meet my needs. I've never had an Asus fail on me it must be said, (aside from an incident stemming purely from my own cackhandedness around a CPU socket) I've never had an issue and they've always seemed well built. My last Asus board was bought around 5-6 years ago and was just the same as the rest in terms of build. Is the general consensus that Asus's build quality has dropped and they should be treated with caution?
I've had plenty of Asus and they have all been fine. I've got a X99 Asus at the moment and its been pretty solid. To be honest I've not seen a failed motherboard since the removal of electrolytic caps around 2006. I remember around 2009-2011 I was getting masses of PCs in from 2003/2006 with 'issues' and as soon as I opened them up I would see the crusty bulging caps.
Boards have physically improved all round I would say. Now if they could get the bugs out of the BIOS...
Yes, bugs, pathetic. PC industry is supposedly competitive, yet so many present half baked solutions, and lots of marketing talk. ASUS charges extra premium and still doesn't deliver a bug free BIOS.
I have a ROG board that stopped receiving driver and BIOS updates less than 2 years after launch. This wouldn't matter much if the existing drivers worked properly under Win10 and offered all the functionality advertised in big letters on the box. It didn't. I simply take that as a lesson to stay away from the praised high end. At least when they inevitably sh*t the pants I will not be too disappointed.
But yes, it's still working although with just the occasional reset in memory parameters where it defaults to the lowest memory frequency. And some other small annoyances with the drive order. Etc. Great job they did on that BIOS.
For the last 12 years or so I've been exclusively buying Asus motherboards for desktop machines, and Supermicro motherboards for servers & rendernodes. Very roughly 25 of the former and 30 of the latter. So far I've had zero failures. The older of the Supermicro boards have seen 24/7 use for about 8 years!
Their build quality remains excellent. Their efforts in VRMs in recent times (if you're in to heavy overclocking) are a different story though. Gigabyte's range is utterly dominant there at the moment (although their UEFI really sucks in terms of UX).
Personally, Asus have usually been the boards I've ended up opting for. Gigabyte's boards are better overclockers and have better power delivery and NOBODY has worse service...but they just nail everything else.
That said, I'm planning on a new build later in the year (hopefully with Zen 2/Ryzen 3000, but let's see how it turns out first) and am currently leaning towards a Gigabyte board.
Supermicro has been known for server and workstation motherboards, so the idea is that they are obviously going after the common "work hard, play hard" manifesto that has been around for ages
THis has everything I would ever want in a board. I even like the styling (after that dumb tag lien is covered up by components). Except it doesn't have Thunderbolt 3. Why is this so hard to find in a desktop system? Surely they could had dropped some USB 3.1 ports for at least one?
Actually, Intel dropped licensing on Thunder turd 3... probably because no one would pay for it... and they didn't want the turd to shrivel up and die, yet.
No, they stopped *charging* for it, per-item. It's royalty-free and non-exclusive but still licensed.
We don't know the exact terms under which that license is granted, but I wouldn't be surprised if certification was required and that the license might be revoked if action was taken against Intel.
One interesting fact from https://thunderbolttechnology.net/tech/certificati... - "Peripheral devices are certified to be compatible with specific operating systems and *the devices are not end user upgradeable* for additional operating system compatibility." Not sure if this is just Intel covering their ass for incompatibility or an actual restriction on licensees providing user-upgradability.
The truly bizarre thing about this motherboard's existance is that SLI and CF are seeing little support in new titles lately, and last I checked you have to get hacked nvidia drivers to support more than 2 cards in SLI for the 10+ series anyways.
Might have been useful 5-10 years ago when AFR mGPU support didn't stink so much.
The 10 series cards don't have enough SLI connectors to run in more than 2x SLI without severely downgraded bandwidth. It takes both connnectors for a 2 card HB SLI bridge. For the 20 series the RTX 2080 is the cheapest card that supports SLI. Nvidia is definitely limiting SLI to the most expensive cards.
Pretty much abandonware. The z390 Master from Gigabyte is significantly better. Audio improvements, 10gb internet, 3x NVME slots (although capped to 3 gb/s), and significantly better support than Asus. In addition, supermicro is way too cautious in the bios department than other manufacturers. (Bought a brand new Skylake Xeon motherboard from them, and had to select "experimental" NVME x4 to run a 760p from Intel!) Not worthy of mainstream.
I've had similar issues with a SuperO board with PLX chip in the past, probably 2 years. Very common 16GB DDR4 set of Corsair Memory (DDR4-3000, CL15), wouldn't post with XMP enabled. I had to submit a support ticket to see what they could do about getting the memory kit supported. They had to order the memory in and find some stable memory timings, and sent me over the list of timings I should change. (Yeah, quite a few of the memory subtimings beyond the main 5 or so needed configuring... One of those subtimings went from a default 8 cycles to 22 cycles or something.)
I wouldn't say they provide terrible value or terrible boards, but uhhh... Sometimes you just want stuff to work out of the box, and right now they're still a bit behind the usual motherboard vendors in regards to ease of use to work with.
Another swing and miss from Supermicro. SLI is going the way of the dodo and quad-SLI especially is already extinct, so the PLX chip is well-nigh useless. Then we have not one, but two U.2 ports which are also useless.
I dunno, man. Every time I read a review of a SM board it's like they just grabbed a bunch of arbitrary features and threw them together without any thought as to whether they even make sense together.
"There are two USB 2.0 headers, two USB 3.1 G1 headers and one USB 3.1 G2 Type-C header."
It's pretty interesting how you supposedly did a visual inspection yet failed to notice there are not two, but one USB 2.0 header; ditto for the USB 3.0 header.
- Testing this board with a 8700K is not exactly ideal; one would have expected a 9900K, which you do own. A very bad start from the get go. No one can know what the board's really capable of, so there goes the entire review. Well done. - You mention an LLC of 6, but without specifying what it entails; there are board manufacturers that use high numbers for lowest, others for highest. A serious omission here. Were you even aware? - To add to the confusion (one thinks lack of understanding's the issue here), you criticise vdroop twice and that's before you even mention the LLC.. - You mention the PLX chip time and time again, but fail to even convince us you grasp of its downsides. A testing of any PCIe interface card other than GPUs would have been a good start towards that. Further testing performance with just a GPU to compare would've been even better. - We've zero interest in the thermal throttling of your specific CPU, nor any knowledge as to whether it might be a 'dud', or not. Using its thermal throttling so as to 'comment' on the board's OC capabilities is.. ridiculous, sorry. - Automated RAM timings are what's usually the make or break in terms of frequencies.. and you don't even mention how they're handled when on a manual RAM overclocking. - Personally? I'm not sure you should be doing board reviews in the first place.
Not the first Anandtech review i see that is bad, but.. do please improve.
I'm curious if that switch in the way of PCIe lanes is a performance bottleneck. And it would have been nice if you benchmarked 10G ethernet to see if it reaches its advertised speed during peak load (doubt it since it shares bandwidth with so many devices on the PCH).
welcome to the lack of foresight on intels part, and not increasing the lanes available on their platforms... 40/44 lanes is just not enough any more for the HEDT
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
shabby - Friday, February 1, 2019 - link
Play harder? These guys need some new marketing people.crotach - Friday, February 1, 2019 - link
They could include a sticker in the box that says "My other motherboard is an Asus"prophet001 - Friday, February 1, 2019 - link
Seems like Asus' quality has gone down in recent years?GTVic - Friday, February 1, 2019 - link
I base my buying decisions exclusively on online comments like this.close - Tuesday, February 5, 2019 - link
Your alternative is to buy a significant number of boards (hundreds and up) and use them over a significant period (years) in order to have any hope of a reliability statistic based on a remotely relevant data set. Are you? Because otherwise when it comes to "quality" I agree, you DO base your buying decisions exclusively on online comments like this.GreenReaper - Friday, February 1, 2019 - link
Well that's why you got a Supermicro.philehidiot - Saturday, February 2, 2019 - link
Just out of interest, banter aside, I've been an incidental fan of Asus mobos for many decades in that, whilst I consider all the major brands, I always seem to end up with an Asus. It's not intentional, it's just perhaps that their selection of feature sets to market perhaps always seems to meet my needs. I've never had an Asus fail on me it must be said, (aside from an incident stemming purely from my own cackhandedness around a CPU socket) I've never had an issue and they've always seemed well built. My last Asus board was bought around 5-6 years ago and was just the same as the rest in terms of build. Is the general consensus that Asus's build quality has dropped and they should be treated with caution?jabber - Sunday, February 3, 2019 - link
I've had plenty of Asus and they have all been fine. I've got a X99 Asus at the moment and its been pretty solid. To be honest I've not seen a failed motherboard since the removal of electrolytic caps around 2006. I remember around 2009-2011 I was getting masses of PCs in from 2003/2006 with 'issues' and as soon as I opened them up I would see the crusty bulging caps.Boards have physically improved all round I would say. Now if they could get the bugs out of the BIOS...
Gadgety - Monday, February 4, 2019 - link
Yes, bugs, pathetic. PC industry is supposedly competitive, yet so many present half baked solutions, and lots of marketing talk. ASUS charges extra premium and still doesn't deliver a bug free BIOS.close - Tuesday, February 5, 2019 - link
I have a ROG board that stopped receiving driver and BIOS updates less than 2 years after launch. This wouldn't matter much if the existing drivers worked properly under Win10 and offered all the functionality advertised in big letters on the box. It didn't. I simply take that as a lesson to stay away from the praised high end. At least when they inevitably sh*t the pants I will not be too disappointed.But yes, it's still working although with just the occasional reset in memory parameters where it defaults to the lowest memory frequency. And some other small annoyances with the drive order. Etc. Great job they did on that BIOS.
colonelclaw - Wednesday, February 6, 2019 - link
For the last 12 years or so I've been exclusively buying Asus motherboards for desktop machines, and Supermicro motherboards for servers & rendernodes. Very roughly 25 of the former and 30 of the latter. So far I've had zero failures. The older of the Supermicro boards have seen 24/7 use for about 8 years!althaz - Tuesday, February 5, 2019 - link
Their build quality remains excellent. Their efforts in VRMs in recent times (if you're in to heavy overclocking) are a different story though. Gigabyte's range is utterly dominant there at the moment (although their UEFI really sucks in terms of UX).Personally, Asus have usually been the boards I've ended up opting for. Gigabyte's boards are better overclockers and have better power delivery and NOBODY has worse service...but they just nail everything else.
That said, I'm planning on a new build later in the year (hopefully with Zen 2/Ryzen 3000, but let's see how it turns out first) and am currently leaning towards a Gigabyte board.
Questor - Thursday, February 7, 2019 - link
I have been using EVGA motherboards for some time now with excellent results. What can I say? I run away from the crowds.Big problem now is, I want to build a Ryzen system with the upcoming CPUs. I am going to have to return to the mainstream 4. It scares me.
bunnyfubbles - Tuesday, February 5, 2019 - link
Supermicro has been known for server and workstation motherboards, so the idea is that they are obviously going after the common "work hard, play hard" manifesto that has been around for agesCoryS - Friday, February 1, 2019 - link
THis has everything I would ever want in a board. I even like the styling (after that dumb tag lien is covered up by components). Except it doesn't have Thunderbolt 3. Why is this so hard to find in a desktop system? Surely they could had dropped some USB 3.1 ports for at least one?GreenReaper - Friday, February 1, 2019 - link
Licensing it probably comes with a covenant not to sue Intel.Freeb!rd - Friday, February 1, 2019 - link
Actually, Intel dropped licensing on Thunder turd 3... probably because no one would pay for it... and they didn't want the turd to shrivel up and die, yet.GreenReaper - Sunday, February 3, 2019 - link
No, they stopped *charging* for it, per-item. It's royalty-free and non-exclusive but still licensed.We don't know the exact terms under which that license is granted, but I wouldn't be surprised if certification was required and that the license might be revoked if action was taken against Intel.
One interesting fact from https://thunderbolttechnology.net/tech/certificati... - "Peripheral devices are certified to be compatible with specific operating systems and *the devices are not end user upgradeable* for additional operating system compatibility." Not sure if this is just Intel covering their ass for incompatibility or an actual restriction on licensees providing user-upgradability.
Tecnoc - Friday, February 1, 2019 - link
As far as I know while the PLX chip allows for x8/x8/x8/x8 SLI is not supported. Does SLI have to be officially supported to work?blppt - Friday, February 1, 2019 - link
The truly bizarre thing about this motherboard's existance is that SLI and CF are seeing little support in new titles lately, and last I checked you have to get hacked nvidia drivers to support more than 2 cards in SLI for the 10+ series anyways.Might have been useful 5-10 years ago when AFR mGPU support didn't stink so much.
Flunk - Sunday, February 3, 2019 - link
The 10 series cards don't have enough SLI connectors to run in more than 2x SLI without severely downgraded bandwidth. It takes both connnectors for a 2 card HB SLI bridge. For the 20 series the RTX 2080 is the cheapest card that supports SLI. Nvidia is definitely limiting SLI to the most expensive cards.FastCarsLike - Friday, February 1, 2019 - link
Pretty much abandonware. The z390 Master from Gigabyte is significantly better. Audio improvements, 10gb internet, 3x NVME slots (although capped to 3 gb/s), and significantly better support than Asus. In addition, supermicro is way too cautious in the bios department than other manufacturers. (Bought a brand new Skylake Xeon motherboard from them, and had to select "experimental" NVME x4 to run a 760p from Intel!) Not worthy of mainstream.JoeyJoJo123 - Friday, February 1, 2019 - link
I've had similar issues with a SuperO board with PLX chip in the past, probably 2 years. Very common 16GB DDR4 set of Corsair Memory (DDR4-3000, CL15), wouldn't post with XMP enabled. I had to submit a support ticket to see what they could do about getting the memory kit supported. They had to order the memory in and find some stable memory timings, and sent me over the list of timings I should change. (Yeah, quite a few of the memory subtimings beyond the main 5 or so needed configuring... One of those subtimings went from a default 8 cycles to 22 cycles or something.)I wouldn't say they provide terrible value or terrible boards, but uhhh... Sometimes you just want stuff to work out of the box, and right now they're still a bit behind the usual motherboard vendors in regards to ease of use to work with.
Rocket321 - Sunday, February 3, 2019 - link
It does, however, sound like they provided you with an amazing level of support. Above and beyond what I would have expected the response to be.StrangerGuy - Saturday, February 2, 2019 - link
Reviewing a mobo with 10GbE/PLX but not do any testing on both features seems to miss the whole point completely.The_Assimilator - Saturday, February 2, 2019 - link
Another swing and miss from Supermicro. SLI is going the way of the dodo and quad-SLI especially is already extinct, so the PLX chip is well-nigh useless. Then we have not one, but two U.2 ports which are also useless.I dunno, man. Every time I read a review of a SM board it's like they just grabbed a bunch of arbitrary features and threw them together without any thought as to whether they even make sense together.
"There are two USB 2.0 headers, two USB 3.1 G1 headers and one USB 3.1 G2 Type-C header."
It's pretty interesting how you supposedly did a visual inspection yet failed to notice there are not two, but one USB 2.0 header; ditto for the USB 3.0 header.
jabber - Sunday, February 3, 2019 - link
Yep I would just prefer 1x16/4x4Aenra - Saturday, February 2, 2019 - link
Lurker that couldn't help but (finally) comment..- Testing this board with a 8700K is not exactly ideal; one would have expected a 9900K, which you do own. A very bad start from the get go. No one can know what the board's really capable of, so there goes the entire review. Well done.
- You mention an LLC of 6, but without specifying what it entails; there are board manufacturers that use high numbers for lowest, others for highest. A serious omission here. Were you even aware?
- To add to the confusion (one thinks lack of understanding's the issue here), you criticise vdroop twice and that's before you even mention the LLC..
- You mention the PLX chip time and time again, but fail to even convince us you grasp of its downsides. A testing of any PCIe interface card other than GPUs would have been a good start towards that. Further testing performance with just a GPU to compare would've been even better.
- We've zero interest in the thermal throttling of your specific CPU, nor any knowledge as to whether it might be a 'dud', or not. Using its thermal throttling so as to 'comment' on the board's OC capabilities is.. ridiculous, sorry.
- Automated RAM timings are what's usually the make or break in terms of frequencies.. and you don't even mention how they're handled when on a manual RAM overclocking.
- Personally? I'm not sure you should be doing board reviews in the first place.
Not the first Anandtech review i see that is bad, but.. do please improve.
Korguz - Sunday, February 3, 2019 - link
Aenra... if you think you can do better.. then please do...Personally ? im not sure you would be able to do a better review...
yobbo7 - Saturday, February 2, 2019 - link
So, TU-62/NP175 are both mid-Tg generic FR4 class materials with a dielectric constant of over 4.This is standard stuff, it is not special at all and I would be surprised if anyone in the industry used anything worse on this class of motherboard.
There are still better materials available from the likes of Isola before you get into the RF materials like Rogers.
Lord of the Bored - Saturday, February 2, 2019 - link
"Both of the power delivery heatsinks ... are constructed of metal."I would certainly hope so. What else would one make a heatsink out of?
GreenReaper - Sunday, February 3, 2019 - link
Diamond? At least in part. There has been some promise shown by diamond-metal composites - the idea is more to spread the heat from the center of the die to the edges for further conduction (thermal conductivity raised ~25% for Noctua):http://www.rhp-technology.com/en/products/diacool-...
https://www.eteknix.com/computex-noctua-show-coppe...
https://noctua.at/media/wysiwyg/images/computex_20...
http://web.archive.org/web/20061004161221/http://w...
https://web.archive.org/web/20181117185351/http://...
https://web.archive.org/web/20071004215124/http://... (see illustration to right on page 2)
https://patents.google.com/patent/US20030183368 (abandoned patent, just an example)
Realistically water works well, but you could use diamond-copper in combination with it. Or if you want an edge, maybe silver?
https://www.overclockers.com/easy-heatsink-mods-to...
And of course, diamond grease:
https://www.overclockers.com/diamond-thermal-greas...
zbc - Monday, February 4, 2019 - link
I'm curious if that switch in the way of PCIe lanes is a performance bottleneck. And it would have been nice if you benchmarked 10G ethernet to see if it reaches its advertised speed during peak load (doubt it since it shares bandwidth with so many devices on the PCH).Korguz - Tuesday, February 5, 2019 - link
welcome to the lack of foresight on intels part, and not increasing the lanes available on their platforms... 40/44 lanes is just not enough any more for the HEDTPeterSun - Wednesday, February 6, 2019 - link
Now I know what the red circle in their company logo means. XDadiots123 - Saturday, February 9, 2019 - link
Sli impossible!!C9z390-pgw doesn't pay for sli certification license!!!
niemi - Thursday, February 14, 2019 - link
Anandtech, which program do you use to measure DPC latency?Is it idle or load numbers, and for how long do you run the test?
Any chance of a ‘highest measured’ or a curve over time as a supplement to the median? That would help when comparing for spikes.
Dan62 - Thursday, March 21, 2019 - link
Is it possible to mirror the 2 M.2 drives?Dan62 - Thursday, March 21, 2019 - link
Will this board support a redundant power supply?