You're not the only one between us that mirrors that sentiment. Even on models that don't cover the intake area, as it stands, it still needs a fan. I intend to investigate a few elements around the cooling of the X570 chipset to see what the ramifications are without active cooling.
PCI Express 4.0 being maxed out will be the primary reason why active cooling would be needed. From the diagram, no PCI Express x16, which isn't impressive, and would make me stick with an older motherboard anyway. PCI Express 3.0 x16 and PCI Express 4.0 x8 would give roughly the same bandwidth, so honestly, why bother going to a motherboard that doesn't give PCIe 4.0 x16 on even a single slot?
"Dominating the lower portion of the MSI MEG X570 Godlike's PCB is four full-length PCIe 4.0 slots which operate at x16, x8/x0/x8, and x8/x4/x4, with the last full-length slot locked down to PCIe 4.0 x4." . . . "In terms of PCIe, the MSI MEG X570 Ace has three full-length PCIe 4.0 slots which run at x16, x8/x8, and x8/x8/x4."
As someone else mentioned, they both have PCIe 4.0 x16 slots that are actually x16. Also, those slots have nothing to do with the chipset so they wouldn't contribute to chipset heating at all.
What are you talking about. Ryzen 3000 series cpus have 24 PCIe 4.0 lanes. 16 for GPU, 4 for m.2 and 4 for connecting the x570 chipset. The x570 chipset then exposes another 16 PCIe 4.0 lanes. The chpset diagram is just the ADDIITONAL PCIe lanes provides by the chipset. That is in addition to the 20 usable lanes (16 for CPU, 4 for m.2) provided by the CPU directly.
PCI-e 4.0 @ x8 is still FASTER than 3.0 x16 (less overhead and far more efficient packing and such)
pci1-2-3 was "near perfect linear" the "4" and "5" of various generations seem to "break the mold" DDR4/5 absolutely took the "norm" and split on it's head did they not? so who is to say that even though on paper 3.0 x16 or better is a MUST that does not always hold merit either.
You are correct that the TRANSFER RATE has overhead, however the THROUGHPUT is the amount of actual data that can be transferred after accounting for the OVERHEAD bits... I think where you went wrong is that PCIe v4.0 did not change the encoding scheme thus doubling the transfer rate doubles the throughput... therefore PCIe v3.0 x16 has the same THROUGHPUT capability as PCIe v4.0 x8
Regarding the upcoming x570 reviews, would you consider noting which - if any - boards support mounting standard 3rd party 40mm fans in said reviews?
As in; can you replace the existing fan setup with an off-the-shelf fan, whether to replace a broken one or to address heat/noise concerns?
I were around the last time chipset fans were a thing and I'm less than keen on their reintroduction. Knowing they can be easily replaced would at least offer some peace of mind!
Yeah, it was not uncommon to have these tiny whiny bastards fail after 1-2 years. But then it's said that on X570 they don't spin at 8k rpm all the time, but are mostly off and when active typically rev at ~3k rpm. I think Igor from THG posted a youtube video on that.
The chipset fan on my Rampage IV Extreme, certainly a more modern example of such a design, has been running flawlessly for several years now, so I'm not too worried. It was an open box board for like half the price too :D
Most of the time the fans are going to be stirring air silently. In fact, 99.9% of desktop users won't be hammering the chipset hard enough (load all the PCIe 4.0 lanes, NVMe RAID) to really stress the chipset.
With that being said there are some passive arrangements that ensure that the chipset will be cool regardless of load, but they tend to involve heatpipes (for PCIe card clearance purposes).
yeh is "stupid as hell" no doubt to do that, however, there is something to be said about "stacking" different materials for the varying properties, often plastic does NOT allow "drums" to happen or the reverb of sound like wood or metal can "enchance" and project it, unless custom designed etc, much less costly to just do plastic (for some parts)
where it becomes a problem is for a certain Ass us? company, is pretty bad when you take off the "fancy" locked back IO shield and remove m.2 heatshield that operating temerpatures and power use "magically" drop....that is where they did NOT bother to take the time to "test it" they likely just though it would look "cool" did a quick test and maybe it "just passed QT time"
who knows, I personally have nothing against most any material, how they make them and how they are used however is a VERY different story (not cheap out, use high quality in design, even the crap materials can give a very high quality product worth it's weight in gold)
With regard to actually using a motherboard, the fan controls matter to me. I hope the review, when it drops, can at least comment on the practicality / range / effectiveness of the BIOS fan header settings. These are never addressed by reviewers, who test on open chassis and instead devote hours to finding the maximum OC. My hours go into getting the cooling right, and too much of that effort can only begin with the board installed and running.
These fan features can disappoint; for example on the Asus Crosshair Hero VI the control range is artificially limited at the low end and only allows two breakpoints between min and max. Also, not all the headers have delay settings, which are very important in preventing surging. One good thing is that some of the fans can control on the max of two sensed temperatures eg CPU / VRM. Again, all these things are never even mentioned.
(Don't both telling me to use water, or anything else. I've been doing this long enough to know what I need in a PC and in a review, thanks.)
In addition to some of the questionable design elements (plastic grill/airflow restrictions - why?), please also mention the expected or manufacturer suggested price for these. I have the strong suspicion that these MoBos will cost more than my entire (planned) HTPC build, which will be around the 3400G. I guess these MoBos are "bleeding edge" indeed.
These aren't the only mobos having a mix of two different M.2 SSD lengths. I wonder why. Is it simply an extended cooler on the longer ones? The extra length appears to prevent a PCIe x1 socket from taking the space. The Ace has such a socket for the shorter M.2; unfortunately the Godlike doesn't. I'd much rather have the socket than an extended cooler, if it's that simple (and yeah I know that none of this is actually simple). Especially of the mobo M.2 cooler is expected to be discarded if the SSD has its own - another topic where some clarification would be useful.
TBH, I haven't done much research on the M.2 SSDs yet; will do that when I'm ready to build, especially given all the new PCIE4.0 announcements. Maybe there's two different lengths of SSDs, hence the different length socket areas... I haven't happened to run into that so far.
There are in fact several lengths. Most commonly 80mm, and then occasionally you will see 110mm. There are also some much shorter ones that are typically not used in desktops at 42mm, although the spec supports many different lengths.
The 110mm M.2 format gives more space thus can offer higher capacities which is why they tend to be aimed at data centres. Consumer drives are usually 80mm and can be less on some tablets.
Glad to hear that the prices quoted for the Godlike in various Internet domains are *wrong*--at least, I think I'm glad to hear it...;) (Lord help us all if the real pricing goes the other way!)
The usual prices for x570 will be from 200 to 600 USD. Of course, there should be some exceptions on both ends, like 140 or 1000 (that limited edition Asrock with liquid cooling).
Almost double from what they used to be 10 years ago. And in two more generation they will increase even further, what is now a 400 USD mobo will be 600 in two years time. The sales go down and the prices go up to recoup.
benchies? would be curious to know if there are any measurable performance benefits of the new chipsets vs last gen. Obv, new Ryzen's are still under embargo, but could you run a quick test even with a last gen chip just to see if there are any interesting or even noticeable performance differences?
THANK YOU for including VRMs! i don't even OC and i'm eyeing them closely. hopefully, it'll lead to a board that will live waaaay too ridiculously long
Ryzen 3000 CPUs themselves will have total of 24 PCIe 4.0 lanes: x4 of which to the chipset and x4 to the primary M.2 slot. The remaining x16 are usually routed to the primary and secondary expansion slots, to be used in either x16/x0 or x8/x8 configuration, although motherboards can arrange these in other ways too.
Modern chipsets are essentially PCIe hubs. The X570 is connected the the CPU using PCIe 4.0 x4 (around 8 GB/s) and splits this into more connections. The total bandwidth is limited by the uplink, but since you'll want to connect your most bandwidth-hungry devices directly to the CPU, it's really not an issue for most people.
Yeah that is exactly how every PCIe lanes off every chipset for the past decade or so has worked. The 300 series Ryzen have 24 native PCIe lanes. 16 for GPU, 4 for m.2 and 4 to connect to the chipset. The chipest then shares those for 20 PCIe worth of "stuff". Every other AMD and Intel CPU and chipset combo has worked exactly the same way except with PCIe 3 lanes not PCIe 4 lanes.
It is 8 lanes not necessarily a single x8 slot. So it could be configured by the OEM as a pair of x4 or two x1 an x4 and using the other 2 lanes for additional onboard stuff (2.5, 5, or 10 Gb ethernet, TB3, etc).
The general idea is that nothing ever gets used at the same time. Also, SLI has an x8 requirement, so even if it is a bottleneck x4 slot in the end, SLI would still work without registry or driver hacks. But I doubt many motherboards will have an x8 slot from the chipset. Some might, but not a lot.
Consider that it will take a long time before devices support PCIe 4. In the meantime, you'll need PCIe 3 x8 or PCIe 2 x16 to saturate the 8 GB/s PCIe 4 x4 uplink.
Having said that, I think x8 on the chipset is unpractical in general purpose PCs for GPUs for latency reasons, and for anything for consuming too much of the connectivity.
The x16 and x8 options may have been included simply because the silicon can do it - the X570 chipset design is repurposed from the Ryzen 3000 CPU I/O die design.
Who knows, maybe a motherboard or system manufacturer comes up with a special case where using up 8 lanes for a device makes sense?
Do you intend to review lower-end motherboards? Maybe I am the only person who wants no more from a motherboard than a processor socket, some memory, one NVMe slot, and one PCIe slot for the lowest-power video card sufficient to make the board work; but these reviews of ludicrously overbuilt things aren't terribly interesting to me.
AnandTech will review whatever manufacturers send to them, AFAIK. It's just that halo products generally tend to garner more attention and hence more traffic, so there is an obvious bias towards reviewing such.
Considering the feeding frenzy around the Ryzen 3000 launch, it's unlikely that we'll be seeing "bare-bones" X570 boards for a while yet, especially since the lower-end chipsets are only scheduled to arrive next year. If you want something with less frills, you'll either have to wait a few months or commit to Ryzen 2000 and X470/X370.
Safety, since those can apparently run very hot. But the good news is that it won't actually turn on unless it has to, so the heatpipe should hopefully reduce that to a minimum.
I don't get this fascination with 2.5G NICs by manufacturers. There are very few switches that support it, and that doesn't look likely to change before the 10G switches drop to equal prices. Why not just put a 10G NIC on there? They aren't that much more than the 2.5G NICs.
My mistake, 10G switches already are cheaper. I found five models of switches with 4X 10G ports for $100 cheaper than the cheapest multi-gigabit (2.5G/5G/10G) switch. 2.5G networking is DOA. Why use it in new motherboards?
2.5G networking has a clear but small use case - it's for businesses who can't afford to tear out cat5 and re-cable, but want backhaul from wifi access points that runs at the speed of newer kinds of wifi.
I agree that it doesn't make sense for home users, because it hasn't got to the little Netgear five-port switches yet; on the other hand there's no sign of 10GbaseT getting over the chasm from £100/port to £5/port.
until there is a 12+ port switch for 2.5 to 10g that is reasonably priced .. i'll be sticking with my current 1 gig setup, anything less then 10 ports.. is not enough....
What idiot brands these things? Godlike and Ace? MSI isn't the only company that has a marketing department filled with cluelessly lost individuals that still think UT99 is a thing, but how long are they going to lag behind before the department heads figure it out?
Yes, the names do seem odd, no question...;) It appears as if MSI simply took those names from its Zxxx motherboards for Intel of the same names...! Still, though, your question does deserve an answer as the names are fairly punk, I agree. My only guess would be that foreign marketing departments often get basic concepts in English and American idioms somewhat wrong.
Amazon US has hide nor hair of any x570 mboards listed as of today. About that Amazon UK listing, anyone know what the exchange rate is--or just close, not exact--and how much the VAT would be--since that would naturally come off of US pricing. Just curious--really surprised not to see Amazon US arranging for pre-orders--somebody is asleep at the switch. I think I'll be going with the Aorus Master, actually.
The standard VAT rate in the UK is 20%. (You could have googled "uk vat rate" and gotten it right in the search results, or you can check all countries' various rates at https://en.wikipedia.org/wiki/European_Union_value... The standard rate applies for most things.)
My bank currently offers 1 GBP = 1.27864 USD.
The total conversion ratio becomes 1.27864 / 1.20 = 1.06553.
So 777.77 GBP (incl. VAT) becomes about 829 USD (tax-free).
*
But note that the £777.77 price would indicate that it's a price holder; AMD is using "seven" heavily in the launch due to the parts being on 7nm, including launcing on the 7th of the 7th month despite it being a Sunday.
Thanks much--yes, I could have done it myself--was just hoping that someone (you?) who maybe has to do the conversions regularly could give me an off-the-cuff approximation. That's a good guess about the 777.77 placeholder! Pretty obvious--thanks again...;) Guess we're still waiting on some of these prices!
I simply DO. NOT. GET. the necessity for active cooling for a chipset that supposedly has a TDP between 11 and 15W.
My old ASUS Maximus Formula X38 motherboard had passive cooling and the X37 chipset had a TDP of 26.5W (https://ark.intel.com/content/www/us/en/ark/produc... Granted, it had a heatpipe cooling solution, but at the pricepoint of X570 motherboards one would expect this not to be an issue.
Ten years ago someone on SPCR made the effort to collect the TDP of all Intel chipsets at that time. Most used in excess of 10W. Some even idled around that mark. http://www.silentpcreview.com/forums/viewtopic.php...
Denser process nodes may concentrate the heat more than chipsets of the past? Especially since the cooling is supposed to be required (only) for dual NVMe operation in RAID 0, in which case all the heat would be concentrated on that area.
If NVMe RAID 0 throttled - or made the system unstable - people would complain too.
This is exactly it. The density has improved a lot, and the ~15W are in a much smaller area then those old chipsets had. Less area to dissipate the heat means higher temperatures, and need for more immediate cooling.
So you can preorder the MEG X570 Godlike on Amazon UK for GBP 777.77, to be delivered on 7 July. How cute. Oh, and "delivery at no extra cost for Prime members". Bargain right there!
Does anyone else have the MEG X570 GODLIKE and confirm what I am seeing?
There appears to be a serious design flaw with the MEG X570 GODLIKE. If true, this is a deal killer for me and I will have no choice but to return this board to the store.
It appears that with my PCI-E x16 video card installed, if I then install ANY other expansion card in ANY of the three, seemingly available, expansion slots, my video card is forced down to x8 only! I have an x8 AVAGO RAID card I must have, so again, this will be a complete deal killer.
If this IS true, this is one of the dumbest design mistakes I've seen in a long time. That is NOT "GODLIKE" at all. Who in their right mind would want to pay $700.00 and only get x8 from their GPU? That is just asinine!! This board comes with TWO add in cards; so using any of those, (let alone BOTH) cripples the GPU down to x8?!!!!
I can't fathom how this severe design flaw got green lit. Does Charles Chiang know about this? Someone should tell him.
I will be doing my final testing today, and if this is the case, it's going back to the store.
I suppose I should have just waited for the 3K THREADRIPPER.
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.
87 Comments
Back to Article
Gunbuster - Friday, June 28, 2019 - link
So the chipset needs a fan? but then they cover 45% of the intake surface area with plastic grill. Not a solid design choice IMHO.gavbon - Friday, June 28, 2019 - link
You're not the only one between us that mirrors that sentiment. Even on models that don't cover the intake area, as it stands, it still needs a fan. I intend to investigate a few elements around the cooling of the X570 chipset to see what the ramifications are without active cooling.Targon - Friday, June 28, 2019 - link
PCI Express 4.0 being maxed out will be the primary reason why active cooling would be needed. From the diagram, no PCI Express x16, which isn't impressive, and would make me stick with an older motherboard anyway. PCI Express 3.0 x16 and PCI Express 4.0 x8 would give roughly the same bandwidth, so honestly, why bother going to a motherboard that doesn't give PCIe 4.0 x16 on even a single slot?Irata - Friday, June 28, 2019 - link
"Dominating the lower portion of the MSI MEG X570 Godlike's PCB is four full-length PCIe 4.0 slots which operate at x16, x8/x0/x8, and x8/x4/x4, with the last full-length slot locked down to PCIe 4.0 x4.".
.
.
"In terms of PCIe, the MSI MEG X570 Ace has three full-length PCIe 4.0 slots which run at x16, x8/x8, and x8/x8/x4."
extide - Friday, June 28, 2019 - link
As someone else mentioned, they both have PCIe 4.0 x16 slots that are actually x16. Also, those slots have nothing to do with the chipset so they wouldn't contribute to chipset heating at all.TheUnhandledException - Friday, June 28, 2019 - link
What are you talking about. Ryzen 3000 series cpus have 24 PCIe 4.0 lanes. 16 for GPU, 4 for m.2 and 4 for connecting the x570 chipset. The x570 chipset then exposes another 16 PCIe 4.0 lanes. The chpset diagram is just the ADDIITONAL PCIe lanes provides by the chipset. That is in addition to the 20 usable lanes (16 for CPU, 4 for m.2) provided by the CPU directly.Dragonstongue - Saturday, June 29, 2019 - link
? huh......PCI-e 4.0 @ x8 is still FASTER than 3.0 x16 (less overhead and far more efficient packing and such)
pci1-2-3 was "near perfect linear" the "4" and "5" of various generations seem to "break the mold" DDR4/5 absolutely took the "norm" and split on it's head did they not? so who is to say that even though on paper 3.0 x16 or better is a MUST that does not always hold merit either.
my .02c of "nothing"
photonboy - Sunday, June 30, 2019 - link
Dragonstongue,NO... the throughput for PCIe lanes has DOUBLED from PCIe v3.0 to v4.0
https://en.wikipedia.org/wiki/PCI_Express#History_...
You are correct that the TRANSFER RATE has overhead, however the THROUGHPUT is the amount of actual data that can be transferred after accounting for the OVERHEAD bits... I think where you went wrong is that PCIe v4.0 did not change the encoding scheme thus doubling the transfer rate doubles the throughput... therefore PCIe v3.0 x16 has the same THROUGHPUT capability as PCIe v4.0 x8
Gastec - Tuesday, July 16, 2019 - link
Mind fart?Exodite - Friday, June 28, 2019 - link
Regarding the upcoming x570 reviews, would you consider noting which - if any - boards support mounting standard 3rd party 40mm fans in said reviews?As in; can you replace the existing fan setup with an off-the-shelf fan, whether to replace a broken one or to address heat/noise concerns?
I were around the last time chipset fans were a thing and I'm less than keen on their reintroduction. Knowing they can be easily replaced would at least offer some peace of mind!
Thanks. :)
MrSpadge - Friday, June 28, 2019 - link
Yeah, it was not uncommon to have these tiny whiny bastards fail after 1-2 years. But then it's said that on X570 they don't spin at 8k rpm all the time, but are mostly off and when active typically rev at ~3k rpm. I think Igor from THG posted a youtube video on that.Assimilator87 - Saturday, June 29, 2019 - link
The chipset fan on my Rampage IV Extreme, certainly a more modern example of such a design, has been running flawlessly for several years now, so I'm not too worried. It was an open box board for like half the price too :Dplonk420 - Friday, June 28, 2019 - link
yeah, i wanted to slap my sad and unused HR-05 IFX/SLI on the chipset back from my X58peevee - Friday, June 28, 2019 - link
A full feature comparison table, with sizes and prices, would be nice.Alexvrb - Saturday, June 29, 2019 - link
Most of the time the fans are going to be stirring air silently. In fact, 99.9% of desktop users won't be hammering the chipset hard enough (load all the PCIe 4.0 lanes, NVMe RAID) to really stress the chipset.With that being said there are some passive arrangements that ensure that the chipset will be cool regardless of load, but they tend to involve heatpipes (for PCIe card clearance purposes).
Dragonstongue - Saturday, June 29, 2019 - link
yeh is "stupid as hell" no doubt to do that, however, there is something to be said about "stacking" different materials for the varying properties, often plastic does NOT allow "drums" to happen or the reverb of sound like wood or metal can "enchance" and project it, unless custom designed etc, much less costly to just do plastic (for some parts)where it becomes a problem is for a certain Ass us? company, is pretty bad when you take off the "fancy" locked back IO shield and remove m.2 heatshield that operating temerpatures and power use "magically" drop....that is where they did NOT bother to take the time to "test it" they likely just though it would look "cool" did a quick test and maybe it "just passed QT time"
who knows, I personally have nothing against most any material, how they make them and how they are used however is a VERY different story (not cheap out, use high quality in design, even the crap materials can give a very high quality product worth it's weight in gold)
Arbie - Friday, June 28, 2019 - link
With regard to actually using a motherboard, the fan controls matter to me. I hope the review, when it drops, can at least comment on the practicality / range / effectiveness of the BIOS fan header settings. These are never addressed by reviewers, who test on open chassis and instead devote hours to finding the maximum OC. My hours go into getting the cooling right, and too much of that effort can only begin with the board installed and running.These fan features can disappoint; for example on the Asus Crosshair Hero VI the control range is artificially limited at the low end and only allows two breakpoints between min and max. Also, not all the headers have delay settings, which are very important in preventing surging. One good thing is that some of the fans can control on the max of two sensed temperatures eg CPU / VRM. Again, all these things are never even mentioned.
(Don't both telling me to use water, or anything else. I've been doing this long enough to know what I need in a PC and in a review, thanks.)
Arbie - Friday, June 28, 2019 - link
Marketing marches on. Now a trivial temperature-controlled fan is called "Zero Frozr". I wonder if they patented the idea.eastcoast_pete - Friday, June 28, 2019 - link
In addition to some of the questionable design elements (plastic grill/airflow restrictions - why?), please also mention the expected or manufacturer suggested price for these. I have the strong suspicion that these MoBos will cost more than my entire (planned) HTPC build, which will be around the 3400G. I guess these MoBos are "bleeding edge" indeed.gavbon - Friday, June 28, 2019 - link
We can't reveal that until 7/7 as it's part of the embargoshabby - Friday, June 28, 2019 - link
They'll be around $40 more than the current crop.Death666Angel - Friday, June 28, 2019 - link
Why go for the x570 platform with an APU that won't support PCIe 4.0? That doesn't make much sense anyway.1_rick - Saturday, June 29, 2019 - link
Faster m.2 drives, USB 3.2?Death666Angel - Saturday, June 29, 2019 - link
The link between the APU and the chipset will still be 3.0, so there won't be a lot of acceleration.DigitalFreak - Sunday, June 30, 2019 - link
x370 and x470 only supported PCIe 2.0. With x570, any PCIe 3.0 device you plug will run at PCIe 3.0 speeds.Karmena - Tuesday, July 2, 2019 - link
you what? All AM4 socket mobos where PCIe3. Even lousy A320 chipset.Arbie - Friday, June 28, 2019 - link
These aren't the only mobos having a mix of two different M.2 SSD lengths. I wonder why. Is it simply an extended cooler on the longer ones? The extra length appears to prevent a PCIe x1 socket from taking the space. The Ace has such a socket for the shorter M.2; unfortunately the Godlike doesn't. I'd much rather have the socket than an extended cooler, if it's that simple (and yeah I know that none of this is actually simple). Especially of the mobo M.2 cooler is expected to be discarded if the SSD has its own - another topic where some clarification would be useful.Arbie - Friday, June 28, 2019 - link
TBH, I haven't done much research on the M.2 SSDs yet; will do that when I'm ready to build, especially given all the new PCIE4.0 announcements. Maybe there's two different lengths of SSDs, hence the different length socket areas... I haven't happened to run into that so far.extide - Friday, June 28, 2019 - link
There are in fact several lengths. Most commonly 80mm, and then occasionally you will see 110mm. There are also some much shorter ones that are typically not used in desktops at 42mm, although the spec supports many different lengths.smilingcrow - Friday, June 28, 2019 - link
The 110mm M.2 format gives more space thus can offer higher capacities which is why they tend to be aimed at data centres. Consumer drives are usually 80mm and can be less on some tablets.WaltC - Friday, June 28, 2019 - link
Glad to hear that the prices quoted for the Godlike in various Internet domains are *wrong*--at least, I think I'm glad to hear it...;) (Lord help us all if the real pricing goes the other way!)Mugur - Friday, June 28, 2019 - link
The usual prices for x570 will be from 200 to 600 USD. Of course, there should be some exceptions on both ends, like 140 or 1000 (that limited edition Asrock with liquid cooling).Scary prices.
Gastec - Tuesday, July 16, 2019 - link
Almost double from what they used to be 10 years ago. And in two more generation they will increase even further, what is now a 400 USD mobo will be 600 in two years time. The sales go down and the prices go up to recoup.1_rick - Friday, June 28, 2019 - link
"orientated"Please don't do this. Just say oriented.
SkyBill40 - Friday, June 28, 2019 - link
I see I'm not the only one who cringed at that.DigitalFreak - Friday, June 28, 2019 - link
Seems stupid to include 2 NICs onboard, and then the 10Gig as an add-in card.Targon - Friday, June 28, 2019 - link
In some business environments, you need two ethernet connections for security and such.DigitalFreak - Sunday, June 30, 2019 - link
I don't see many businesses using an $800 motherboard.tpurves - Friday, June 28, 2019 - link
benchies? would be curious to know if there are any measurable performance benefits of the new chipsets vs last gen. Obv, new Ryzen's are still under embargo, but could you run a quick test even with a last gen chip just to see if there are any interesting or even noticeable performance differences?gavbon - Friday, June 28, 2019 - link
Performance data is under embargo until 7/7. Just wanted people to see what both these boards had to offer prior to the launchDeath666Angel - Friday, June 28, 2019 - link
Thanks for that! Any mATX variants on the way? :D Or is it just the AsRock Pro4? Would be more than we had X470 mATX motherboards.plonk420 - Friday, June 28, 2019 - link
THANK YOU for including VRMs! i don't even OC and i'm eyeing them closely. hopefully, it'll lead to a board that will live waaaay too ridiculously longlevizx - Friday, June 28, 2019 - link
I don't understand how this works. The CPU and X570 are linked by PCIe 4.0 x4, but the X570 then provides roughly PCIe 4.0 x20 worth of I/O?Hul8 - Friday, June 28, 2019 - link
Ryzen 3000 CPUs themselves will have total of 24 PCIe 4.0 lanes: x4 of which to the chipset and x4 to the primary M.2 slot. The remaining x16 are usually routed to the primary and secondary expansion slots, to be used in either x16/x0 or x8/x8 configuration, although motherboards can arrange these in other ways too.Modern chipsets are essentially PCIe hubs. The X570 is connected the the CPU using PCIe 4.0 x4 (around 8 GB/s) and splits this into more connections. The total bandwidth is limited by the uplink, but since you'll want to connect your most bandwidth-hungry devices directly to the CPU, it's really not an issue for most people.
levizx - Saturday, June 29, 2019 - link
Yea, but what's the point of supporting x8 link?TheUnhandledException - Saturday, June 29, 2019 - link
Yeah that is exactly how every PCIe lanes off every chipset for the past decade or so has worked. The 300 series Ryzen have 24 native PCIe lanes. 16 for GPU, 4 for m.2 and 4 to connect to the chipset. The chipest then shares those for 20 PCIe worth of "stuff". Every other AMD and Intel CPU and chipset combo has worked exactly the same way except with PCIe 3 lanes not PCIe 4 lanes.levizx - Saturday, June 29, 2019 - link
I simply don't understand what's the point of having x8? Sure 2 or more x4 links can share one x4 upstream, but how do you do that with x8?TheUnhandledException - Saturday, June 29, 2019 - link
It is 8 lanes not necessarily a single x8 slot. So it could be configured by the OEM as a pair of x4 or two x1 an x4 and using the other 2 lanes for additional onboard stuff (2.5, 5, or 10 Gb ethernet, TB3, etc).Death666Angel - Saturday, June 29, 2019 - link
The general idea is that nothing ever gets used at the same time. Also, SLI has an x8 requirement, so even if it is a bottleneck x4 slot in the end, SLI would still work without registry or driver hacks. But I doubt many motherboards will have an x8 slot from the chipset. Some might, but not a lot.Hul8 - Monday, July 1, 2019 - link
Consider that it will take a long time before devices support PCIe 4. In the meantime, you'll need PCIe 3 x8 or PCIe 2 x16 to saturate the 8 GB/s PCIe 4 x4 uplink.Having said that, I think x8 on the chipset is unpractical in general purpose PCs for GPUs for latency reasons, and for anything for consuming too much of the connectivity.
The x16 and x8 options may have been included simply because the silicon can do it - the X570 chipset design is repurposed from the Ryzen 3000 CPU I/O die design.
Who knows, maybe a motherboard or system manufacturer comes up with a special case where using up 8 lanes for a device makes sense?
Hul8 - Friday, June 28, 2019 - link
The article said:> The X570 chipset swaps out PCIe 4.0 lanes over the previous X470 and X370's use of PCIe 3.0
That's only true for the uplink lanes (between the CPU and the chipset). Downstream lanes on the earlier chipsets were all PCIe 2.
TomWomack - Saturday, June 29, 2019 - link
Do you intend to review lower-end motherboards? Maybe I am the only person who wants no more from a motherboard than a processor socket, some memory, one NVMe slot, and one PCIe slot for the lowest-power video card sufficient to make the board work; but these reviews of ludicrously overbuilt things aren't terribly interesting to me.The_Assimilator - Saturday, June 29, 2019 - link
AnandTech will review whatever manufacturers send to them, AFAIK. It's just that halo products generally tend to garner more attention and hence more traffic, so there is an obvious bias towards reviewing such.Considering the feeding frenzy around the Ryzen 3000 launch, it's unlikely that we'll be seeing "bare-bones" X570 boards for a while yet, especially since the lower-end chipsets are only scheduled to arrive next year. If you want something with less frills, you'll either have to wait a few months or commit to Ryzen 2000 and X470/X370.
haukionkannel - Sunday, June 30, 2019 - link
Ryzen 3000 and 470 or 450 works just fine!But yeah, best bang for the buck most likely from 2000 series!
PVG - Saturday, June 29, 2019 - link
They pull a heatpipe all the way down to the chipset and still put a fan there? What's the point, then?nevcairiel - Wednesday, July 3, 2019 - link
Safety, since those can apparently run very hot. But the good news is that it won't actually turn on unless it has to, so the heatpipe should hopefully reduce that to a minimum.dgingeri - Saturday, June 29, 2019 - link
I don't get this fascination with 2.5G NICs by manufacturers. There are very few switches that support it, and that doesn't look likely to change before the 10G switches drop to equal prices. Why not just put a 10G NIC on there? They aren't that much more than the 2.5G NICs.dgingeri - Saturday, June 29, 2019 - link
My mistake, 10G switches already are cheaper. I found five models of switches with 4X 10G ports for $100 cheaper than the cheapest multi-gigabit (2.5G/5G/10G) switch. 2.5G networking is DOA. Why use it in new motherboards?TomWomack - Sunday, June 30, 2019 - link
2.5G networking has a clear but small use case - it's for businesses who can't afford to tear out cat5 and re-cable, but want backhaul from wifi access points that runs at the speed of newer kinds of wifi.I agree that it doesn't make sense for home users, because it hasn't got to the little Netgear five-port switches yet; on the other hand there's no sign of 10GbaseT getting over the chasm from £100/port to £5/port.
Qasar - Monday, July 1, 2019 - link
until there is a 12+ port switch for 2.5 to 10g that is reasonably priced .. i'll be sticking with my current 1 gig setup, anything less then 10 ports.. is not enough....piroroadkill - Monday, July 1, 2019 - link
I have one (Netgear MS510TX) that supports 10/5/2.5. 2.5Gig is a good idea.. it works over Cat5e up to 100mPeachNCream - Sunday, June 30, 2019 - link
What idiot brands these things? Godlike and Ace? MSI isn't the only company that has a marketing department filled with cluelessly lost individuals that still think UT99 is a thing, but how long are they going to lag behind before the department heads figure it out?WaltC - Sunday, June 30, 2019 - link
Yes, the names do seem odd, no question...;) It appears as if MSI simply took those names from its Zxxx motherboards for Intel of the same names...! Still, though, your question does deserve an answer as the names are fairly punk, I agree. My only guess would be that foreign marketing departments often get basic concepts in English and American idioms somewhat wrong.WaltC - Sunday, June 30, 2019 - link
Amazon US has hide nor hair of any x570 mboards listed as of today. About that Amazon UK listing, anyone know what the exchange rate is--or just close, not exact--and how much the VAT would be--since that would naturally come off of US pricing. Just curious--really surprised not to see Amazon US arranging for pre-orders--somebody is asleep at the switch. I think I'll be going with the Aorus Master, actually.Qasar - Sunday, June 30, 2019 - link
there is the possibility of the NDA still in effect...Hul8 - Monday, July 1, 2019 - link
The standard VAT rate in the UK is 20%. (You could have googled "uk vat rate" and gotten it right in the search results, or you can check all countries' various rates at https://en.wikipedia.org/wiki/European_Union_value... The standard rate applies for most things.)My bank currently offers 1 GBP = 1.27864 USD.
The total conversion ratio becomes 1.27864 / 1.20 = 1.06553.
So 777.77 GBP (incl. VAT) becomes about 829 USD (tax-free).
*
But note that the £777.77 price would indicate that it's a price holder; AMD is using "seven" heavily in the launch due to the parts being on 7nm, including launcing on the 7th of the 7th month despite it being a Sunday.
Hul8 - Monday, July 1, 2019 - link
*place holderWaltC - Monday, July 1, 2019 - link
Thanks much--yes, I could have done it myself--was just hoping that someone (you?) who maybe has to do the conversions regularly could give me an off-the-cuff approximation. That's a good guess about the 777.77 placeholder! Pretty obvious--thanks again...;) Guess we're still waiting on some of these prices!Gastec - Tuesday, July 16, 2019 - link
At last they don't call them Pro Gamer...or do they? :)piroroadkill - Monday, July 1, 2019 - link
Almost 800 quid for a.. motherboardHAHAHAHAHAHAHAHA
.. no.
azrael- - Monday, July 1, 2019 - link
I simply DO. NOT. GET. the necessity for active cooling for a chipset that supposedly has a TDP between 11 and 15W.My old ASUS Maximus Formula X38 motherboard had passive cooling and the X37 chipset had a TDP of 26.5W (https://ark.intel.com/content/www/us/en/ark/produc... Granted, it had a heatpipe cooling solution, but at the pricepoint of X570 motherboards one would expect this not to be an issue.
Ten years ago someone on SPCR made the effort to collect the TDP of all Intel chipsets at that time. Most used in excess of 10W. Some even idled around that mark.
http://www.silentpcreview.com/forums/viewtopic.php...
Have engineers suddenly become incompetent?
azrael- - Monday, July 1, 2019 - link
X37 == X38. Too bad Anandtech cba to implement an edit feature...piroroadkill - Monday, July 1, 2019 - link
Marketing took over. It's complete crap. Why can't we have nice heatpipes snaking around the board to large, FINNED heatsinks without bloody logosHul8 - Monday, July 1, 2019 - link
Denser process nodes may concentrate the heat more than chipsets of the past? Especially since the cooling is supposed to be required (only) for dual NVMe operation in RAID 0, in which case all the heat would be concentrated on that area.If NVMe RAID 0 throttled - or made the system unstable - people would complain too.
nevcairiel - Wednesday, July 3, 2019 - link
This is exactly it. The density has improved a lot, and the ~15W are in a much smaller area then those old chipsets had. Less area to dissipate the heat means higher temperatures, and need for more immediate cooling.Ironchef3500 - Monday, July 1, 2019 - link
Jesus, that price...azrael- - Monday, July 1, 2019 - link
So you can preorder the MEG X570 Godlike on Amazon UK for GBP 777.77, to be delivered on 7 July. How cute. Oh, and "delivery at no extra cost for Prime members". Bargain right there!WaltC - Monday, July 1, 2019 - link
Ha-ha..;) Bargains like that give me indigestion....;)just4U - Monday, July 1, 2019 - link
Dear Santa,Since Christmas is coming early this year, I would like a MSI Godlike X570 Motherboard please.
Thx,
Roland
peevee - Tuesday, July 2, 2019 - link
There is no X470 micro-ATX boards. You need to go down to B450 to have that FF.Now, there is no B550 announced. Will there be any micro-ATX X570 boards?
just4U - Wednesday, July 3, 2019 - link
Good chance there will be... as the various motherboard companies have 50+ models their releasing for the x570 chipset.just4U - Wednesday, July 3, 2019 - link
Asrock has pictures of theirs... https://www.anandtech.com/show/14502/asrock-x570-p...GaryTheBusmna - Tuesday, July 9, 2019 - link
for god sake show people UNDER THE CHIPSET FANDBVD - Friday, July 19, 2019 - link
Does anyone else have the MEG X570 GODLIKE and confirm what I am seeing?There appears to be a serious design flaw with the MEG X570 GODLIKE. If true, this is a deal killer for me and I will have no choice but to return this board to the store.
It appears that with my PCI-E x16 video card installed, if I then install ANY other expansion card in ANY of the three, seemingly available, expansion slots, my video card is forced down to x8 only! I have an x8 AVAGO RAID card I must have, so again, this will be a complete deal killer.
If this IS true, this is one of the dumbest design mistakes I've seen in a long time. That is NOT "GODLIKE" at all. Who in their right mind would want to pay $700.00 and only get x8 from their GPU? That is just asinine!! This board comes with TWO add in cards; so using any of those, (let alone BOTH) cripples the GPU down to x8?!!!!
I can't fathom how this severe design flaw got green lit. Does Charles Chiang know about this? Someone should tell him.
I will be doing my final testing today, and if this is the case, it's going back to the store.
I suppose I should have just waited for the 3K THREADRIPPER.
DBVD - Friday, July 19, 2019 - link
Ok, FYILooks like they have removed BIOS 7C34v12 from their site, and that is the BIOS I have been using.
Additionally, I have been experiencing two other issues:
- VGA ERROR displayed on DYNAMIC DISPLAY on every boot, with any video card.
- Occasional CPU ERROR displayed on DYNAMIC DISPLAY. For sure happens with 4 DIMMs installed and setting A-XMP. Hangs machine, reset BIOS required.
Just got a reply from their tech support who have provided me a link to a BETA E7C34AMS.132 BIOS they want me to try. Will try tonight after work.
DBVD - Friday, July 19, 2019 - link
https://www.dropbox.com/s/35d2tcarceaevnw/E7C34AMS...