Peformance Testing & Methodology

For general performance testing, we use the same basic test setup for both CPU and motherboard evaluation making the results comparable. Each test is run multiple times to ensure accuracy. The middle result is used in each case. The following system configurations were used for all benchmarks and general testing.

Due to potential scheduler improvements with Windows 10 that have happened over time as well as other tweaks we are using the latest build available at the time of this writing. For reference, the current Windows 10 build is 2004. We are using Windows 10 Professional for reference. All the latest patches have been applied and the driver versions are noted in the specifications. These are not necessarily the newest as we want the game performance to be more consistent across a broader sample of system configurations. We will update these periodically and retest as needed.

All systems were freshly formatted, and all the latest drivers and OS patches were used. All the systems were updated to their latest BIOS revisions. Finally, for the Intel system, I did install the CPU microcode updates relevant to that CPU. It’s important to note that build 1909 does contain improved mitigations for several security flaws on Intel processors. However, I did not go out of my way to download any additional or optional mitigation patches. Hyperthreading (SMT for AMD) also remained enabled for all testing.

We are using the performance power plan on all our test configurations. Essentially, we created a “best case” scenario for each system outside of the hardware configurations. For the hardware, it was impossible to use the same memory modules on all the test systems due to the nature of memory compatibility on different motherboards. That said, we were able to use common frequencies and keep the timings relatively close for the most part. The memory timings and the speeds we used are referenced in the specification tables above.

One thing to keep in mind is that performance is largely determined by CPU and memory configuration, or that of the graphics card rather than the motherboard itself. The benchmarks here ensure that the motherboard was able to hold up to the testing and that there were no problems with firmware, thermals, or power delivery, which caused performance to drop off. It’s also worth noting that overclocked results are provided to not only showcase what the motherboard can do when pushed but that it was able to sustain the necessary power delivery to make it through the torture testing another time with greater demand placed on it than that of stock operation.

Finally, systems were run at stock and overclocked values. For clarification: “Stock” settings are their automatic or default values in BIOS which allows the CPU’s tested to operate using their default base and boost clocks. Overclocked values provided are all core overclocks unless otherwise noted. These clock speeds are the maximum 24/7 stable overclock we could achieve for each configuration. AVX offsets are not used unless stated. Results, where clock speeds are not explicitly shown, are “stock” values. These are checked to ensure that boost clock behavior is normal and that temperatures are in the correct ranges to avoid throttling and performance anomalies as much as possible.

Join the Conversation

17 Comments

  1. Nice review – I built a system using the non-wifi version last week (well my niece did most the work I just had surgery).

    The board was easy to work with and while things went slow (for obvious reasons) putting it together was easy enough.

    One point I didnt see and may have missed in the review is the two m.2 slots are not equal. The one with the heat sink is Pcie Gen 4 and on the main Pcie bus, the other m.2 is Gen 3 and connected to the chipset. If you have a Gen 4 NVME drive this could be important.

    Would have been nice if the PCIE x1 slots allowed for larger cards to be plugged in – I have a PCIE card that can hold two m.2 sata and one m.2 nvme (uses pcie for power for all 3 drives) and had to put it in the second x16 socket. No big deal but would have been nice to space it differently.

    I’m open to getting a TUF board again for my next rebuild.

    The Q-Connector was missing from the box but since my last board was Asus I just used the old one.

    The new system has been running great, boots super fast – even better (much faster) then the ROG Crosshair Hero VI x370 with an 1800x.

    I used to use Ryzen Master to check temps from time to time but that is not compatible with this system so I am looking for a new tool to use.

  2. One point I didnt see and may have missed in the review is the two m.2 slots are not equal. The one with the heat sink is Pcie Gen 4 and on the main Pcie bus, the other m.2 is Gen 3 and connected to the chipset. If you have a Gen 4 NVME drive this could be important.

    You didn’t see that in the review, because that’s not true. Both M.2 slots are PCIe Gen 4.0 compliant. With X570 boards, one M.2 slot is always direct to the CPU’s PCIe controller, while the other is always via the PCH. In the case of X570, all PCIe slots are gen 4.0 compliant. This of course only applies to Ryzen 3000 series CPU’s without an integrated APU. The ones with integrated APU’s, are actually Zen+, and do not support PCIe 4.0.

    X570 is PCIe Gen 4.0 all the way. B550 isn’t.

  3. You didn’t see that in the review, because that’s not true. Both M.2 slots are PCIe Gen 4.0 compliant. With X570 boards, one M.2 slot is always direct to the CPU’s PCIe controller, while the other is always via the PCH. In the case of X570, all PCIe slots are gen 4.0 compliant. This of course only applies to Ryzen 3000 series CPU’s without an integrated APU. The ones with integrated APU’s, are actually Zen+, and do not support PCIe 4.0.

    X570 is PCIe Gen 4.0 all the way. B550 isn’t.

    I wonder if my book is old then? Or if I am getting the various models I looked at confused. Both being Gen 4 would be nice.

  4. I wonder if my book is old then? Or if I am getting the various models I looked at confused. Both being Gen 4 would be nice.

    You may have misread it given how the specifications are written. The slot is only Gen 4.0 compliant with a Ryzen 3000 series CPU. Ryzen 1000
    CPU’s, Ryzen 2000 CPU’s, APU’s, and Ryzen 3000 series APU’s do not support PCI-Express 4.0. Therefore, the M.2 slot would only operate at Gen 3.0 speeds. The way this is written in the specifications can be kind of confusing.

    All X570 motherboards are fully PCI-Express 4.0 compliant when using the right CPU. If it was a B550 chipset, only the M.2 slot that connects directly to the CPU would be PCIe 4.0 compliant. The secondary M.2 slot connected to the PCH would be Gen 3.0.

  5. How does the bandwidth compare with identical pcie gen 4 nvme drives from one socket to the other – I wonder… I had wondered if the one socket got a heat shroud because it was expected to stress the drive more.
  6. How does the bandwidth compare with identical pcie gen 4 nvme drives from one socket to the other – I wonder… I had wondered if the one socket got a heat shroud because it was expected to stress the drive more.

    I’ve tested this a number of times. There is no difference between the two slots. The reason why one slot gets the heat sink and the other doesn’t is simply cost cutting. The ASUS TUF GAMING X570 Plus WiFi is a budget offering. Budget offerings cut corners to hit those lower price points. Not reinforcing all the expansion slots, not providing heat sinks for every M.2 slot, weaker VRM’s, etc. are all methods used to do that.

  7. I could pick up a one or a pair of these: https://www.amazon.com/heatsink-double-sided-matching-thermal-silicone/dp/B07KDDKDNN or maybe one that lights up to match my Dimms – Any FPS reviews for m.2 cooling options?

    Gonna be a few weeks before I am up for installing though – could ask my niece over again and cook her dinner.

    On a second TUF conversion issues… I am checking out CPUID Hardware Monitor Pro for monitoring thermals — any feedback? Is there something better? As of right now my Ryzen temps are Value: 44.3 Min: 40.3 Max: 70.5 – Max was prob while playing Black Desert Online.

  8. Excellent review, thanks. I’ve built 3 systems with this board, pretty rock-solid and stable, and easy to work with. Indeed though, it’s not the best board for overclocking. "This is a motherboard best paired with 8c/16t CPUs or a 12c/24t CPU at stock in my opinion." I agree. And yeah, CPU boost leaves something to be desired, compared to other boards.

    "… although it’s not a 12+2 phase solution like ASUS says it is. This is a 4×3 phase solution that uses three power stages per phase. I dislike ASUS’ recent trend towards its marketing practices concerning its VRM’s. This was a company that used to offer the same VRM’s down most of the line at one point (years ago) and has built a lot of credibility with the enthusiast community, and they are taking advantage of that." Sure seems like it, and yeah I didn’t know that **** about the phase solution being different than what ASUS claimed it was. Asshats.

  9. Excellent review, thanks. I’ve built 3 systems with this board, pretty rock-solid and stable, and easy to work with. Indeed though, it’s not the best board for overclocking. "This is a motherboard best paired with 8c/16t CPUs or a 12c/24t CPU at stock in my opinion." I agree. And yeah, CPU boost leaves something to be desired, compared to other boards.

    "… although it’s not a 12+2 phase solution like ASUS says it is. This is a 4×3 phase solution that uses three power stages per phase. I dislike ASUS’ recent trend towards its marketing practices concerning its VRM’s. This was a company that used to offer the same VRM’s down most of the line at one point (years ago) and has built a lot of credibility with the enthusiast community, and they are taking advantage of that." Sure seems like it, and yeah I didn’t know that **** about the phase solution being different than what ASUS claimed it was. Asshats.

    Had ASUS simply been honest about the phase count, I wouldn’t give them so much crap for the design. It’s obviously capable enough for most people, but sometimes people will pick up the box and look at the back when shopping at their local Microcenter and think they are getting a monster overclocking board for $200 and that’s just not the case.

  10. Had ASUS simply been honest about the phase count, I wouldn’t give them so much crap for the design. It’s obviously capable enough for most people, but sometimes people will pick up the box and look at the back when shopping at their local Microcenter and think they are getting a monster overclocking board for $200 and that’s just not the case.

    In the more recent past, my motherboard decision has come down to deciding to get a board with 10Gbit built in, and I assume that that’s pushed me toward higher-end boards. I haven’t really done much ‘VRM shopping’, or shopping for memory topology, so I was absolutely impressed to find those details mentioned in the review!

    And now that more PCIe lanes are being included in CPUs and Chipsets, I probably won’t have to deal with the limitations that pushed me toward getting the feature built-in; which is just as good since the boards with that feature are starting to get a big extreme.

    [also, your pictures looked pretty good sir ;) ]

  11. In the more recent past, my motherboard decision has come down to deciding to get a board with 10Gbit built in, and I assume that that’s pushed me toward higher-end boards. I haven’t really done much ‘VRM shopping’, or shopping for memory topology, so I was absolutely impressed to find those details mentioned in the review!

    And now that more PCIe lanes are being included in CPUs and Chipsets, I probably won’t have to deal with the limitations that pushed me toward getting the feature built-in; which is just as good since the boards with that feature are starting to get a big extreme.

    [also, your pictures looked pretty good sir ;) ]

    Getting a board with 10GbE controllers built into it typically pushes you towards HEDT offerings or super high end motherboards like the MSI MEG X570 GODLIKE which come with separate 10GbE adapters. There are workstation oriented motherboards, or offerings that are targeted more towards the content creator which also may offer 10GbE controllers. As for VRM’s, its complicated. On one hand, lower end motherboards tend to cheap out on them but usually have VRM’s that are good enough for most CPU’s assuming they are being run at stock clocks. That includes boost clocking, although the CPU may not boost quite as well on cheaper motherboards.

    The only real reason to place a ton of focus on VRM capability is when you are overclocking. Even then, unless its a super cut rate VRM solution, overclocking is still on the table. Where you really run into issues is taking something like MSI’s X570-A Pro, which has notoriously bad VRMs and overclocking a Ryzen 9 3950X to an all core clock of 4.2GHz+. If you try that, they are going to get hot and while they may handle it for a time, running it hard like that may shorten the life of the board. You will also fight much harder to achieve overclocks on boards with worse VRM’s as those boards have additional hardware constraints that make overclocking more difficult.

    Generally, if I can achieve a specific overclock on one motherboard, I can achieve the same overclock on 90% of them. However, getting to that clock speed with stability may be considerably more difficult. The board in question here, required more work than I am used to putting in with this specific test CPU. On my MSI MEG X570 GODLIKE, it takes almost nothing to get it to overclock. All I have to do is set my vCore and my frequency multiplier and that’s really it. Sometimes load-line calibration needs to be adjusted as well.

    Now, a lower end VRM implementation also gets you by on air or watercooling. When LN2 comes into play, you can basically forget a board like the ASUS TUF GAMING X570 Plus [WiFi]. It would probably **** sparks and die if you tried to push a 3950X over 5.0GHz or something like that.

    Memory topology is really something that’s a bigger concern on the AMD side of things. On Intel motherboards, its less of an issue as Intel systems can generally clock RAM higher and do so with four modules fairly easily. The same rules apply to both, but they come into play at lower clock speeds on the AMD side. With Intel motherboards, its rarely a concern until you are well on your way to DDR4 5000MHz speeds. So, upwards of 4133MHz or so you are probably fine even using four modules. On X570, that’s not going to happen on a daisy chain motherboard using four DIMMs. Conversely, while you can use four modules more easily on a T-Topology board, your clock speeds will be more limited.

  12. "The ASUS TUF Gaming X570 Plus WiFi supports the following PCIe lane configurations: x16/x0 or x8/x8."

    The second PCIe 16x slot is only x4 capable and is run from the PCH. So x8/x8 is not possible on this board. From your image you can see it is wired up as x4.

    View attachment 716

    Good review and was utterly surprised it is an actual 4 phase VRM design, bummer. Seems rather positive from user reviews in general. Anyways may end up buying a Cheap X570 or B550 board. MSI Tomahawk X570 would be my first choice if I could find one at MSRP.

  13. Getting a board with 10GbE controllers built into it typically pushes you towards HEDT offerings or super high end motherboards like the MSI MEG X570 GODLIKE which come with separate 10GbE adapters.

    Coming back to this for a couple of reasons (you can blame @noko too).
    First, the Aquantia NICs are basically not worth it. I’d take a Realtek crab-LAN controller if they made one in 10Gbit over these (which they don’t… yet). Instead, I bay’ed an X550-T2 to toss in the bottom PCIe 3.0 x4 slot. Definitely a crap-shoot with used parts, but it can pay off too, and this one worked.

    Main complaints with the Aquantias are that they like to decide that the network cable that is clearly connected suddenly isn’t, and require a disable / enable cycle in the OS to ‘find’ the cable again. Side complaint is that I like TrueNAS (nee FreeNAS) and the stable BSD-based branch of that distro does not yet have the Aquantia driver built in yet, and won’t, until they rebase (if ever!) on FreeBSD 13, which I believe hasn’t actually been released yet.

    The used Intel NIC seems to work fine so far. I got an X540-T2 for the fileserver since it is sharing the main PCIe x16 allocation from the CPU with an SAS controller which is PCIe 2.0, since the prices on used PCIe 3.0 controllers are pretty high still and they provide no advantage at my level with SATA drives.

    So in summary… I’ve backed off of NICs as being a requirement for my desktop boards. At this point I pretty much feel that fan and RGB control are top of the list after having the basics covered, ideally just a solid VRM and non-retarded slot layout, as I’ll carry my Intel NICs forward.

    On the other hand, I’d really like the option to add more RAM easily, and I am starting to warm up to the utility of having more PCIe lanes, so I’m now eyeing HEDT :D

  14. Surprised I missed this one as I have always like motherboard reviews.

    Kinda sucks the TUF name is now kind of a budget entry. I bought a 990FX TUF Sabertooth back in the day and that thing was indestructible!! I absolutely hammered a Phenom II X6 1080 at 3.9, then beat the balls of a FX 8150 at 4.3 and finally a 8350 at 4.8! That was when I was most into overclocking and I ran OCCT and Prime95 a zillion times. That board never so much as whimpered.

    That’s what I like about them… They didn’t have a bunch of fancy features, just bulletproof build quality. And they weren’t a budget alternative to the Crosshair boards.

  15. Surprised I missed this one as I have always like motherboard reviews.

    Kinda sucks the TUF name is now kind of a budget entry. I bought a 990FX TUF Sabertooth back in the day and that thing was indestructible!! I absolutely hammered a Phenom II X6 1080 at 3.9, then beat the balls of a FX 8150 at 4.3 and finally a 8350 at 4.8! That was when I was most into overckocking and I ran OCCT and Prime95 a zillion times. That board never so much as whimpered.

    That’s what I like about them… They didn’t have a bunch of fancy features, just bulletproof build quality. And they weren’t a budget alternative to the Crosshair boards.

    Owned one also, pushing way over 250w out of the FX9590. This board does not seem remotely in the same class as that Sabertooth, which I ended throwing away, still good but just taking up space never to be used again most likely.

  16. Coming back to this for a couple of reasons (you can blame @noko too).
    First, the Aquantia NICs are basically not worth it. I’d take a Realtek crab-LAN controller if they made one in 10Gbit over these (which they don’t… yet). Instead, I bay’ed an X550-T2 to toss in the bottom PCIe 3.0 x4 slot. Definitely a crap-shoot with used parts, but it can pay off too, and this one worked.

    Main complaints with the Aquantias are that they like to decide that the network cable that is clearly connected suddenly isn’t, and require a disable / enable cycle in the OS to ‘find’ the cable again. Side complaint is that I like TrueNAS (nee FreeNAS) and the stable BSD-based branch of that distro does not yet have the Aquantia driver built in yet, and won’t, until they rebase (if ever!) on FreeBSD 13, which I believe hasn’t actually been released yet.

    The used Intel NIC seems to work fine so far. I got an X540-T2 for the fileserver since it is sharing the main PCIe x16 allocation from the CPU with an SAS controller which is PCIe 2.0, since the prices on used PCIe 3.0 controllers are pretty high still and they provide no advantage at my level with SATA drives.

    So in summary… I’ve backed off of NICs as being a requirement for my desktop boards. At this point I pretty much feel that fan and RGB control are top of the list after having the basics covered, ideally just a solid VRM and non-retarded slot layout, as I’ll carry my Intel NICs forward.

    On the other hand, I’d really like the option to add more RAM easily, and I am starting to warm up to the utility of having more PCIe lanes, so I’m now eyeing HEDT :D

    I’ve tested a bunch of the Aquantia NICs and I’ve never had any such problems with them. The problem I have with the Intel NIC’s such as my X540-T2 is that they will not negotiate down to 5GbE or 2.5GbE like the Aquantia NICs do.

    Surprised I missed this one as I have always like motherboard reviews.

    Kinda sucks the TUF name is now kind of a budget entry. I bought a 990FX TUF Sabertooth back in the day and that thing was indestructible!! I absolutely hammered a Phenom II X6 1080 at 3.9, then beat the balls of a FX 8150 at 4.3 and finally a 8350 at 4.8! That was when I was most into overckocking and I ran OCCT and Prime95 a zillion times. That board never so much as whimpered.

    That’s what I like about them… They didn’t have a bunch of fancy features, just bulletproof build quality. And they weren’t a budget alternative to the Crosshair boards.

    It does. I don’t care for the new positioning of the brand, but it was pretty redundant the second they introduced multiple ROG offerings for each chipset and form factor.

  17. I’ve tested a bunch of the Aquantia NICs and I’ve never had any such problems with them. The problem I have with the Intel NIC’s such as my X540-T2 is that they will not negotiate down to 5GbE or 2.5GbE like the Aquantia NICs do.

    I’ve had it happen with ones built into boards as well as a discrete one; I guess at that point the problem could be the switch?

    Dunno. Having a 10Gbit switch, I haven’t really worried about 2.5Gbit and 5Gbit, though I believe that the switch can do those too.

    I’m just happy that I have an uninterrupted 10Gbit connection now. And I keep a 1Gbit line connected alongside just in case those act up again.

Leave a comment