Do Not Use Z490 Motherboard Auto Settings: Excessive Voltage, Power, & Heat

Z490 motherboards for Intel's 10th Gen CPUs, including the i9-10900K & 10600K are all over the place for Vcore, power limits, and Turbo duration. MCE is back in a worse way than last time.

In this video, we're talking about MCE on Z490 motherboards. Motherboards are all over the map for auto Vcore, some blasting way by what's necessary to sustain an Intel Core i9-10900K at the functionally pre-overclocked frequencies that are applied. A lot of this comes down to semantics and down to Intel not strongly enforcing its "official guidance" as "official spec," leading to a discrepancy in testing procedures and in fairness between board manufacturers. The excessive voltage, excessive heat, and excessive power don't fit the advertised specifications of the CPU, and at best, are lazy settings by some board makers. They're not all like this, but in this content, we're looking at the "best" Z490 motherboards from ASRock, ASUS, MSI, and Gigabyte. That includes the ASUS Z490 Maximus XII Extreme, the Gigabyte Z490 Master, the MSI Z490 MEG Ace, and ASRock Z490 Velocita. If your Intel 10th Gen CPU is overheating or too hot, like the i9-10900K or 10600K, it's probably because of the default motherboard settings.

I watched the video earlier and for me the biggest problem I can see is when someone isn’t an Overclocker and doesn’t spend much time in the bios. I don’t think this will apply so much to people who buy the 10900K as for the most part those enthusiasts will know what to do.

However for someone dropping in a 10600k or 10700K and just expecting it to work there could be problems. Especially if you buy a certain cooler and expect it to work. If intel advertises that the cpu is a certain tdp and you buy a cooler to go with it and then the motherboard you pair it with basically overclocks it for you, then there could be issues.

Some boards are boosting every core to 5.3 instead of just cores 8 or 9 as per intels recommendation. Other boards are boosting way longer than other boards. Which will build up heat.

This isn’t a problem you will really notice if you have a 360mm Aio, but anyone on a cheaper cooler may run into issues that are not of intels or their own doing.

Sounds like the whole 10th gen Intel/Z490 release is rushed and premature. I imagine there will be a flurry of bios updates in the near future.

I don't get this. The behavior has been in motherboards for generations now.

Some boards are boosting every core to 5.3 instead of just cores 8 or 9 as per intels recommendation. Other boards are boosting way longer than other boards. Which will build up heat.

Click to expand...

Oh wow. I haven't run across this yet (5 board reviews). The boards I've looked at will hold the all core turbo to 4.9 GHz (4.8 GHz is stock, TVB is the +100 MHz).

I also question the details of specific cores boosting... I thought these were intelligent enough to boost the best core(s) to that speed, not sit on specific ones on each and every chip.

That said, I cant say these was rushed or premature considering they have been doing this for a while now. They may update the BIOS and have one that conforms to intel spec (if the option isn't available aleady)... but, considering it has been this way for a while, I wouldn't bet on it.

Feels like GN missed the boat a bit on this one, honestly.

I somewhat agree. I don't have any hands on experience with these boards or CPUs, or previous generations for that matter, as a caveat. However as I see things, the big difference between AMD and Intel's approach is overclocking. If the Intel customer want's something that is predictable and easy to power and cool, then they purchase a non "K" model and it is fully locked down from the factory. If they want to go a bit wild, they purchase a K model and the board partners have some leeway. Since every AMD CPU is overclockable, AMD has to regulate the motherboard power delivery more strictly in order to ensure a mostly functional out of the box experience. There probably is a group however that falls into that rift that knows enough to get a "K" CPU but not enough to adjust the BIOS to their preferences. I see this more with maybe sloppy system integrators, but I imagine some home builders as well. I just don't think it's the "majority" of Intel's market here. Although if boards are running with instability out of the box that can be a problem.

I can see from the GN perspective, where they review chips at both stock and a manual OC, that this is a major headache to their process.

EDIT: I was going back through the review on our page to see what board it used, and I saw those power numbers again. I thought about all the people that come here wondering why their H212 can't keep their 9900k cool. Wonder if we will see people coming in with a 300W out of the box 10900k under a 212 asking what is wrong. Maybe Steve has a point after all.

EarthDog said:

I also question the details of specific cores boosting... I thought these were intelligent enough to boost the best core(s) to that speed, not sit on specific ones on each and every chip.

That said, I cant say these was rushed or premature considering they have been doing this for a while now. They may update the BIOS and have one that conforms to intel spec (if the option isn't available aleady)... but, considering it has been this way for a while, I wouldn't bet on it.

Feels like GN missed the boat a bit on this one, honestly.

Click to expand...

I am guessing a little here. But I think it is Intels recommendation to boost cores 8 or 9 as these may be furthest away from the controllers etc. As with the K series they have reduced the silicone thickness. Maybe this could cause heat issues if core 0 is boosting all the time, I am not sure at all why, but intel must have some reason to state cores 8 or 9.

I don’t think the boards or bios were rushed at all, this is just different board makers either interpreting the rules differently or wanting to get ahead of the competition. GN said in the video that they need something to stand out from each other as they all pretty much have the same features.

Links instead of thinks, btd?

Now that i think about it, according to Hwinfo, I've seen all cores (eventually, not all at once) hit 5.3 Ghz.




In the tests I've run so far (ASRock Velocita/Taichi, MXIIHero, Aorus Master, MSI MEG Ace), these boards/cpu at stock speeds using AIDA64 stress test pulled less than 200W stock according to hwinfo. The whole system creept up towards 300W...

As far as review methods...thats how I do it too. Out of the box results and a manual overclock. I feel this makes the most sense because that is how users will use it, right? Either you get a box, enable xmp and go, or manually overclock. Performance can vary like this and specifically why I test this way (for years...here, and Anand and Toms).

Edit: I see.. 5.3 ghz on some tests. I've only watched clocks on a stress test for the most part.

Edit2: I see the core 8/9 thing too now, btd.

bigtallanddopey said:

I am guessing a little here. But I think it is Intels recommendation to boost cores 8 or 9 as these may be furthest away from the controllers etc. As with the K series they have reduced the silicone thickness. Maybe this could cause heat issues if core 0 is boosting all the time, I am not sure at all why, but intel must have some reason to state cores 8 or 9.

I don’t think the boards or bios were rushed at all, this is just different board makers either interpreting the rules differently or wanting to get ahead of the competition. GN said in the video that they need something to stand out from each other as they all pretty much have the same features.

Click to expand...

This is what I meant by "rushed". Making bios configuration decisions that were not tested thoroughly or taking time to gain clarity on "the rules". All this is part of "trying to get ahead of the competition."

trents said:

This is what I meant by "rushed". Making bios configuration decisions that were not tested thoroughly or taking time to gain clarity on "the rules". All this is part of "trying to get ahead of the competition."

Click to expand...

Thats the thing. They know the 'rules'. They are provided the parameters so they can create the boards. These were tested just as any other generation has been and the board partners are well aware of how it works. This is intentional and again, not the first generation this happened with.

This go around, the Asus boards (Hero and Extreme at least) pop up two options in the bios (after CMOS reset or h/w change, so first boot). First one (pressing F1) adheres to stock parameters, the other (F3) unlocks things.

The other boards are already loosey goosey out of the box/optimized defaults.

I'm on my 3rd board right now including mentioned ASRock Velocita and I haven't seen mentioned issues. I'm not saying that everything in this video is wrong but after watching videos like that, people are starting to share BS around the web. Maybe they tested initial BIOS releases as they clearly had to test mobos before the premiere, I don't know as it's not described. Every brand released 1-2 new BIOS after 20 May. For ASRock Velocita, there were 2 new BIOS and were fixing some little issues but I haven't seen any power and voltage problems even at the initial one which was already marked as 1.20. I've seen better memory support using 1.30 or 1.40 BIOS.

I dislike videos and tests like that because these are early products and often not the same as can be purchased in retail stores. We can already read around the web how bad is 10900K (heat, wattage etc.) and how bad are Z490 motherboards ... from people who had no chance to even test it, just post random thoughts about new products based on "news".
We could see "up to 300W+ from a stock 10900K" news 3-4 weeks ago. Somehow, I haven't seen more than ~246W under full load so not much more than the 9900K has. I still can read on various websites and forums that 10900K has 300W+ at stock settings... based on what?
We could see that "all Z490 motherboards have faulty LAN" ( i225) but a big part of motherboards have older LAN (219v) or use it with Realtek/Aquantia.

Some motherboard series, including ASRock, have auto OC, bclk boost and other features included. It may cause clock and voltage differences. Also, not every motherboard works the same or has the same power states. We could see that in the previous series too so it's not a big discovery. Somehow videos like that will cause that we will find many forum threads about it and "global panic".

I do concur with the motherboards and I have only used the one but didn't appear to me it was overvolting. Now when it comes to power usage that really depends on the test. When I ran Prime95 small FFTs with MCE on (4.9 all-core) then system usage was up to 370+ W. So saying it's a 300 W chip really isn't that far off IMO.

I haven't seen more than ~240W at auto, unless I set manually 1.42V+ on my CPU. Max I think was something near 300W but at manual OC, not auto.

Btw. afaik power limit at stock is 250W so as long as the motherboard uses default Intel settings then it shouldn't pass 250W. If it goes up to 300W then motherboard has unlocked power limit and clearly bumps voltages. This can be the issue mentioned in the video.


This is 247W in AIDA64, 100% load at all cores, CPU+FPU 5.1GHz AVX2 with manual 1.37V in BIOS. Unlocked power limit to 4096W.
Screenshot during the test on ASRock Z490 PG Velocita with 1.40 BIOS.