At what temperatures will the VRMs on the VIII and IX Hero cause throttling?

[Noirgheos]

I currently have a 6700K at stock (removed the OC for now), and an NH-U12S cooling it.

Attached are my temperatures after 15 minutes of running around Novigrad in Witcher 3. I'd like to overclock, but I've heard that VRM throttling can cause stutters in games, and I have been experiencing some odd ones. Do my VRM temperatures look good? 

[Deli]

Highend boards have overkill VRMs. As long as this is a little airflow over the VRM heatsink. It will never reach the point of throttling.

[weberdarren97]

I don't think any of those temperatures are your VRMs, it would be labeled something like VRT (maybe a number after t if there's multiple sensors). It's not uncommon for motherboards to not report VRM temperatures to the operating system. Most UEFI BIOS keep that information for themselves, which means you'll never see the temps of the VRMs during a stress test.

 

Anyways, it's not something you need to worry about. You're good until about 130C which would require you to take the system fans out of the case, seal the case shut, replace the CPU air cooler with an AiO, and put the radiator outside the case where it can still keep the CPU cool. Basically, you'd have to completely starve the area around the socket from getting any airflow at all and it would still take almost an hour to get to dangerously high VRM temps.

[TheRandomness]

13 minutes ago, weberdarren97 said:

I don't think any of those temperatures are your VRMs, it would be labeled something like VRT (maybe a number after t if there's multiple sensors). It's not uncommon for motherboards to not report VRM temperatures to the operating system. Most UEFI BIOS keep that information for themselves, which means you'll never see the temps of the VRMs during a stress test.

 

Anyways, it's not something you need to worry about. You're good until about 130C which would require you to take the system fans out of the case, seal the case shut, replace the CPU air cooler with an AiO, and put the radiator outside the case where it can still keep the CPU cool. Basically, you'd have to completely starve the area around the socket from getting any airflow at all and it would still take almost an hour to get to dangerously high VRM temps.

I'm fairly certain VRMs start throttling at around 90 Celsius before they just give up at 130 Celsius... according to my 290, anyway

[Noirgheos]

20 minutes ago, weberdarren97 said:

I don't think any of those temperatures are your VRMs, it would be labeled something like VRT (maybe a number after t if there's multiple sensors). It's not uncommon for motherboards to not report VRM temperatures to the operating system. Most UEFI BIOS keep that information for themselves, which means you'll never see the temps of the VRMs during a stress test.

 

Anyways, it's not something you need to worry about. You're good until about 130C which would require you to take the system fans out of the case, seal the case shut, replace the CPU air cooler with an AiO, and put the radiator outside the case where it can still keep the CPU cool. Basically, you'd have to completely starve the area around the socket from getting any airflow at all and it would still take almost an hour to get to dangerously high VRM temps.

Hmm, I guess I should do my first custom loop then.

[weberdarren97]

8 minutes ago, TheRandomness said:

I'm fairly certain VRMs start throttling at around 90 Celsius before they just give up at 130 Celsius... according to my 290, anyway

Is your (I'm assuming R9 290) a custom PCB with aftermarket VRMs on it? If not, it's MOSFETs are not comparable to such an extreme VRM config like what's seen on ridiculously high end Asus boards. Then there's that one Gigabyte board with 32 power phases on it (not kidding), which seems to have proven that any more than 10 doesn't make much of a difference.

[weberdarren97]

The Hero boards from Asus are ridiculous. In my opinion, they're a waste of your money. The only way to really get any benefit out of the extremely clean power that the VRMs are capable of supplying is to win the silicon lottery and end up with a perfect CPU. We all know there's no such thing as a perfect CPU because they're made so quickly that each of them carries a unique imperfection. The chances of getting a better or worse chip is known as the silicon lottery, and the fact is that the chances of winning are less than the chances of winning the actual lottery here in the United States. Ending up with a CPU that doesn't overclock too well isn't something you can just tell the seller to avoid.

[weberdarren97]

5 minutes ago, Noirgheos said:

Hmm, I guess I should do my first custom loop then.

Even if you delid the CPU, use the best block available on the market and use the largest most efficient radiators you can find, you'll still hit the thermal limits of the CPU before you max out that mobo's VRMs.

[TheRandomness]

2 minutes ago, weberdarren97 said:

Is your (I'm assuming R9 290) a custom PCB with aftermarket VRMs on it? If not, it's MOSFETs are not comparable to such an extreme VRM config like what's seen on ridiculously high end Asus boards. Then there's that one Gigabyte board with 32 power phases on it (not kidding), which seems to have proven that any more than 10 doesn't make much of a difference.

Consider the fact that the 5 core phases of the reference 290 can output around 500 amps at 90 Celsius. Also, any amount of phases above 8 uses doublers to get the PWM signals for all the other phases. 

[Noirgheos]

1 minute ago, weberdarren97 said:

Even if you delid the CPU, use the best block available on the market and use the largest most efficient radiators you can find, you'll still hit the thermal limits of the CPU before you max out that mobo's VRMs.

Yep, but I've also heard that with Skylake, once you hit 64C, thermal things will begin to come into play. I hit 69C-73C while gaming, as you can see.

[weberdarren97]

25 minutes ago, Noirgheos said:

Yep, but I've also heard that with Skylake, once you hit 64C, thermal things will begin to come into play. I hit 69C-73C while gaming, as you can see.

I believe you're talking about the temperatures of the CPU itself, not components on the motherboard. Asus wouldn't charge $200+ for a board that can barely run an i7 at stock speeds.

[Noirgheos]

1 minute ago, weberdarren97 said:

I believe you're talking about the temperatures of the CPU itself, not components on the motherboard. Asus wouldn't charge $200+ for a board that can barely run an i7 at stock speeds.

Yep, you guys have convinced me that its probably not my motherboard components. Just the CPU itself. I've heard that Skylake is more aggressive when it comes to thermal limits.

[weberdarren97]

1 minute ago, Noirgheos said:

I've heard that Skylake is more aggressive when it comes to thermal limits.

Actually, it's the modern UEFI BIOS that companies are putting on their boards that are posing this restriction, not the CPU itself. Somewhere in BIOS you should see a setting that's named something along the lines of CPU Thermal Target. If the CPU goes x degrees over that target, the board will begin to tone down the OC. In many Asus boards, x is somewhere around 20C. That doesn't mean that you can set the Thermal Target to 90C and then allow the CPU to go to 110C before it throttles, it will still throttle at just under 100C. That is a fail safe feature, it cannot be viewed or changed.

[Noirgheos]

3 minutes ago, weberdarren97 said:

Actually, it's the modern UEFI BIOS that companies are putting on their boards that are posing this restriction, not the CPU itself. Somewhere in BIOS you should see a setting that's named something along the lines of CPU Thermal Target. If the CPU goes x degrees over that target, the board will begin to tone down the OC. In many Asus boards, x is somewhere around 20C. That doesn't mean that you can set the Thermal Target to 90C and then allow the CPU to go to 110C before it throttles, it will still throttle at just under 100C. That is a fail safe feature, it cannot be viewed or changed.

Oh, alright. Just odd that that would be around 20C. Don't know many people who can keep it below that. I understand its incremental, but still. That's a little extreme. Setting it to start... playing with clocks around 60C is good? I mean if I pass 60C I shouldn't see anything out of the ordinary right?
 

Just odd, my clocks stay at 4.2GHz even when going towards the high 60s with my default BIOS.