Understanding the Variables.....

[NoddyPirate]

Hi all,

I am finding myself fascinated by the whole area of overclocking - and while I don't really have any intention to do it myself, I would really appreciate any insights and experience to help me better understand all the bits and pieces you need to ponder. Assuming it is OK with the Mods to ask here....

One area in particular I would love some info on is VRM's and Chipset temperatures. I have an ASUS PRIME B550 PLUS in my build, which has big, but passive, heatsinks on the VRM and also the Chipset. I see that the ROS STRIX 570 boards have heat pipes as well in the VRM sinks and active cooling on the Chipset.

A few Q's!!

What kind of temperatures do you typically see on VRMS and Chipsets when overclocking and how do they compare to a stock set up?

I imagine a board like my B550 could only handle very mild overclocking because the VRM and Chipset cooling is limited?

Do Motherboards protect themselves from overtemps in VRMs and Chipsets? i.e. Would your CPU be throttled if they get too hot?

It strikes me that stock downdraft coolers do provide some cooling airflow for the VRM and chipset. Can you have issues with aftermarket coolers - particularly AIO's - because the VRM's and Chipsets actually run hotter?

Thanks in advance for furthering my education!!!

 

[Scott]

There are other variables to consider:

What is your usage? If it's primarily multi-threaded application then overclocking may be beneficial, if it's primarily single threaded (ie gaming) then overclocking is a waste of time as all chips boost better on a single core than they ever would overclocked.

What is the chip? The overclocking ability and VRM usage is heavily impacted, and directly related, to the core count. Newer chips also overclock FAR better than older chips do. The less cores you have, the more OC you can apply.

Most people won't monitor the VRM temperature so I think you are unlikely to see much information in this regard. When I was doing my OC i did check out my VRM temperature and IIRC it was over 100°C when being hammered to death by my stress testing, but otherwise I don't really recall or monitor. I have no idea what the temperatures were at stock levels either, I put trust in the system to throttle for any issues (which I believe it does quite readily).

B550 Plus isn't geared for heavy overclocking. The actual physical size of the VRM chips has little impact, it's more the quality and cooling ability that comes into play. That being said, the overall size obviously helps with the heat transfer so being bigger will help.... it's just not an indicator of ability.

Generally I would recommend the B550 Plus for gaming rigs up to and including the 5800X. For more intensive applications the X570 would always be favoured as it has more capacity to help sustain the voltages required for sustained boosting. Overclocking on a B550 won't really net you much on anything over a 5600X, for anything but brief bursts of operation (pub talk). The Strix and Hero options are where the workloads get serious, and the processor core counts get serious as well. These boards are geared up to heavy usage for any of the chips but if you want to OC the 5900X or the 5950X I don't think there is any option other than the HERO if you want it to run properly with sustained loads.

No matter which route you go down, the motherboard and the chip will keep itself pretty safe by throttling. If things get dicy, it'll hard shutdown on you to protect itself. This isn't always fool proof but it lets you know when you have gotten things very wrong and, hopefully, gives you the chance to correct this error.

Not sure what you mean with cooling. Air coolers circulate the air inside the case and increase the temperature of the airflow. You need to ensure the other airflow contributors (intake fans and exhaust fans) are well proportioned and directed well in order to keep the internal temperatures down. AIOs actually contribute massively to the airflow through the case, thus keeping the temperatures inside the case lower than they would be otherwise. If you use the AIO as an exhaust, the temperature delta that would increase from the CPU is eliminated altogether, but thanks to the way AIOs operate this actually doesn't make as much of a difference as you would think. Airflow doesn't matter to components, it's not like windchill. What does matter is the temperature of the air surrounding the components, which is ironically where airflow does matter. If the air surrounding the components is lower with an AIO than with a fan then they are going to be cooler regardless of how fast that air is passing over them.

 

[NoddyPirate]

That's brilliant Scott thanks so much for taking the time to explain!

I am shocked at VRM's getting over 100 degrees but it shows just how hard they can be working....

In terms of cooling - I have have read a lot that says one side effect of upgrading to a tower cooler from a stock cooler is that you lose the downdraft airflow across the motherboard components in the vicinity of the CPU - which can increase the temperature of the VRM and chipset. With an AIO you don't have any airflow around the CPU of any sort - so I was curious to see what impact they might have. Without overclocking of course who cares really I suppose!!

But I think you have answered my question perfectly anyway - the whole set up must consider airflow through the entire case. Just sticking an AIO into a case with rubbish ariflow characteristics won't necessarily keep everything under control? Using an AIO as an exhaust seems sensible to me. Using it as an intake would surely dump every bit of heat removed from the CPU - plus the little bit heat of heat generated by the AIO pump and fans themselves - back into the case - so if you don't exhaust that effectively your case temperature would increase too?

Where I was coming from is - yes components don't get windchill but they need the air to moving past them to effectively remove the heat from their heatsinks. Static airflow will not cool components anywhere even remotely close to as well as a dynamic or moving flow.

Anyway, brilliant reply Scott thank you again! More to think about!! The learning continues......

 

[NoddyPirate]

One more Q:

What causes the Chipset to get toasty? Many or most boards these days seem to provide a direct data transfer between the CPU and the first M.2 drive and one or more PCIex16 slot(s). So the GPU and first storage device bypass the Chipset. Would it only be the case that if you are heavily using peripherals or storage devices that run through the Chipset that it gets a heavy load?

 

[Scott]

That's brilliant Scott thanks so much for taking the time to explain!

I am shocked at VRM's getting over 100 degrees but it shows just how hard they can be working....

In terms of cooling - I have have read a lot that says one side effect of upgrading to a tower cooler from a stock cooler is that you lose the downdraft airflow across the motherboard components in the vicinity of the CPU - which can increase the temperature of the VRM and chipset. With an AIO you don't have any airflow around the CPU of any sort - so I was curious to see what impact they might have. Without overclocking of course who cares really I suppose!!

But I think you have answered my question perfectly anyway - the whole set up must consider airflow through the entire case. Just sticking an AIO into a case with rubbish ariflow characteristics won't necessarily keep everything under control? Using an AIO as an exhaust seems sensible to me. Using it as an intake would surely dump every bit of heat removed from the CPU - plus the little bit heat of heat generated by the AIO pump and fans themselves - back into the case - so if you don't exhaust that effectively your case temperature would increase too?

Where I was coming from is - yes components don't get windchill but they need the air to moving past them to effectively remove the heat from their heatsinks. Static airflow will not cool components anywhere even remotely close to as well as a dynamic or moving flow.

Anyway, brilliant reply Scott thank you again! More to think about!! The learning continues......

I would be careful to what you take as gospel when it comes to reading links etc. A popular opinion can often be suggested as fact when it comes to forums and the likes. What is proven for setup A, cannot be taken as fact in setup B. This is the most common rabbit hole with information on the interweb. A respected journalist/tester carries out a test on a bench rig and all of a sudden the results extrapolated are generalised across the entire net. If only things were that simple

Disregarding the cooler entirely, the temperature of the air flowing over the chipset is all that matters. Moving to a larger towered cooler could potentially impact the flow of air over the VRM, but what would be more important would be the temperature of that air. The temperature of the air is more impacted with the airflow through the case than it ever will be by the cooler on the CPU. If everything was equal regarding airflow then the smaller, more compact and direct cooler on the CPU would have the best increased airflow over the VRM chipset, which would definitely make a difference. There are so many if's but's and maybe's in all of that which make it completely impossible to state as fact.

The coolest air in a case, with everything else being equal, is an AIO. Believe it or not, the AIO exhausting or intaking doesn't actually have that much of an impact. The reason for this is that the coldest air is going through it initially, which does soak up the heat from the CPU and the pump, but it's such a difference in delta that the temperature of the flow over the GPU and chipset will only be marginally higher than the outside temperature (this has been proven). The significance of the cooling performance to the CPU, with the cold air, far outweighs the slight detriment to the GPU and other components (showed no impact in performance during the testing). So when configured correctly, in a case with good airflow and a decent exhaust setup, having the AIO as an intake actually benefits overall performance.

Now, with regards to the static... there will never be static air in a case with any sort of GPU or fan structure in it so you won't really need to worry about static air. There will always be an airflow of sorts. The closer the temperature delta of the heatsink and the flowing air, the more flowing air you need to cool the chipset. The most efficient way to impact this is to reduce the air temperature, the second most efficient way is to increase the airflow. Having lower case temperatures with good airflow significantly beats higher case temperatures with good airflow. AIOs always reduce case temperatures simply due to how they work. They automatically increase the airflow through the case, rather than simply internally around it, and the heat exchange is more efficient per m cubed of air flowing through, which makes better use of the air going through the entire system by default.

Incidentally, I was surprised to find that AIOs have gotten so efficient with their exchange that they will actually outperform high end tower coolers like the best Noctua in an open air environment. Previously AIOs only made the most sense when measured in a case due to, as described above, the way they operate and help with a closed space.

 

[Scott]

One more Q:

What causes the Chipset to get toasty? Many or most boards these days seem to provide a direct data transfer between the CPU and the first M.2 drive and one or more PCIex16 slot(s). So the GPU and first storage device bypass the Chipset. Would it only be the case that if you are heavily using peripherals or storage devices that run through the Chipset that it gets a heavy load?

The VRM is the voltage regulation module/manager. It controls the voltage/power to the CPU. The more intense the operation, the more power the CPU will draw. It's basically the link between the PSU and the CPU. If the CPU was allowed to, it would just pull all the watts it could. The VRM is what manages this draw. The more stable and less stress the VRM is under, the more stable and efficient the CPU runs. When you add cores and more power draw, the VRM comes under strain. This is why, when opting for heavy core utilisation, you want the beefiest VRM you can get.

 

[NoddyPirate]

Brillo Scott - thanks again!

Just thinking about it as you have laid it out makes so much sense. More to digest again for me.

I totally agree with the rabbit holes on the Interweb. I have tried to be careful about sources and their attention to detail. Gamers Nexus has to be my favourite for their literally OCD like attention to detail which appeals to me quite a bit!

 

[NoddyPirate]

The VRM is the voltage regulation module/manager. It controls the voltage/power to the CPU. The more intense the operation, the more power the CPU will draw. It's basically the link between the PSU and the CPU. If the CPU was allowed to, it would just pull all the watts it could. The VRM is what manages this draw. The more stable and less stress the VRM is under, the more stable and efficient the CPU runs. When you add cores and more power draw, the VRM comes under strain. This is why, when opting for heavy core utilisation, you want the beefiest VRM you can get.

Yep that’s great. Thanks!

In terms of Chipset temps (as opposed to VRM temps) I would imagine ensuring your heaviest workload devices are installed in locations that bypass the chipset would keep the chipset temps down?

So, installing your GPU in an alternate PCIe slot that runs through the Chipset rather than direct to CPU would make it run hotter? Or having a load of SSD reading activity that is Chipset managed rather than CPU direct?

 

[Scott]

Brillo Scott - thanks again!

Just thinking about it as you have laid it out makes so much sense. More to digest again for me.

I totally agree with the rabbit holes on the Interweb. I have tried to be careful about sources and their attention to detail. Gamers Nexus has to be my favourite for their literally OCD like attention to detail which appeals to me quite a bit!

GN is definitely one of the best, and one of my favourites. They are high end, extreme testers though and they test in very controlled environments that can be replicated. This means, basically, that they tend to use bench rigs (open air chassis). These aren't very indicative of how a users system will operate.

There are obviously limits to the testing they can carry out though, and what makes sense to compare. My only gripe with what they do is that people who don't really understand what they are watching/reading/listening to, tend to fill in their own blanks and then regurgitate their hybrid opinion as fact/gospel without truly understanding the variations of their own systems.

 

[Scott]

Yep that’s great. Thanks!

In terms of Chipset temps (as opposed to VRM temps) I would imagine ensuring your heaviest workload devices are installed in locations that bypass the chipset would keep the chipset temps down?

So, installing your GPU in an alternate PCIe slot that runs through the Chipset rather than direct to CPU would make it run hotter? Or having a load of SSD reading activity that is Chipset managed rather than CPU direct?

I've never had any reason to be concerned over chipset temperatures. It's unlikely you would do anything that would cause issue. Most of the chipsets are designed for maximum limit of operation through the board. Ergo, the X570 chipset will have a far better cooling system with the dual PCIe4.0 in place over the B550.

I would guess your theory is correct in that regard though, but I honestly haven't given it much thought or even looked into it as it's just never came up for me

 

[NoddyPirate]

I've never had any reason to be concerned over chipset temperatures. It's unlikely you would do anything that would cause issue. Most of the chipsets are designed for maximum limit of operation through the board. Ergo, the X570 chipset will have a far better cooling system with the dual PCIe4.0 in place over the B550.

I would guess your theory is correct in that regard though, but I honestly haven't given it much thought or even looked into it as it's just never came up for me

The fact it never came up for you tells me it likely isn’t an issue at all as you suggest.

Anyway, I’m having fun pondering all this stuff. Thanks so much for indulging me Scott!

 

[NoddyPirate]

Oooh - just finding that some AIO models actually have VRM cooling fans on the CPU cooler block.

Gamers Nexus makes the point in this article that VRM cooling is a bit of a gimmick as VRM temps are rarely limiting. I guess if they are it's because you just need a new mobo which can handle your overclock better.....

 

[Scott]

Oooh - just finding that some AIO models actually have VRM cooling fans on the CPU cooler block.

Gamers Nexus makes the point in this article that VRM cooling is a bit of a gimmick as VRM temps are rarely limiting. I guess if they are it's because you just need a new mobo which can handle your overclock better.....

Totally agree. The limitation with the VRM that when it is overworked it will throttle the CPU performance. If you purchase the appropriate motherboard for the CPU (which you will see suggestions of in my desktop building guide) it should never be a factor.

The trouble comes from trying to run a 5950X overclocked on a B550 Gaming Plus and wondering why it's not running as well as it could.

It's like trying to run a V8 with a Corsa 1.0L fuel setup.

 

[NoddyPirate]

Totally agree. The limitation with the VRM that when it is overworked it will throttle the CPU performance. If you purchase the appropriate motherboard for the CPU (which you will see suggestions of in my desktop building guide) it should never be a factor.

The trouble comes from trying to run a 5950X overclocked on a B550 Gaming Plus and wondering why it's not running as well as it could.

It's like trying to run a V8 with a Corsa 1.0L fuel setup.

Super - all coming together in my head now. Again I have no intention to overclock myself - but it’s interesting to see that just picking a chip and slapping an AIO on it might not work so well. There’s more to consider to ensure your intended use can actually be achieved.....

 

[Scott]

Exactly, compatibility does not equal suitability

 

[NoddyPirate]

Exactly, compatibility does not equal suitability

Thanks again Mr Scott sir......