If you have not lived under a rock lately, you most surely have heard of German overclocker der8auer complaining about overheating VRMs in the newest motherboards of several manufacturers, stating the heatsink designs that should remove the heat from these important components were in reality creating a “VRM Disaster“, overheating the components and causing the motherboards to throttle the CPU frequencies down to safer specification thresholds. He even went so far to say that removing the heatsinks from some motherboards and cooling directly the naked VRMs with a normal 120mm fan resulted in better temperatures and no throttling whatsoever on the CPU under Prime95 (a CPU stress tool used by overclockers worldwide to verify the stability of the overclocked CPU)
At some point in the video he asks: “are mainboard vendors testing these boards before shipping…?” While I cannot answer that specific question, I can say he has a point, since as he himself noted earlier in the video that Intel rushed the launch of the X299 onto mainboard vendors to come out in June instead of August, basically cutting development time and testing by the same vendors who, probably, found themselves in the need to slap on the VRMs of their motherboards whatever they had in stock without proper testing.
Now even if this was the case, let#s take a step back and see what is really going on here, shall we? I have been testing motherboards, CPUs and the such since my first Celeron 400 overclocked to 600MHz on chilled water. I am talking of a CPU from 1999, here. Many of you younger readers my find it laughable, but in 1999 this was a major feat. Since 1998 Intel had locked the multiplier of their CPUs (contrary to AMD where you still had full control over multiplier and FSB) so the only way to overclock an Intel CPU was increasing the FSB values, which for that particular CPU was 66MHz. You could then keep the overclock stable enough increasing the Vcore, the voltage fed to the CPU, which for that Celeron started at 2.0V (yes, you read it right… 2 freaking volts!).
Things have gotten better in all these years, and I managed to get my hands on more CPUs I can remember of, always testing their limits on air and then under waterblocks (both self-made and off the shelf products). But one thing was always clear to me: overclocking makes not only the CPU run hotter, it also makes the surrounding components run hotter, too. This is something we learned immediately back n those days, where manufacturing quality was not what it is today, and motherboards just died on us if we pushed their limits a bit too far without proper cooling countermeasures.
Some of these countermeasures were, of course, strategically placed fans (see where I am going?) on top of the northbridge and southbridge chipsets, which usually came with inferior cooling solutions like only a passive heatsink and very poor thermal paste, but also providing active cooling to the VRMs that surrounded the CPU socket. In forums around the world you could find hundred of threads of people who even designed and manufactured at their home watercooling blocks specifically for the VRMs. Big names like Swiftech and Alphacool noticed the trend years after, and started selling dedicated waterblocks for VRMs. I was one of the early adopters of such solution, and I can honestly say, I would not have achieved some overclocks without these small copper blocks cooling my motherboards’ VRMs.
It was as simple as that: you had to cool everything in order to have a stable overclock, not only the CPU.
Getting back to the present: I see the point of der8auer when he says that proper VRM cooling on the new Intel X299 chip motherboard is the key to a stable overclock with acceptable temperatures and no throttling, but I cannot avoid to think that this is something we already knew back then when we only had our fingertips to measure the temperature of a component, and thermocouples and fancy setups were tools only professional labs could afford. We used aquarium pumps inside water buckets full of water and ice to cool a 23,7W – 2.0V Vcore – 66MHz FSB Celeron 400 CPU to get to 600MHz stable and emitting some 45W, with loud 80mm fans blowing air on the waterblock itself to cool the surrounding VRMs and motherboard northbridge chipset. Nowadays we are overclocking monsters of CPUs (if compared to those days…) up to 5GHz with 1.35V of Vcore and capable of more than 200W under stress.
The object of this article was, of course, not to make fun of an internationally renown overclocker as der8auer; who, from my very personal point of view, just pointed out (seemingly discovering it himself for the first time) something we already knew 10 years ago and that have been practicing since. The article points out really on one specific phrase from the overclocker: the mainboard vendors had probably no time to properly devise a cooling solution for the VRMs that you could trust your overclock to without taking some active action and cool it yourself.
As an overclocker myself, and past cooling products developer, I cannot really expect that a motherboard vendor, even if they launch a product specifically targeted to overclockers, takes care of every little factor that will make your overclocking experience the best (thus including a good VRM cooling design). Especially if they are on a rush imposed by CPU manufacturers. It is just not their interest to focus on a small group of people capable to push the limits of a CPU so high like der8auer, and not respect time constraints imposed by business partners.Vendors can always come out with a revised version of the motherboard later, when their initial investment has been recovered with the sales of the products rushed out of the factory and that millions will buy, notwithstanding the comments of a single (or maybe two) overclockers.
People like der8auer and Tom Logan are perfectly right: motherboard vendors were on a rush and launched products that had probably poorly designed elements if we only take in consideration their specific needs (overclocking capabilities). But what they fail to understand, or even if they do understand, they still fail to point it out, is that on the mass scale of the hardware market overclockers are still a very small niche. Motherboard vendors will not put their comments on the balance of their business calculations, no matter how many youtube subscribers they may have.
der8auer made an update video (so has Tom Logan) explaining more in detail their testing methodology and the reasons why they got so high temps on the VRMs. Again, both overclockers/youtubers were testing the products to their limits. der8auer specifically stated the reason behind such practice was his partnership with german distributor Caseking.de: testing pre-assembled PCs that were sold to customers. This is a normal practice for system integrators and pre-overclocked systems like Caseking, but honestly thi si snot something that normal users will ever have to do, once their system has arrived. It is still a good thing that such things are told and made public, but I still do not understand why the media got so fussed up about an issue on which vendors had only partially control over.
Let me quote der8auer: “…if you are a vendor …. just go back to some proper heatsink design that are working well… because I know that all the consumers…..” I wonder if he ever realized not really ALL consumers are aware of the overclocking capabilities of a specific motherboard, or even have the knowledge, that clearly der8auer has, to push such components so hard. der8auer belongs to a niche market, a very small group of people who do overclock their CPUs to the limit: say 1% of the whole market for motherboards?. Vendors care of the rest 99%, unfortunately. It has always been like that, and it will always be so.