One of the stories we missed over the break involved reports of exploding VRMs on a Gigabyte X79-UD3 motherboard. This centred around a Youtube video posted just before Christmas by user japan0827. In the four minute video he is seen running overclocking stability tests until one of the MOSFETS on the motherboard smokes, sparks and dies (We'd embed the video but suspiciously both embedding and commenting have been disabled).
This little video has sparked a storm among overclockers, who understandably get a little edgy when bits of motherboards blow up. This was exacerbated when a Google translation of a press release on the Gigabyte Taiwan website was picked up by some outlets. The translation in question mentions the word ‘recall’, which was picked up and run with. However not only was this a case of potential translation issues, the release specifically says “This statement is limited to Taiwan, Penghu, Kinmen and Matsu area, if its content updated, will be announced.”
So no recall, and even more importantly, the video in question appears to show user error more than a hardware fault. One thing we have learned over the past 11 years at Atomic is that an open testbench needs adequate airflow. This is a concession that needs to be made when testing outside of a case – inside one the case fans ensure that air is moving over any heatsinks on a motherboard, but on an open testbench you have to rely on ambient airflow from the CPU cooler .
When benching with a closed-loop water cooler one doesn’t have this kind of airflow, so common practice is to sit a case fan in such a way that it blows over the power circuitry. That way you have heat being removed from the heatsinks on top of the VRMs – without airflow these are just lumps on metal that heat up until they match the temperature of the underlying components. When reviewing X79 motherboards at launch both ASUS and Gigabyte stressed to us that we should have air flowing over these parts of the boards, especially because Intel supplied its closed loop water cooler with its review kits.
Watch the video in question and there is absolutely zero air flowing over the motherboard. This means that over the four minutes of footage there is very little heat being removed from the VRMs (there will be a small amount of transfer to the surrounding air, but with nothing to move it away the effect becomes negligible). If anything it is an example of the worst possible scenario for this kind of testing – the lack of airflow over the motherboard is also going to severely limit the overclocking headroom of the system.
Gigabyte’s answer to this was not to recall boards, but rather to issue a new F7 BIOS that removed a couple of the more extreme voltage settings on the X79-UD3, UD5, UD7 and G1.Assassin 2. This release was met with comments around the web suggesting that it had locked off the overclocking abilities on the motherboards. In reality it appears that the changes stop the VRMs from being pushed beyond their capabilities, as appears to be the case in the video.
To Gigabyte’s credit its response to these allegations was about as comprehensive as possible. Its head office overclocking guru, Hi Cookie, took the same model of motherboard used in the video, the X79-UD3. He then installed the new F7 BIOS and broke a bunch of records. Using Liquid Nitrogen he was able to push the Core i7-3930X to a record frequency of 5643.3MHz while also knocking over some benchmark records.
As for the recall, we contacted Gigabyte and were pointed to a statement on its Australian website for clarification. This statement recommends updating to the new BIOS and highlights that it is making an effort to ensure that motherboards sold in the future will come with F7 as standard. There is also a post on Gigabyte’s motherboard blog that points out the new BIOS is only really needed if you are overclocking – for those running boards at stock there is no need to update.