This is some good information and it does at least seem that some of the coolers have a problem. I most certainly don't like the sensationalism that he appears to have decided to add with the title of the video as well as with his concluding remarks, I think it tarnishes the scientific data he got and presented, and obviously attracts behavior of fanboyism. But given I always like figuring things out, I do have a variety of remarks and questions of my own. First, I am not very knowledgeable about the specifics of vapor chambers, but I am able to think of a couple of things that would be useful to know.
1. As always, having an example of a working one is always useful for these tests. I believe he had no working ones? But having one that presumably is functioning properly would be good as a point of comparison. It could help narrow down the exact reason the non-working ones are dysfunctional.
2. It is odd that no reviewers ever got this issue? I guess it only is happening with the reference model? And possible no reviewers ever actually test by putting it in a case horizontally or only very few? Which is a little funny, as we can of course make critiques of Amd's QC and testing, but this could also be directed towards reviewers as well, I think? In that them testing in ways that are more comparable to what an end user might experience would be what is most beneficial to end users. So maybe this can be a learning experience for reviewers as well to update their own testing parameters given they are suppose to be reviewing it for the end user's benefit.
3. I would have liked to see some testing on other ways to mitigate the issue, or testing on how certain "fixes" might mitigate the issue. What I mean would be for example, based on his last test, where he flipped it during load and then the issue occurred, but then flipping it back didn't fix it. Does it only not fix it while it is continuously under load? If he shut the system down and allowed the gpu and vapor chamber to cool off, and then placed it back in the vertical position, does it then cool properly again? Also let's say you undervolt and even power limit the card, does this them mitigate the observation he got in his flip test? This of course would not mean there isn't a problem but would be helpful for end users as something they may be able to do to still use their card well until they are able to RMA it, or if they are someone who doesn't want to RMA it.
So these are the thoughts that came to mind when I watched the video. I think cutting right to saying Amd has to recall the cards, or bringing up silly marketing jabs Amd directed towards Nvidia doesn't really help anyone who has the issue or adds anything besides attracting behavior of fanboyism as I mentioned. This is why that is the aspect I didn't like. As I said, what would be helpful would have been more testing on things customers could do to not have the gpu reach max temp. Getting in contact with Amd and relaying this to them and trying to get them to take the proper course of action is also good, but parts of the video comes off, as I said, as merely jabs at Amd.
But anyway, what's most important is understanding exactly what is causing the issue, and from that how many units may be affected. Whether it is a design flaw, or a manufacturing flaw, or a combination where the design had too low margin of error which can then lead to a lot of defects in manufacturing.
(and also ugly), some users didn't realize they did not fully plug in the connector
which one??? That's horrible!
