It seems that cases of NVIDIA GTX 780, GTX 780 Ti, TITAN and GTX 980 Ti graphics cards are beginning to appear, which would be breaking, due to some problems in the memory inductors in the reference models.
Technological products are characterized by planned obsolescence and one of the clearest and most flagrant examples is HP printers. It seems that a new case of programmed obsolescence has been uncovered, but in this case in NVIDIA products. Specifically, these cases would be the GTX 780, GTX 780 Ti, TITAN and GTX 980 Ti, which would be presenting problems, breaking the memory inductors, but only in the reference models of NVIDIA itself. It would not occur in the custom models, which have better components and higher quality heatsinks.
These accusations come from the YouTube video: 'Actually Hardcore Overclocking', where it is said that a large number of the company's graphics cards have suffered a severe deterioration in a very short time, breaking these. The affected components are quite complex to repair, since we are talking about components with very specific soldering, not suitable for users without knowledge of SMD soldering, more complex than soldering normal size components. The problem is in the inducers of the memories, which would be the ones that would be degraded.
Thinking about planned obsolescence is logical, but we also do not know what workloads these graphics cards have been subjected to. We would not rule out that they have been used for cryptocurrency mining or long gaming sessions. The problem also lies with NVIDIA's benchmark heatsink, which is by no means the best solution on the market. It could also be the case that these have some kind of manufacturing deficiency, due to a batch of components that is not of adequate quality or other problems derived from the passage of time. We will be attentive to this.
I have always understood by programmed obsolescence a team that even being in full operation is forced to discontinue by software or otherwise (by stopping supplying some input such as ink), if the failure is that they burn for me it may be more a design failure , or it was simply not made to last which also are not models with a few months. Perhaps a sign that it may be better to wait for the assemblers and not go for the reference ones, which are models used more to measure the performance of the chip at its launch.
No. If a graphics chip company sells you their reference models directly so that you can start playing, they should at least work correctly without problems of any kind ... if they want to sell you by lowering the quality of the components they let you know and that They sell it very cheaply, but it is not like that in the case of Nvidia or even in AMD because there is a duopoly that is becoming a monopoly (Nvidia) ... if this continues, I want to believe that Nvidia will sell me its reference models with the best possible quality (and the AIBs too)…. If you think that it is normal to change the gpu every year with each new shipment of triple A titles, we are in the oven!
The truth is that all these cards are overclocked by software, insert hidden code to increase the temperature of a specific component with voltage, or that a fan stops working as well as the one that does not want the thing so that the components burn that you must protect, it is something that can be safely programmed taking advantage of the business secret protection law (which prevents decryption of the source code). I am in favor of a European body reviewing random products every x time to see what it finds. In this specific case of confirming the manipulation, that organ forces envy to change all the affected cards for new for free, no derisory fines, costing them money is the only way that they do not even consider it. But of course, these types of measures are neither there nor expected.