If you are one of the lucky ones who has been able to buy a graphics card at your MSRP or at least close to him and on top of that you haven’t had a single problem so far, our most sincere congratulations. But unfortunately and although few cases come to the fore, certain rumors affirm that the number of RMAs is being the highest in remembrance.
The price skyrockets and … Is the construction of graphics cards getting worse?
Over the months we have been dealing with everything related to the quality of current graphics cards, how they have performed, how they have been manufactured and of course the errors they have had. The last one with EVGA in focus has been really surprising and evidence that something is not working in general in the manufacturers.
The number of errors and problems is the largest that we remember if we remember, starting with the problems of frequencies and black screens of the RTX 30 and the NVIDIA solution with its driver, going through the SMDs for the die, thermal pads without placed or misplaced, with the protection stickers on and the last so famous of the thimble in the VRAM memory.
This may be specific errors, but behind there are heatsink alignment problems, fans that fail, PCBs with bad solders and layers that short between them, leaving a nice paperweight for GPUs or thermal pastes of doubtful quality.
Why are all these problems happening?
There is no single answer and no official statements (as a general rule and except for EVGA) that explain this scenario. But the industry points to several factors to explain the worst construction of current graphics cards:
- Little time to develop PCBs and their heatsinks. Apparently and to avoid leaks, NVIDIA and AMD would have passed the first chips of their GPUs to the AIBs too late, with little time for the corresponding R&D. The hardware requirements have been much higher and demanding and therefore hardly any manufacturer chose the correct SMDs for their cards and voltage drivers.
- Bad planning on some models. The line of GPUs tries to expand so that a brand occupies the largest number of sectors within the market and thus sell more. This implies that if you launch different models (and not the same one with limitations to segment models) you have to have the resources and the time to get it right. The problem is that brands often respond a posteriori when the competition launches a product where they are not competing. This causes speeding up the process and making mistakes due to lack of prior planning.
- Haste for scarcity. We’ve been out of stock for GPUs for many months, so the assembly lines have been working around the clock to meet the needs of consumers. This as a rule implies that operators work longer hours and with this wear leads to errors in the chain. A single thermal pad can send RMA to a GPU, so workers are gradually being replaced by assembly line robots.
- Human error. This point is linked to the previous one, since if the workers are tired, human error in the chain multiplies, which in an extreme environment of stress leads the company to introduce new personnel. Even with this newly trained staff, inexperience is the reason for possible errors, which also affects the final quality of the product.
- Worse overall build quality. It is not the trend far from it, since if we look only two years ago we will see that the PCBs and the heatsinks have increased their quality in almost all the segments, from the low range to the extreme. It is not usually to save or skimp, but it is true that in the models that are going to compete for price, coils, voltage controllers and voltage regulators of second brands are being mounted, as well as an abuse of poor quality plastic.
It is possible that we leave something in the pipeline, but roughly these are the main factors that we can find today. Manufacturers and assembly lines must improve, because between the price and the RMA rate they are going to turn graphics cards into a luxury item, if not already, because apparently only the reliability of the GPU seems to be maintained.