As you know, over the last few days I’ve had the opportunity to test the MSI RTX 5090 Suprim SOC and the MSI MEG AI1600T PCIe 5 power supply. Obviously, I’ve already launched a battery of tests, and the first impressions are promising. But before I give you the full reviews, I’d like to take a look at the evolution of graphics cards. While performance continues to increase with the RTX 5000, there is also a trend towards lock-in, with increasingly strict restrictions on overclocking. Enthusiasts have less and less freedom to exploit their hardware to the fullest. The question is: how far will this trend take us?
Changes :
I’m in the middle of testing a new generation of graphics card and discovering the changes it brings. With this new generation, GDDR7 replaces GDDR6X, resulting in a change in behaviour and an increase in performance, as well as changes in the possibilities for overclocking the memory. On the core side, there are few differences from the RTX 4000: the architecture and manufacturing process remain similar.
I’m not talking here about the on-board technologies that give this card a considerable boost, whether for gaming or professional tasks.
This generation of graphics cards is more plug and play than ever. NVIDIA has imposed more limitations on its chips, reducing the freedom of its partners, such as MSI, to modify them.
Software limitations?
We’re talking here about a graphics card with a blocked PL (Power Limit), i.e. a maximum power consumption in watts that remains limited, as on all graphics cards of this generation. This means that in Afterburner, even if you use a custom card – a version modified by a manufacturer such as MSI, with a different PCB (printed circuit board) and cooling system from the NVIDIA reference model – you will still be subject to this power consumption limit. Unlike previous generations, it is no longer possible to adjust it, whether you have an ultra high-end or entry-level model.
The Power Limit of this graphics card, like that of its competitors, is locked at 600 watts in the BIOS, and no software can be used to modify it. This corresponds to its maximum power consumption. There is also a maximum voltage lockout, as well as other restrictions, further limiting overclocking possibilities. These constraints are in addition to those of the Power Limit, further reducing the freedom of overclocking for enthusiasts and overclockers.
Hardware also?
In addition to software limitations, there are also hardware restrictions, particularly on the PCBs of these graphics cards. This trend has been going on for some time, but it is becoming even more pronounced with the RTX 5000. Manufacturers like MSI now have less freedom to modify the design of NVIDIA cards, resulting in more standardised PCBs.
Of course, there’s still a noticeable difference between an entry-level card and a high-end model, but the scope for overclocking on premium models is gradually shrinking. Where top-of-the-range versions used to allow advanced optimisations, they are now more limited, leaving less room for enthusiasts to exploit their full potential.
Why is this?
For several generations now, NVIDIA has been trying to restrict more and more the modifications made to its cards and the freedom given to manufacturers. Since the RTX 2000, these limitations have become apparent and have become more pronounced over time. Overclocking, which used to be more accessible, is becoming increasingly difficult and limited, to the point where we might wonder if it isn’t coming to an end on graphics cards.
It begs the question: why is NVIDIA adopting this strategy? As I see it, there are several objectives. Firstly, it is undoubtedly a desire to improve the reliability of graphics cards. By limiting manufacturers’ room for manoeuvre, particularly in terms of hardware design and BIOS, NVIDIA is ensuring that cards comply with stricter standards. Less freedom means fewer risks associated with excessive modifications, guaranteeing controlled performance and a longer lifespan, without pushing the hardware beyond its original limits.
There is also another reason for these restrictions: to prevent the different ranges from overlapping too much. Overclocking too effectively could allow a lower-range card to catch up or come close in performance to a high-end model, which would be detrimental to the product segmentation established by NVIDIA.
NVIDIA is not the only one concerned!
And yes, NVIDIA is not alone in following this trend. AMD has also imposed restrictions on overclocking, particularly since the RX 7000 series. For example, the use of software such as More Power Tool, which used to allow total control over the card, is now blocked, considerably limiting overclocking possibilities.
Processors are also affected by these limitations, although to a lesser extent than graphics cards. While there is still some room for manoeuvre, particularly when it comes to overclocking, there is a trend towards progressive locking, with restrictions on voltages, frequencies and power consumption management.