Home » Technology » AMD: The competition will increase the consumption of graphics cards significantly more than we do

AMD: The competition will increase the consumption of graphics cards significantly more than we do

About the upcoming generation of Nvidia (Lovelace) is rumored to bring a significant intergenerational increase in energy demands. In other words, while AMD hardware (RDNA 3) is expected to bring the top model around 375W TDP with a 450W limit for OC models, the top model series Lovelace in the optimistic case it does not start below 450 watts for the reference design and some sources even talk about 600W cards.

Some of these rumors (at least the lower consumption of AMD cards) were confirmed by Sam Naffziger. Naffziger is currently a product technology architect, vice president of the company and Corporate Fellow. For about 33 years, he worked in the field at HP, Intel and since 2006 at AMD, where he focused on energy efficiency of processors. After the departure of Raji Koduri at the end of 2017, it began to be more realized in the graphics division, where it is also realized in energy efficiency optimizations.

The demand for gaming and computing power is increasing and at the same time production technologies are slowing down, which also applies to the pace of their improvement. Performance will therefore continue to increase. We have a multi-year plan for a very significant increase in efficiency ahead of us, which will offset this curve, but the trend is clear. […] Although our designs are more energy efficient, this does not mean that we will not increase power limits if the competition does the same. It’s just that they will have to raise them significantly higher than we do.

— Sam Naffziger, AMD

While in the past, shifts in energy efficiency brought mainly new processes and architecture was secondary, in recent decades the situation has changed dramatically. You could even say she turned around. To do this, we are witnessing a rather strange phenomenon when Nvidia, which after experience with Fermi invested in energy efficiency and with generations Kepler and especially Maxwell it has jumped significantly, in recent years it has completely lost its lead and it even seems that with a generation Lovelace in spite of this, the better production process (4nm compared to the combination of 5nm and 6nm at AMD) will be worse than the competition.

AMD in relation to RDNA 3 speaks of ~ 50% increase in energy efficiency compared to RDNA 2, however, according to unofficial sources, the values ​​around 50% refer to 6nm Navi 33whereas for chip products that partially build on the 5nm process, the shift should be over 60%, for some models maybe up to around 70%.

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.