I was told that when comparing graphics cards, you only need to look at the last two digits of the model number to judge performance. The example was that a 3070 could be more powerful than a newer 5040, despite being two generations older. Is there any truth to that naming rule, or is GPU performance more complicated?
3 Answers
The specific comparison is misleading because a 5040 model doesn't exist, but the general idea can sometimes appear true. A higher-tier card from an older generation may beat a lower-tier card from a newer one. For example, a 3070 can outperform a newer entry-level card, depending on the exact models and workload. Check independent benchmarks rather than relying only on the names.
As a very loose rule, a newer card is often similar to a previous-generation card one tier higher, but there are plenty of exceptions. Something like a 5060 Ti, 4070, and 3070 Ti might land in a comparable range, though their features, power use, ray tracing, and upscaling support can differ. Treat the numbering as a hint, not a performance ranking.
That naming system is only a rough guide. For Nvidia cards, the first two digits usually indicate the generation, while the last two indicate the performance tier. Ti and Super versions are upgraded variants. Newer generations can change the comparison quite a bit, so the model number alone isn't enough. AMD uses a similar idea, with labels such as XT instead.
That makes sense—so the generation and tier both matter, and benchmarks are the safest way to compare them.