Why doesn't AMD make graphics cards like the RTX 4090 level?

Why doesn’t AMD make graphics cards like the RTX 4090 level?

Last year, AMD launched a new generation of graphics card RX 7900 series, but the highest specification RX 7900 XTX is not comparable to Nvidia’s RTX 4090 in terms of specifications or performance.

In response to this situation, AMD executives said in an interview that from a technical point of view, AMD is fully capable of developing a graphics card that is comparable to the Nvidia RTX 4090, but due to various considerations such as power consumption and final pricing, it was finally decided not to use it. launch such a product.

Why doesn't AMD make graphics cards like the RTX 4090 level?

Also last year, Nvidia released the RTX 40 series graphics card, among which the flagship RTX 4090 naturally received a lot of attention.

According to the official introduction, the RTX 4090 is the fastest gaming GPU in the world.

In terms of performance, Nvidia claims that in a complete ray tracing game, compared with the previous generation RTX 3090 Ti using DLSS 2, the performance of the RTX 4090 with DLSS 3 can be improved by up to 4 times.

In modern games, the RTX 4090 delivers up to 2x the performance boost while maintaining the same 450W power consumption.

RTX 4090 is equipped with AD102-300 GPU and enables 16384 CUDA cores, which is 52% more than RTX 3090 Ti, but the performance improvement is even greater. The GPU frequency is up to 2520MHz, equipped with 24GB 21 Gbps GDDR6X video memory, reaching a bandwidth of 1TB/s, 450W default TGP, and the maximum configurable TGP is 660W.

With the blessing of TSMC’s 4N process, the acceleration frequency exceeds 2.5GHz, which is a significant improvement over the previous generation of Samsung’s 8nm process.

AMD’s RX 7000 series graphics card using the RDNA 3 architecture still uses TSMC’s 5nm chip, with 1 main graphics computing module and 6 memory cache modules. More than that, AMD has made many low-level changes to the RDNA 3 architecture, including stuffing more computing units and more shaders.

From the official information, this generation of AMD maximizes the power consumption, and the performance of each watt is 50% higher than that of the previous generation.

This strategy is different from Nvidia’s power consumption for performance.

It makes many players feel ashamed, and the increase in AMD’s power consumption is more in line with players’ expectations.

On the power interface, AMD did not learn from Nvidia, but continued to use the 8 PIN interface.

The TDP of the RX 7900 XTX is 355W, and the TDP of the RX 7900 XT is 300W.

Compared with the 335W of the previous generation RX 6950 XT, the 7900XTX has increased slightly, but the 7900 XT has achieved a decrease in power consumption.

More importantly, this performance is more obvious in the comparison with RTX 4090. You must know that the TDP of RTX 4090 is as high as 450W, and the power consumption even exceeds 600W under some extreme overclocking conditions. Even the TDP of RTX 4080 has 320W .

So, when asked why it didn’t develop a benchmark RTX 4090-class graphics card, AMD’s senior vice president and product division manager replied: “Technically, it is feasible to develop a GPU with the specifications of (NVIDIA’s) competitors.

However, a graphics card developed in this way with a TDP of 600W and a reference price of US$1,600 is difficult to be accepted by general PC game fans, and finally chose not to adopt such a strategy after careful consideration. “

From the player’s point of view, AMD’s statement is not unreasonable. After all, there are only a few players who will buy RTX 4090 graphics cards, and more users still prefer dessert-level graphics cards.

At the same time, AMD previously claimed at the press conference that the performance of the RX 7900 XTX is far superior to that of the RTX 4080, but due to AMD’s long-term driver problems, the RX 7900 XTX cannot exert its full strength in the early stage.

AMD graphics cards have always had a tradition of “fighting the future”. With the continuous improvement of the RX 7900 XTX driver, and the prices of the current two RX 7900 series graphics cards have dropped significantly, the later RX 7900 series graphics cards will have greater competitiveness and cost performance.

In addition, AMD has always lagged behind Nvidia in terms of professional capabilities. RTX 4090 is not entirely for game users, but also for some professional users with deep learning and neural network needs.

Therefore, it is a little bit reluctant for AMD to compete with Nvidia in this regard. It is more in line with its own interests to serve game users honestly.

AMD’s current strategy does not seem to be a big problem, but driver optimization is imminent.

Why doesn’t AMD make graphics cards like the RTX 4090 level?