YouTuber RedGamingTech recently unveiled a prototype card of the flagship RX 6000 series graphics card, the RX 6900XT. The card features a triple-fan open cooler design and dual 8-pin connectors, providing 375W of power. While the 6900XT in the photo uses a triple-fan setup, lower models might use dual-fan heatsinks.
The appearance looks sleek, but since many custom designs will eventually be available, the crucial aspects of the RX 6000 series graphics cards are the exact specifications and performance. However, the ray tracing performance of the RDNA2 architecture has not yet been disclosed, and specific hardware specs have not been announced.
There were earlier rumors that the RX 6000 series would feature up to 80 CU units with 5120 stream processors, twice that of the current RX 5700XT graphics card.
Additionally, there have been discussions about the 256-bit bandwidth limitation, which RedGamingTech confirmed as accurate. It is expected that AMD will increase the cache to compensate for the bandwidth shortfall, as they did with their CPUs. The most exaggerated prediction is that the L2 cache could reach up to 128MB, 32 times its current size.
While hardware specs are essential, the most important factor is the real-world performance resulting from the combination of hardware and software. Based on AMD’s track record, it is likely that the actual hardware specs might align with these rumors. Still, the real-world performance might fall short of the rumors, and the officially announced performance might be significantly inflated.
According to this YouTuber’s rumors, the RDNA2 architecture RX 6000 series graphics cards can compete well with the RTX 30 series. The RX 6700XT, for instance, was benchmarked against the RTX 3070, and the flagship RX 6900XT was said to compete with the RTX 3090 and RTX 3090 Ti.
However, considering AMD’s marketing approach—one I personally disdain—they tend to market their products by fostering a friendly narrative among lower-tier targets while building antagonism towards their competitors. This strategy seems to cultivate loyal “fanboys.” This kind of verbal marketing, combined with the fact that AMD often mentions competitors without showing actual benchmarks, makes their claims dubious. Historically, their product announcements have often revealed specs that were overhyped, with previous unfinished generations being marketed as budget-friendly consumer products. Once the products became reasonably good, their prices matched the same lineup from competitors. They marketed themselves as different from their competitors, who they painted as deceiving consumers, but in the end, who really deceived the consumers?
Yet, their loyal followers, having invested so much in defending AMD, now see it as a matter of pride and continue to support them blindly.
Reflecting on this, it seems that my judgment is not just subjective but objective.
It is quite disheartening to see that such patterns of manipulation, seen in both commercial and political arenas, are so similar. People often don’t realize they’re being deceived and continue to play the role of loyal followers without questioning the narrative.