The hosts discuss AMD Radeon’s challenges, highlighting the lack of official driver support for certain products, slower adoption of advanced features like AI-driven frame generation, and a focus on raw GPU performance that leaves them behind Nvidia in competitiveness. They also touch on hardware issues like the 12VHPWR connector, the complexities of GPU memory supply amid AI demand, and suggest that AMD needs to prioritize software improvements to better meet market expectations.
In this Q&A session from Harbor Unbox, the hosts discuss a variety of topics related to AMD Radeon and the broader GPU market. They begin by addressing concerns about AMD’s support for the Ryzen Z series handheld APUs, particularly in the context of Lenovo’s handheld Z1E. The hosts clarify that AMD has never provided official driver support for the Z series; instead, it is the responsibility of the OEMs like Lenovo to manage drivers. This situation contrasts with Valve’s Steam Deck, where Valve actively maintains and updates drivers. The hosts suggest that Lenovo’s complaints about AMD’s driver support are somewhat misplaced, though they acknowledge that AMD could improve by offering official drivers for these products.
The conversation then shifts to the topic of power connectors for GPUs, specifically the controversial 12VHPWR connector used in high-end cards like the RTX 5090. Despite reports of connector issues and even melting, the hosts do not consider this a deal breaker for building systems, especially since such high-end GPUs are not commonly built for friends or typical users. They emphasize that while the connector design is flawed and should not catch fire, proper installation usually prevents problems. The discussion highlights the practical realities of hardware issues versus their perceived severity among enthusiasts.
When asked about the ultimate graphics card feature set combining Nvidia, AMD, and Intel technologies, the hosts pick DLSS from Nvidia for its advanced upscaling, AMD’s superior multi-monitor and display support, and Intel’s reputed video encoding capabilities. They note that AMD has lost its traditional advantage of offering more VRAM and that Radeon software, while competent, doesn’t stand out significantly. The hosts also point out AMD’s better automatic refresh rate support on monitors, which improves user experience, especially for novices with high-refresh-rate displays.
A critical question about why AMD’s Radeon team seems out of touch with their market position is explored in depth. The hosts speculate that AMD may be stuck in an outdated mindset that prioritizes raw GPU performance above all else, ignoring the increasing importance of features like high-quality upscaling, ray tracing, and AI-driven frame generation. They argue that gamers now value a broader feature set, where Nvidia currently leads, and AMD’s slower adoption of these technologies has hurt their competitiveness and market share. The discussion also touches on AMD’s frame generation issues and the company’s need to prioritize software improvements over chasing new but less essential features.
Finally, the hosts discuss broader industry topics such as the challenges GlobalFoundries faces in entering the DRAM market and the current state of GPU memory supply amid AI-driven demand surges. They explain that GlobalFoundries is primarily a foundry without the IP or infrastructure to produce DDR5 memory and that memory manufacturing is a specialized, capital-intensive process. The conversation also highlights skepticism about the longevity of the AI boom, suggesting that the current high demand for GPUs and memory may be a short-term phenomenon. The session concludes with thanks to the community for their questions and a note on the ongoing need to find better formats for these Q&A videos.