AMD RX 9070 Series Graphics Cards to Support ROCm; Team Red Reveals Working Prototype

AMD’s forthcoming RX 9070 GPU series will support the company’s ROCm software stack, as officially stated by a company spokesperson in a recent post on social media. This news is particularly significant for developers and enthusiasts working with AI-driven applications.

Support for ROCm in AMD’s RDNA 4 GPUs: Updates and Expectations

During the recent reveal of its RDNA 4 GPU lineup, AMD introduced several new models aimed at enhancing the gaming experience. Initially, it seemed that the RX 9070 series might not support ROCm at launch, disappointing many who planned to leverage these GPUs for AI workloads. However, Anush Elangovan, AMD’s Vice President of AI Software, has confirmed that these GPUs will indeed support ROCm, and a functional prototype has already been demonstrated.

ROCm’ing on just fine https://t.co/KSquDUq34W pic.twitter.com/xhootvpzuE

— Anush Elangovan (@AnushElangovan) February 28, 2025

This clarification comes after initial speculation arose from a Q&A session, where AMD executives hesitated to confirm RDNA 4’s compatibility with ROCm. Now, with Elangovan demonstrating ROCm on the RX 9070 series, concerns have been alleviated. The pivotal question remains whether ROCm will be accessible on the very first day of the GPUs’ launch, as definitive confirmation has yet to be provided.

AMD Radeon RX 9070 XT For $649 & RX 9070 For $599 Launched: RDNA 4 GPU Up To 2.97 GHz, 16 GB Memory, Up To 68% Faster Than 7900 GRE 6

AMD has been intensively enhancing its ROCm framework to compete effectively with NVIDIA’s CUDA, which has significantly bolstered NVIDIA’s market share and revenue. With the rising demand for edge AI applications, many users are now employing AMD’s high-performance consumer GPUs to handle these tasks locally. This trend highlights ROCm’s increasing adoption, reinforcing AMD’s commitment to this technology.

Overall, the RX 9070 series is poised to make an impactful entrance into the market, offering substantial gaming performance alongside potential applications in AI. Although the VRAM might not cater to the most demanding AI workloads, these GPUs should meet the needs of average users effectively.

Source & Images