Intel has reportedly sought to reserve TSMC’s 7nm Fabrication process for its second generation of in-house developed Intel Iris Xe DG2 dGPU. The company is already deep in development of the discrete graphics chips but needs a reliable fabrication process to accelerate production and deployment.
Intel intends to beat AMD and NVIDIA at their own games. After the successful launch of the Intel Xe Iris MAX discrete GPUs, the company is already on the development of the Intel Xe DG2 GPU. The second-generation graphics chip is reportedly based on the Intel Xe-HPG Architecture. While Intel is actively developing the GPU, it still needs to fabricate the same at scale. It seems Intel has secured or could book TSMC’s 7nm Fabrication Process for the Intel Xe Iris DG2 GPU.
Intel Has Confirmed the Xe DG2 will Be Manufactured Using An External Foundry:
Intel has struggled with its own Fabrication processes for the past two years. Hence, the company confirmed that the DG2 will be manufactured using an external foundry. Intel has recently started mass-manufacturing its Ice Lake Xeon CPUs as well as Tiger Lake mobile CPUs. These CPUs are made on the 10nm Fabrication process developed in-house.
— Dylan Martin (@DylanLJMartin) January 12, 2021
Intel is strongly considering relying on third-party manufacturing to produces its CPUs and GPUs. This could be because other foundries, such as those of Samsung and TSMC, have successfully developed newer processes, while Intel is still stuck with 14nm and 10nm Production Nodes.
The head of Mobileye, Intel’s subsidiary focusing on self-driving technologies, has indicated that the processor for the autonomous vehicle will be manufactured using TSMC 7nm node. Intel has been relying on TSMC for quite some time. Except for its top-end, flagship CPUs, the company regularly gets its mainstream CPUs from TSMC.
Intel Xe Iris DG2 GPU To Compete With Mainstream AMD and Mid-Level NVIDIA GPUs?
It is currently not clear how much different is the Enhanced 7nm TSMC process over the current 7nm technology. Incidentally, AMD also relies on TSMC for its ZEN 3-based Ryzen CPUs. However, according to reports, the 7nm node that Intel Xe DG2 would use, is ‘more advanced’ than Samsung’s 8nm process that NVIDIA is currently using for its Ampere graphics cards.
Reports also claim Intel wants to position its second-generation Xe graphics chip, the Intel Xe Iris DG2, against mainstream GPUs from AMD and NVIDIA. Expected to arrive next year, the new dGPU from Intel will reportedly compete with Nvidia and AMD gaming chips that cost between $400 and $600.
Intel DG2 rumored to be manufactured using enhanced 7nm TSMC node https://t.co/zcpnGY4bvH
— VideoCardz.com (@VideoCardz) January 12, 2021
Previous leaks about the Intel Xe Iris DG2 GPU indicated it will pack 4096 cores (Intel’s Shading Units). In fact, a file inside a driver update had listings about 128 and 512 Execution Units, with each carrying 8 Unified Cores. Rumors even indicate Intel would pack 6GB or perhaps even 8GB GDDR6 VRAM.
Intel’s second-generation Xe DG2 could be offered as an integrated dGPU or discrete graphics chips in Tiger Lake-H and Alder Lake-P gaming laptops. However, Intel was rather silent on these new graphics chips at the CES 2021.