According to a report from Reuters, Intel is going to leverage TSMC’s 7nm+ node for its Xe-HPG (High-Performance Gaming) GPUs, slated to arrive later this year. Otherwise known as the “DG2”, the successor to the Xe-LP based Xe Iris Max “DG1”, Intel’s first wave of offerings in the mainstream gaming market are expected to compete across a wide spectrum, with as many as 4,096 shaders and 32MB of L3 cache:
The graphics cards based on the DG2 die will reportedly compete in the $400 to $600 range, primarily offering performance in-line with the GeForce RTX 3060, 3060 Ti, 3070, and perhaps even the now uncertain 3070 Ti. The top-end offerings such as the RTX 3080 and the Radeon RX 6800 XT shouldn’t get a rival from Team Blue anytime soon.
DG2 will be fabricated either on the N7P or the N7+ node, both of which are more power-efficient and faster than the vanilla 7nm process used to manufacture AMD’s Navi and Navi 2x GPUs. They’re also a step ahead of Samsung’s 8nm LPP node (which is essentially a 10nm process) on which NVIDIA’s RTX 30 series “Ampere” graphics cards are fabbed.
- Intel may use a Large L3 Cache in its Xe DG2 GPUs, Similar to AMD’s Infinity Cache
- Intel Xe-HPG Gaming GPUs Powered On, Xe-HP Sampling and Ponte Vecchio (HPC) in Manufacturing Stage