Once again, the specifications of Intel’s upcoming Xe-HPG (High-Performance Gaming) graphics cards have been leaked. The details are more or less in line with the previously shared data, with the entry-level part starting with just 96 EUs or 768 shaders and a 4GB VRAM buffer paired with a paltry 64-bit bus. The top-end part is a beastly 512EU GPU with 4,096 shaders and an 8-16GB GDDR6 memory buffer paired with a 256-bit bus.
The flagship part features specifications very similar to the Radeon RX 6800 with roughly the same bus-width, memory, and shader count. However, the former boasts a much higher effective bandwidth thanks to the use of the Infinity Cache and high boost clocks. It was reported earlier that Intel is looking at a similar L3 caching technology, but we haven’t anything since. As for the clock speeds, they will mainly depend on the process node used. We’re likely looking at TSMC’s 7nm process, but the boost clocks should generally be lower than Navi 2x.
The model below the 512EU part features 384EUs or 3,072 cores and a 192-bit bus paired with 6-12GB of GDDR6 memory. This is likely going to be a direct competitor to the RX 6700 and the RTX 3050 Ti/3060. Further down, we have a 256EU (2,048 cores) and a 192EU (1,536 cores) part with a 128-bit bus and up to 8GB of GDDR6 memory. These will go against the Radeon RX 6600 and the RTX 3050. Keep in mind that we’ll most likely only see 3-4 graphics cards hit retail (half as many mentioned above). The six variants mentioned above are likely engineering samples and not all will make the final cut.
- Intel Teases Xe-HPG Graphics Cards with Support for Mesh Shading and Other DX12 Ultimate Features
- Intel “DG2” Xe-HPG Graphics Cards to be Fabbed on TSMC’s 7nm Node