Intel Xe GPUs are blossoming on all fronts, such as Tiger Lake, a thin, mid-year platform that will integrate the graphics core of the Xe architecture for the first time, DG1, DG2 discrete graphics for the mainstream market, and Ponte Vecchio for data centers and large-scale computing.
Recently, Intel official show called “baap of all” (Heavenly Parent) a giant GPU chip, the package area of about 3,696 square millimeters, the effective chip area of about 2343 square millimeters, compared to the 10th generation of 10 core 20 threadCore i9-10900K is a full 1.6 times larger.
At the same time, the back of the chip can be seen divided into eight groups of components, obviously corresponding to the front of the eight GPU core, unified package in a single chip, which is also the same as Intel previously announced the design idea.
Intel doesn’t disclose the number of transistors it has, but uses the term “many tens of billions of billions of” to describe it.
This figure out, so that many people and professionals eat melons are very excited, there aremedia began YY it and NVIDIA RTX 3080 Ti high-end duel, but it is clear that playing the game is not the task of this Heavenly Parent GPU.
Raja Koduri has sent several tweets explaining some of intel GPU’s directional plans, highlighting that the Chips of Ponte Vecchio and Xe HP architecture of the Xe HPC architecture are market-oriented for the data center market because from edge computing to cloud computing requires exceptionally powerful computing resources to handle the massive amounts of data generated in real time.
He also said he understood the excitement of gamers hearing about Intel’s GPU, and that Intel was committed to continuing to improve the experience of hundreds of millions of PC players, with a lot of improvements to the next generation of PC products (Tiger Lake).
However, Intel GPU’s main focus in PC graphics today is addressing some of the long-standing pain points, especially the set-up, which is why the Xe architecture was first applied to core graphics cards.
According to Intel’s plan, the Xe LP, Xe HP, Xe HPC three levels of GPUs, although the application market is different, but essentially based on the same underlying architecture, but the microarchitecture is different.