Intel's new CEO, Pat Gelsinger, says the company's 3D packaging technology is "flawless" and thanks to it "we have the ability to make tiles instead of chiplets."

AMD is using TSMC's chiplet packaging technology to make affordable, high core count CPUs, but Gelsinger suggests that Intel's stacked designs are so superior that Intel does not have to go that route and can use a higher-performance tile stacking approach.

The final Q&A session at last night's Intel Unleashed event highlighted the differences between Intel's Tile design and AMD's Chiplet model. Thanks to advanced packaging technology, Intel suggests that its modular stack approach is superior to AMD's Chiplet design.

One major difference between the two is how the individual elements communicate with each other: in AMD's chiplet design, chip-to-chip communication is fast but requires a bus to move from one chiplet to another, whereas this is not the case in Intel's tiled design.

"One of the cool things I went back and found is that even if there were some issues with process technology and 3D packaging technology, [Chef's Kiss] was perfect," says Gelsinger. Unquestionable leadership." This technology allowed us to make tiles instead of chiplets. This packaging technology eliminates the need to buffer the interconnects, making them like long wires on-chip"

. [This packaging technology offers a real advantage in the next-generation process roadmap, as it allows tiles from different process technologies to be mixed and matched as if they were one chip. We are moving from system-on-chip to system-on-package"

.

As part of his presentation, Pat Gelsinger showed off Ponte Vecchio, Intel's XPU for exascale computing and AI. While not the kind of chip you would find in a gaming PC, it is an interesting practical example of Intel's tile design, consisting of 47 tiles with over 100 billion transistors.

In other words, Intel's technology works with a variety of modules. This is due to both Foveros' 3D packaging technology and EMIB, remember how it was once used in Kaby Lake-G to push AMD GPUs into packages along with Intel CPUs?

Intel may have stumbled on process technology, but they still have a lot of technical know-how. Being able to produce chips in-house means that they don't suffer from supply shortages like AMD and Nvidia. And it means, ultimately, that the company is making profits where AMD and Nvidia cannot.