Modern CPU architecture for example the Versal architecture from AMD has seamless integration with compute accelerators namely FPGA, GPU, DSP, etc [1]. They e

Modern CPU architecture for example the Versal architecture from AMD has seamless integration with compute accelerators namely FPGA, GPU, DSP, etc [1]. They even has built-in ADC/DAC for communicatiobmn, sensing and data acquisition (DAQ) [2]. On top of that they have native support for dataflow library and framework including linear algebra that is more suitable for data-intensive in which AI is only one of the intended applications [3]. I was recently asked by my colleague why we still need CPU, since GPU is very dominant now and it seems it's all that we need. I just pointed out GPU is only one of the many accelerators available to the CPU, but a very good one that happened to be very useful and biased towards fancy killer applications namely graphic (game), blockchain (bitcoin) and AI (LLM/ChatGPT). [1] AMD Adaptive SoC Platform: Versal Architecture [pdf]: https://www.amd.com/content/dam/amd/en/documents/products/adaptive-socs-and-fpgas/versal/adaptive-soc-paper.pdf https://www.amd.com/content/dam/amd/en/documents/products/ad... [2] AMD adds RF-sampling data converters to Versal adaptive SoCs (2024): https://news.ycombinator.com/item?id=42899304 https://news.ycombinator.com/item?id=42899304 [3] AIEBLAS: Open-Source Expandable BLAS Implementation for AMD/Xilinx Versal Device: https://github.com/atlarge-research/AIE-BLAS https://github.com/atlarge-research/AIE-BLAS