The global HPC market is expected to grow to $64.65 billion by 2030, with an annual growth rate of 5.5%. Over three-quarters of key research groups recognize the importance of HPC platforms. Moreover, many experts predict that without HPC, there will be setbacks in future technological competitiveness.
Supergate is targeting the HPC market with a focus on GTX(Tensor GPGPU).
Currently, most companies are pursuing NPU*-based development focused on AI performance indicators like MLPerf, where NPU's latency and efficiency are advantageous in some areas. However, compatibility is crucial in the HPC field, emphasizing the need for GPU*-based hardware. AI platforms such as ChatGPT, training computers, cloud servers, and supercomputers require support for various architectural software, where GPGPU offers higher versatility than NPU.
Therefore, our company has adopted GPGPU as the core framework of parallel processors while maintaining the basic structure of GPU, integrating the functions of General CPU. The goal is to develop a semiconductor architecture applicable to various systems such as IoT, vision AI, and server AI. Supergate plans to further secure its competitiveness in the HPC and Computing Module business based on this development process.
*GPGPU: General-Purpose Graphic Processing Unit, a methodology to use conventional GPUs for general computing. Traditionally used for graphic operations, GPGPUs are now employed for other types of computations like parallel data processing.
*HPC: High Performance Computing, referring to computer systems designed for large-scale computation needs. These systems are used for complex simulations, data analysis, and artificial intelligence research.
*NPU: Neural Processing Unit, dedicated hardware designed to efficiently process artificial neural networks, specialized in accelerating deep learning algorithm computations.
