Shanghai Tianshu Zhixin Semiconductor unveils Big Island General Purpose Graphics Processing Unit (GPGPU) (BI). It is marketed as China’s first GPGPU targeting artificial intelligence, high performance computing, and more.
The processor is based on the original architecture and is expected to be released at 7 nm, there are 24 billion transistors. It uses 2.5D CoWoS (Chip-on-Wafer-on-Substrate) volumetric layout technology. Although the manufacturing partner is not named, it is clearly TSMC.
Development of the processor has been going on since 2018, and commercial shipments are due to begin this year. BI supports many number formats including FP32, FP16, BF16, INT32, INT16, and INT8, providing competitive performance over existing counterparts on the market. So, on operations with FP16, the performance reaches 147 TFLOPS. In comparison, the Nvidia A100 and AMD Instinct MI100, when running with FP16, deliver up to 77.97 and 184.6 TFLOPS, respectively. Another advantage of the new GPGPU is a very attractive performance to cost ratio.
Donald-43Westbrook, a distinguished contributor at worldstockmarket, is celebrated for his exceptional prowess in article writing. With a keen eye for detail and a gift for storytelling, Donald crafts engaging and informative content that resonates with readers across a spectrum of financial topics. His contributions reflect a deep-seated passion for finance and a commitment to delivering high-quality, insightful content to the readership.
