InspireSemi's Thunderbird I Chip A Milestone in Accelerated Computing

TapTechNews June 27th news, the chip design company InspireSemi recently announced that it has successfully completed the tapeout of the Thunderbird I accelerated computing chip using TSMC's process.

InspireSemi's partners on this chip also include ASE (TapTechNews Note: Expected to provide packaging services) and the Belgian Imec.

Thunderbird I is a chip-based supercomputer cluster or many-core processor chip for HPC, AI and other computationally intensive loads: The chip is equipped with 1536 64-bit custom RISC-V cores, while integrating high-speed memory and a low-latency on-chip network.

In InspireSemi's view, since most software is written for CPUs, rather than rebuilding the software to run on GPUs, it is better to replace it with a many-core CPU with the same computing power to enjoy the convenience of the existing CPU software ecosystem.

InspireSemi claims that this SoC can provide unprecedented performance for a variety of applications, with first-class energy efficiency and disruptive prices, and is an ideal platform for accelerating large-scale image analysis and other workloads.

InspireSemi stated last July that each Thunderbird I chip can provide 8 TFLOPS of FP64 computing power, which is very important in HPC applications, and the energy efficiency is 50 GFLOPS/W (in FP64 data format).

The company's first batch of products based on Thunderbird I will be a PCIe accelerator card containing four of these chips. This card will have more than 6000+ CPU cores and is expected to be delivered to customers in the fourth quarter.

Also, the Thunderbird I chip can support up to 256 interconnections.

InspireSemi founder, chief technology officer and president Alex Gray said:

This is an important milestone for our company and an exciting moment as we bring this multifunctional accelerated computing solution to the market.

Thunderbird can accelerate many key applications in important industries that other methods cannot do, including life sciences, genomics, medical devices, climate change research, and applications that require deep simulation and modeling.