Samsung Electronics has recently announced that it has developed the world’s first supercomputer by connecting PIM (processing-in-memory) and GPUs (graphics processing units) together. The company has also disclosed the process at the Artificial Intelligence (AI) Semiconductor Future Technology Conference 2022.
Join SFC Nation On Telegram
At the conference, the head of the AI Research Center at Samsung Electronics Advanced Institute of Technology vice president Choi Chang-kyu explained that they have constructed a one of kind computing system by combining 96 GPUs (MI100) from AMD of the US with each of them loaded with a PIM chip.
The head of the AI research center Choi has further explained that the use of AI is increasing in semiconductor production processes. And that’s why he has planned to drive Samsung as a semiconductor company that uses AI better than any other company in the world.
Earlier, Samsung also succeeded in demonstrating the performance of a processing in memory (PIM) chip by installing it on a graphics processing unit (GPU) accelerator card from AMD. After that, it has trained the Text-to-Test Transfer Transformer language model algorithm T5 which improved its performance by 2.5 times and its power consumption dropped by 2.67 times, compared to when PIMs were not used.
At last, the Korean tech giant is also analyzing methods to combine the know-how contained in the notes of engineers working on-site with inspection data and automatically use it in production processes.
