"The company's most recent products of note are the Gemini-I and Gemini-II associative processing units (APUs). These APUs use Compute-In-Memory (CIM) architectures, which can process tasks inside the memory itself without the need to transfer the data to an external processor. Shares in GSI Technology surged 155% yesterday after the company announced the publication of a paper by researchers at Cornell University."
"This paper, titled "Characterizing and Optimizing Realistic Workloads on a Commercial Compute-in-SRAM Device," found that the company's associative processing units' CIM architecture can match GPU-level performance of certain AI workloads but at a fraction of the energy consumption. And that energy consumption reduction is significant-up to 98% less energy is required compared to GPUs. Additionally, the company's APU allow for the processing of retrieval tasks faster than traditional CPUs, which can reduce processing time by up to 80%."
GSI Technology is a semiconductor memory solutions company founded in 1995 and headquartered in Sunnyvale, California. The company employs 158 people and holds more than 120 granted patents. Its recent products include the Gemini-I and Gemini-II associative processing units (APUs) that use Compute-In-Memory (CIM) architectures to process tasks inside memory without transferring data to an external processor. A Cornell University paper found the APUs' CIM architecture can match GPU-level performance on certain AI workloads while using up to 98% less energy. The APUs can also process retrieval tasks faster than traditional CPUs, potentially reducing processing time by up to 80%. Shares surged 155% yesterday and rose another 39% in premarket trading today.
#compute-in-memory #associative-processing-units #energy-efficiency #semiconductor-memory #ai-workloads
Read at Fast Company
Unable to calculate read time
Collection
[
|
...
]