UPMEM has developed a scalable and programmable Processing-In-Memory (PIM) solution that performs computation directly within memory, eliminating the need for costly data movement. This technology enhances performance by 15 times and reduces energy consumption by a factor of 10, significantly lowering the total cost of ownership for data-intensive applications compared to traditional FPGA or GPU solutions.
Funding
$15.9M raised to dateRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
Founders
Product
Problem
Modern data-intensive applications face performance bottlenecks due to the physical separation of processing units and memory, leading to excessive data movement and increased energy consumption. Traditional architectures like CPUs, GPUs, and FPGAs struggle to efficiently handle the growing demands of big data and AI workloads.
Solution
UPMEM offers a Processing-In-Memory (PIM) solution that integrates computation directly into memory chips, eliminating the need to move data between the processor and memory. This approach significantly reduces latency and energy consumption, leading to substantial performance gains for data-intensive tasks. UPMEM's PIM technology is fully programmable and scalable, allowing it to be applied to a wide range of applications without requiring significant changes to existing hardware architectures. By minimizing data movement, UPMEM's solution lowers the total cost of ownership for data centers and other environments dealing with large datasets. The PIM architecture enables efficient and sustainable acceleration of compute-intensive workloads.
Target Audience
The primary target audience includes data centers, enterprises, and research institutions dealing with large-scale data processing, AI, and high-performance computing applications.
Features
- PIM architecture that performs computation directly within the memory chip
- Scalable and programmable solution adaptable to various data-intensive applications
- Significant reduction in data movement, leading to lower latency and energy consumption
- Compatibility with existing hardware architectures, minimizing the need for extensive modifications
- Optimized for big data and AI workloads, delivering substantial performance improvements