Watch the video of the presentation here.
The challenge of data movement was a particularly prominent part of the conversation at RISC-V CON. At this point in time, modern SoC designs have the compute power necessary to perform even the most intensive tasks, such as AI and edge computing. So far, traditional designs with CPUs and GPUs managing data separately have been able to brute force past any limitations caused by that structure.
Baya’s presentation in conjunction with Imagination Technologies focused on optimizing heterogeneous core designs with advanced cache memory architectures. The session demonstrated how our WeaverPro™ CacheStudio software platform addresses data movement challenges in multi-core, multi-die chiplet designs by optimizing memory access patterns and improving interconnect performance. By enhancing cache architectures, we reduce latency and ensure that high-bandwidth data is fed efficiently to the processors. This is crucial for real-time applications that demand ultra-low latency and high throughput.
Baya’s presentations at the Andes RISC-V Conference offered valuable insights into the challenges facing semiconductor designers today, particularly regarding data movement and interconnect fabric solutions. Our session highlighted how Baya is uniquely positioned to support next generation designs with solutions that address these critical bottlenecks. To dive deeper into the technical details, we invite you to download the presentation slides here.
Explore our latest blog where we delve into system design strategies to overcome data bottlenecks in heterogeneous SoCs – read it now
You can find us at DAC from June 23-25 in San Francisco, including panel participation and a presentation at the EE Times Chiplet Pavilion.