ConvNext is a Winner!
On Chimera QC Ultra it runs 28x faster than DSP+NPU
ConvNext is one of today’s leading new ML networks. Frankly, it just doesn’t run on most NPUs. So if your AI/ML solution relies on an NPU plus DSP for fallback, you can bet ConvNext needs to run on that DSP as a fallback operation. Our tests of expected performance on a DSP + NPU is dismal – less than one inference per second trying to run the ConvNext backbone if the novel Operators – GeLu and LayerNorm – are not present in the hardwired NPU accelerator and instead those Ops ran on the largest and most powerful DSP core on the market today.
This poor performance makes most of today’s DSP + NPU silicon solutions obsolete. Because that NPU is hardwired, not programmable, new networks cannot efficiently be run.
To read the full article, click here
Related Semiconductor IP
- Frequency Divider
- Specialized Video Processing NPU IP for SR, NR, Demosaic, AI ISP, Object Detection, Semantic Segmentation
- Ultra-Low-Power Temperature/Voltage Monitor
- Multi-channel Ultra Ethernet TSS Transform Engine
- Configurable CPU tailored precisely to your needs
Latest Blogs
- Silicon Insurance: Why eFPGA is Cheaper Than a Respin
- One Bit Error is Not Like Another: Understanding Failure Mechanisms in NVM
- Introducing CoreCollective for the next era of open collaboration for the Arm software ecosystem
- Integrating eFPGA for Hybrid Signal Processing Architectures
- eUSB2V2: Trends and Innovations Shaping the Future of Embedded Connectivity