Wafer-Scale Engine
A wafer-scale AI accelerator chip developed by Cerebras Systems. By integrating an extremely large number of compute cores and on-chip memory across an entire silicon wafer, it eliminates the interconnect bottlenecks of conventional multi-chip designs, delivers high memory bandwidth, and is suited for large model training and inference.
Core Facts
Timeline (last 90 days)
A wafer-scale engine (WSE) is a processor architecture that designs and manufactures an entire wafer as a single complete chip, bypassing manufacturing defects through redundant cores and reconfigurable interconnects.
Cerebras's WSE integrates SRAM directly next to each core on the wafer, forming distributed on-chip memory, with other cores' memory accessed via the on-chip network.
Specialized inference chips (such as Cerebras's wafer-scale engine, Groq's LPU, etc.) and model optimization techniques are driving AI inference costs down rapidly.
The Cerebras wafer-scale engine has an area exceeding 46,000 square millimeters, while a conventional GPU is approximately 800 square millimeters.
All Facts (5)
Cerebras Systems focuses on wafer-scale chips (Wafer-Scale Engine), fabricating an entire silicon wafer into a single processor
75%UnverifiedA wafer-scale engine (WSE) is a processor architecture that designs and manufactures an entire wafer as a single complete chip, bypassing manufacturing defects through redundant cores and reconfigurable interconnects.
50%UnverifiedCerebras's WSE integrates SRAM directly next to each core on the wafer, forming distributed on-chip memory, with other cores' memory accessed via the on-chip network.
50%UnverifiedSpecialized inference chips (such as Cerebras's wafer-scale engine, Groq's LPU, etc.) and model optimization techniques are driving AI inference costs down rapidly.
50%UnverifiedThe Cerebras wafer-scale engine has an area exceeding 46,000 square millimeters, while a conventional GPU is approximately 800 square millimeters.
50%