1024 Neuron Semiconductor Chip CM1K

CogniMem's CM1K pattern recognition chip with 1024 neurons in parallel

Image of Cognimem's 1024 Neuron Semiconductor Chip CM1K SeriesCogniMem Technologies' CM1K is the first ASIC version of the CogniMem neural network. It features 1024 neurons working in parallel, capable of learning and recognizing patterns of up to 256 bytes in a few microseconds. The two non-linear classifiers (RBF and KNN) supported by the CM1K can classify patterns while coping with ill-defined data, unknown events, and changes of contexts and working conditions. The parallel architecture of the CM1K allows daisy-chaining many chips together to increase the size of the neural network in increments of 1024 neurons. Its low pin-count and low power consumption make it an ideal companion chip for smart sensors and cameras. Its cascadability makes it an ideal candidate for large data mining systems.

  • Unique: The only commercially available hardware Neural Network
  • Scalable: Expandable to fit application needs
  • Fast application design: Eliminates complex algorithms
  • Portable: Knowledge can be copied, restored, and appended
  • Low power: < 300 mW
  • General purpose: Data agnostic
  • Low cost: Reduced component count
  • High-speed: Constant 10 μs recognition time

CM1K Series

ImageManufacturer Part NumberDescriptionTypeApplicationsMounting TypeAvailable Quantity
IC ASIC CHIP CM1K 1024 NEURONS110-0002IC ASIC CHIP CM1K 1024 NEURONSPattern RecognitionCo-ProcessingSurface Mount172 - Immediate
110-0002 product page link
English  |  Español
Published: 2012-12-26