TPU (Tensor Processing Unit)

GPTKB entity

Predicate	Object
gptkbp:instanceOf	gptkb:application-specific_integrated_circuit
gptkbp:announced	2016
gptkbp:architecture	matrix multiplication optimized
gptkbp:availableOn	gptkb:Google_Cloud_Platform
gptkbp:designedFor	gptkb:machine_learning
gptkbp:developedBy	gptkb:Google
gptkbp:generation	gptkb:TPU_v4 gptkb:TPU_v1 gptkb:TPU_v2 gptkb:TPU_v3
gptkbp:location	gptkb:United_States global (via cloud)
gptkbp:notableFeature	energy efficiency custom ML hardware high throughput for ML workloads
gptkbp:notableProject	gptkb:Gemini gptkb:AlphaGo gptkb:LaMDA gptkb:PaLM gptkb:BERT gptkb:Google_Bard gptkb:Imagen
gptkbp:notableUser	gptkb:researchers academia enterprises
gptkbp:openSource	gptkb:JAX TensorFlow XLA
gptkbp:optimizedFor	low-precision arithmetic
gptkbp:platform	gptkb:TensorFlow ASIC
gptkbp:publiclyAvailableSince	2018 (Cloud TPU)
gptkbp:relatedTo	gptkb:microprocessor gptkb:graphics_card gptkb:FPGA
gptkbp:successor	GPU (for some ML workloads)
gptkbp:supports	bfloat16 float32 int8
gptkbp:usedBy	gptkb:Google_Assistant gptkb:Google_Photos gptkb:Google_Search gptkb:Google_Translate
gptkbp:usedFor	gptkb:artificial_intelligence deep learning neural network training neural network inference
gptkbp:usedIn	gptkb:Google_Data_Centers
gptkbp:bfsParent	gptkb:Google_Data_Centers
gptkbp:bfsLayer	7
https://www.w3.org/2000/01/rdf-schema#label	TPU (Tensor Processing Unit)