Google TPU
E660962
Google TPU is a custom-designed application-specific integrated circuit (ASIC) developed by Google to accelerate machine learning workloads, particularly deep learning inference and training in its data centers.
All labels observed (3)
| Label | Occurrences |
|---|---|
| Cloud TPU | 1 |
| Google TPU canonical | 1 |
| Google custom processors | 1 |
How this entity was disambiguated
This entity first appeared as the object of triple T7388296 — resolving that mention is where its identity was fixed. The disambiguator weighed these candidate entities and picked the highlighted one (or “None”, minting a new entity). This is how homonymy is resolved: the same surface form can point to different entities.
NED1
Entity disambiguation (via context triple)
gpt-5-mini-2025-08-07
Target entity: Google TPU Context triple: [Graphcore, competition, Google TPU]
-
A.
Google Tensor
Google Tensor is Google's custom-designed system-on-a-chip (SoC) platform created to power Pixel devices with advanced AI and machine learning capabilities.
-
B.
Google Cloud TPU V2
Google Cloud TPU V2 is a second-generation tensor processing unit offered as a cloud service by Google, designed to accelerate large-scale machine learning workloads such as deep neural network training and inference.
-
C.
Google Cloud TPU V3
Google Cloud TPU v3 is a high-performance, third-generation tensor processing unit offered on Google Cloud for accelerating large-scale machine learning and deep learning workloads.
-
D.
Google Cloud TPU V4
Google Cloud TPU v4 is a high-performance, fourth-generation tensor processing unit accelerator offered on Google Cloud, optimized for large-scale machine learning and deep learning workloads.
-
E.
Graphcore
Graphcore is a British semiconductor company that develops specialized intelligence processing units (IPUs) and systems designed to accelerate artificial intelligence and machine learning workloads.
- F. None of above. chosen
- G. Unsure - the case is ambiguous/there is not enough information to decide.
NED2
Entity disambiguation (via description)
gpt-5-mini-2025-08-07
Target entity: Google TPU Target entity description: Google TPU is a custom-designed application-specific integrated circuit (ASIC) developed by Google to accelerate machine learning workloads, particularly deep learning inference and training in its data centers.
-
A.
Google Tensor
Google Tensor is Google's custom-designed system-on-a-chip (SoC) platform created to power Pixel devices with advanced AI and machine learning capabilities.
-
B.
Google Cloud TPU V2
Google Cloud TPU V2 is a second-generation tensor processing unit offered as a cloud service by Google, designed to accelerate large-scale machine learning workloads such as deep neural network training and inference.
-
C.
Google Cloud TPU V3
Google Cloud TPU v3 is a high-performance, third-generation tensor processing unit offered on Google Cloud for accelerating large-scale machine learning and deep learning workloads.
-
D.
Google Cloud TPU V4
Google Cloud TPU v4 is a high-performance, fourth-generation tensor processing unit accelerator offered on Google Cloud, optimized for large-scale machine learning and deep learning workloads.
-
E.
Graphcore
Graphcore is a British semiconductor company that develops specialized intelligence processing units (IPUs) and systems designed to accelerate artificial intelligence and machine learning workloads.
- F. None of above. chosen
Statements (46)
| Predicate | Object |
|---|---|
| instanceOf |
application-specific integrated circuit
ⓘ
tensor processing unit ⓘ |
| announcedAt | Google I/O 2016 NERFINISHED ⓘ |
| architecture | matrix-multiply optimized architecture ⓘ |
| availableVia | Google Cloud NERFINISHED ⓘ |
| category |
data center hardware
ⓘ
machine learning hardware ⓘ |
| competesWith |
AMD GPU
NERFINISHED
ⓘ
NVIDIA GPU NERFINISHED ⓘ custom AI accelerators ⓘ |
| designedFor |
high throughput matrix operations
ⓘ
low-precision arithmetic ⓘ |
| developer | Google ⓘ |
| firstAnnouncement | 2016 ⓘ |
| firstPublicUse | 2015 ⓘ |
| hasVersion |
TPU v1
NERFINISHED
ⓘ
TPU v2 NERFINISHED ⓘ TPU v3 NERFINISHED ⓘ TPU v4 NERFINISHED ⓘ TPU v5e NERFINISHED ⓘ TPU v5lite NERFINISHED ⓘ TPU v5p NERFINISHED ⓘ |
| integratedWith | Google Cloud AI Platform NERFINISHED ⓘ |
| manufacturer | Google NERFINISHED ⓘ |
| notableFeature |
high performance per watt for ML workloads
ⓘ
systolic array matrix unit ⓘ |
| offeredAs | Cloud TPU NERFINISHED ⓘ |
| optimizedFor | large-scale data center deployment ⓘ |
| purpose |
accelerate deep learning inference
ⓘ
accelerate deep learning training ⓘ accelerate machine learning workloads ⓘ |
| region | global deployment in Google data centers ⓘ |
| supports |
TensorFlow
NERFINISHED
ⓘ
neural network inference ⓘ neural network training ⓘ |
| supportsDataType |
32-bit floating point
ⓘ
8-bit integer ⓘ bfloat16 ⓘ |
| technologyNode | advanced CMOS process ⓘ |
| usedBy |
external Google Cloud customers
ⓘ
internal Google services ⓘ |
| usedFor |
Google Assistant
NERFINISHED
ⓘ
Google Photos NERFINISHED ⓘ Google Search NERFINISHED ⓘ Google Translate NERFINISHED ⓘ |
| usedIn | Google data centers ⓘ |
How these facts were elicited
The pipeline generated the facts above by prompting gpt-5.1 with this entity's name + description and the instruction below.
Instruction
You are a knowledge base construction expert. Given a subject entity and a description of it, return factual statements that you know for the subject as a JSON list of dictionaries(triples), where keys must be "subject", "predicate" and "object". The number of facts may be very high, between 25 to 50 or more, for very popular subjects. For less popular subjects, the number of facts can be very low, like 5 or 10. # Requirements - If you don't know the subject at all, return an empty list. - If the subject is not a named entity, return an empty list. - Include at least one triple where predicate is "instanceOf". - Do not get too wordy. - Separate several objects into multiple triples with one object.
Input
Subject: Google TPU Description of subject: Google TPU is a custom-designed application-specific integrated circuit (ASIC) developed by Google to accelerate machine learning workloads, particularly deep learning inference and training in its data centers.
Referenced by (3)
Full triples — surface form annotated when it differs from this entity's canonical label.
this entity surface form:
Cloud TPU
this entity surface form:
Google custom processors