Method and device for simplifying tcon signal processing
US-2019304351-A1 · Oct 3, 2019 · US
US11689443B2 · US · B2
| Field | Value |
|---|---|
| Publication number | US-11689443-B2 |
| Application number | US-202117334703-A |
| Country | US |
| Kind code | B2 |
| Filing date | May 29, 2021 |
| Priority date | May 29, 2021 |
| Publication date | Jun 27, 2023 |
| Grant date | Jun 27, 2023 |
A practical reading order for non-experts. Skip the full description unless you need deep technical detail.
What the patent document calls the invention.
A short plain-language summary of the technical disclosure.
Who owns or filed the patent and who is credited as inventor.
Filing, priority, publication, and grant dates set the timeline.
The legal scope of protection — read this for what is actually claimed.
Technology tags used to group this patent with similar filings.
Prior art links and similar publications in this corpus.
Official abstract text for this publication.
A node mesh contains an originating node and several node groups, each node group consisting of one or more nodes with interfaces connected to other nodes of the node group. Each node of a node group has an associated route table with an association between an applied DC voltage and an output interface to couple the input signal to. When the originating node outputs a DC voltage accompanied by differential signaling, each node in turn directs the DC voltage and differential signaling to an output interface as directed by the node local route table to a local termination in a node, which may be coupled to a training processor of inference processor for machine learning applications.
Opening claim text (preview).
We claim: 1. A node fabric for machine learning, the node fabric comprising: an originating node having a plurality of communication interfaces, each communication interface of the originating node having an input and an output; a plurality of node groups, each node group comprising a plurality of nodes and having at least one communication interface of the plurality of nodes coupled to a communication interface of the originating node, each communications interface of a node in a node group having an input and an output; each node of a node group having a route table initialized with a correspondence between an incoming DC voltage applied to a communication interface input and either an communication interface output or a local termination; the originating node sending a DC voltage superimposed with a differential signal to a node group; where each node of a node group which receives the DC voltage and differential signal is configured to associate the DC voltage with a communication interface from the route table, a node of the node group which receives the DC voltage and differential signal thereafter coupling the DC voltage and differential signal to the associated communication interface output or local termination. 2. The node fabric of claim 1 where the originating node has four communication interfaces. 3. The node fabric of claim 2 where each of the originating node communication interface outputs is coupled to a different node group. 4. The node fabric of claim 1 where at least one node of a node group has four communication interfaces. 5. The node fabric of claim 1 where at least one node group comprises six nodes, each of the six nodes having as many as four communication interfaces. 6. The node fabric of claim 1 where the route table comprises an association between a DC voltage applied to a node communications interface input and one of the node communications interface outputs for coupling the DC voltage and the differential signal. 7. The node fabric of claim 1 where a route table for each node is configured through an out-of-band interface to the node. 8. The node fabric of claim 1 where a route table for each node is initialized to a default route which provides initial connectivity, and where the default routes are subsequently replaced with routes provided through a communication interface input of the node. 9. The node fabric of claim 1 where the route table for a node contains route information for machine learning training during one interval of time and route information for machine learning inference during a different interval of time. 10. The node fabric of claim 1 where at least one node of a node group further comprises a machine learning processor coupled to the local termination of a node. 11. A node fabric comprising: an originating node having a plurality of communication interfaces, at least one communication interface output transmitting a DC voltage accompanied by differential signaling; a plurality of node groups, each node group comprising a plurality of nodes, each node of a node group coupled to at least one other node of the node group through a communication interface; each node of a node group maintaining a route table associating a DC voltage applied to a communication interface input with either a communication interface output or a local termination; whereby the originating node outputs the DC voltage accompanied by differential signaling, and each node receiving the DC voltage and differential signaling is configured to couple through each node of a node group to a local termination node of the node group based on the DC voltage and according to the route table of each node of a node group. 12. The node fabric of claim 11 where the differential signaling conforms to an LVDS standard, and the DC voltage is a common mode voltage added to the differential signaling. 13. The node fabric of claim 11 where a number of unique DC voltages for routing to an end node of a node group is equal to a number of nodes in the node group. 14. The node fabric of claim 13 where the number of DC voltages is six. 15. The node fabric of claim 11 where the differential signaling comprises a stream of digital data containing at least one of training data or inference data. 16. The node fabric of claim 11 where at least one node local termination is coupled to an inference processor or a training processor. 17. The node fabric of claim 11 where the route table contains routes for training during a first interval of time and routes for inferences during a second interval of time. 18. A method for routing differential signaling in a mesh fabric comprising an originating node coupled to a plurality of node groups, each node group comprising a plurality of nodes, each node of a node group having at least one communication interface coupled to other nodes of a node group, each node having a route table containing a relationship between a DC input voltage applied to a communication interface input and a communication interface output, the method comprising: the originating node outputting the DC input voltage accompanied by differential signaling to a node of a node group; each node of the node group coupling an input interface having the DC input voltage accompanied by differential signaling to an output interface according to the route table and the DC input voltage; at least one node of the node group terminating the differential signaling to a processor. 19. The method of claim 18 where the processor is at least one of a training processor or an inference processor.
Learning-based routing, e.g. using neural networks or artificial intelligence · CPC title
Related publications grouped by family.
Answers are generated from the same data shown on this page.