IP Library › Granted Patent US 7,720,377
Granted Patent B2
US 7,720,377 · App. 11/337,328 · Granted May 18, 2010

Compute clusters employing photonic interconnections for transmitting optical signals between compute cluster nodes

Assignee: Hewlett-Packard Development Company, L.P.
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 7,720,377
App. No.
11/337,328
Filed
Jan 23, 2006
Granted
May 18, 2010
Kind
B2
Examiner
PHAN, HANH
Art Unit
2613
USPC
398/45
Abstract

Various embodiments of the present invention are directed to photonic-interconnection-based compute clusters that provide high-speed, high-bandwidth interconnections between compute cluster nodes. In one embodiment of the present invention, the compute cluster includes a photonic interconnection having one or more optical transmission paths for transmitting independent frequency channels within an optical signal to each node in a set of nodes. The compute cluster includes one or more photonic-interconnection-based writers, each writer associated with a particular node, and each writer encoding information generated by the node into one of the independent frequency channels. A switch fabric directs the information encoded in the independent frequency channels to one or more nodes in the compute cluster. The compute cluster also includes one or more photonic-interconnection-based readers, each reader associated with a particular node, and each reader extracting the information encoded in the independent frequency channels directed to the node for processing.

Claims (39)

1. A compute cluster comprising:

a photonic interconnection having one or more branching optical transmission paths for transmitting independent frequency channels within an optical signal to each node in a set of nodes;

one or more photonic-interconnection-based writers, each writer associated with a particular node and disposed adjacent to one of the branching optical transmission paths, each writer enodes information generated by the associated node into one of a first subset of the independent frequency channels that travel in the transmission paths in a first direction;

a switch fabric that receives the first subset of independent frequency channels and encodes the information encoded in the first subset into a second subset of the independent frequency channels and sends the second subset on the transmission paths to the one or more nodes in a second direction, the second direction opposite the first direction; and

one or more photonic-interconnection-based readers, each reader associated with a particular node and disposed adjacent to one of the branching optical transmission paths, each reader extracts the information encoded in one of the encoded independent frequency channels of the second subset directed to the associated node for processing.

2. The compute cluster of claim 1 further comprising a clock signal source.

3. The compute cluster of claim 2 wherein the switch fabric further comprises one or more cyclic permutation networks for distributing continuous bit streams of information and one or more virtual output queues for queuing packets of information into one or more buffers.

4. The compute cluster of claim 1 wherein the nodes further comprise any one of a processor, memory, computer server, storage server, an external network connection, a data transmitting device or any electrical circuit or mosaic of electrical circuits.

5. The compute cluster of claim 4 where the electrical circuits have microscale dimensions or nanoscale dimensions.

6. The compute cluster of claim 1 wherein the photonic interconnection further comprises a two-dimensional photonic crystal with waveguides, a number of independent optical fibers, or free space.

7. The compute cluster of claim 1 wherein the photonic-interconnection-based writer further comprises:

a drop filter that extracts a frequency channel from the optical transmission path,

a local waveguide that extracts the frequency channel from the drop filter and transmits the frequency channel,

a modulator that extracts the frequency channel from the local waveguide and produces a modulated frequency channel by modulating the frequency channel, and

an add filter that extracts the modulated frequency channel and inserts the modulated frequency channel into the optical transmission path.

8. The compute cluster of claim 1 wherein the photonic-interconnection-based reader further comprises;

a drop filter that extracts a frequency channel from the optical transmission path, and

a detector that extracts the frequency channel from the drop filter and demodulates the information encoded in the frequency channel.

9. A switch fabric comprising:

one or more optical transmission paths that each transmit a first set of independent frequency channels and a second set of independent frequency channels, the first set of independent frequency channels encoding information output by nodes in a compute cluster and travel in the one or more optical transmission paths in a first direction;

one or more photonic-interconnection-based readers, each reader extracting a frequency channel in the first set of independent frequency channels and encoding the information into one or more electronic input bit streams;

an electronic-based switch fabric that partitions and assembles the electronic input bit streams into electronic output bit streams, each electronic output bit stream directed to a particular node in the compute cluster; and

one or more photonic-interconnection-based writers, each encoding the information encoded in the electronic output bit streams into frequency channels in the second set of independent frequency channels, the second set travel in the one or more optical transmission paths in a second direction, the second direction opposite the first direction.

10. The switch fabric of claim 9 further comprising a clock signal source.

11. The switch fabric of claim 10 wherein the switch fabric further comprises one or more cyclic permutation networks for distributing continuous bit streams of information and one or more virtual output queues for queuing packets of information into one or more buffers.

12. The switch fabric of claim 9 wherein the nodes further comprise any one of a processor, memory, computer server, storage server, an external network connection, a data transmitting device or any electrical circuit or mosaic of electrical circuits.

13. The switch fabric of claim 12 where the electrical circuits have microscale dimensions or nanoscale dimensions.

14. The switch fabric of claim 9 wherein the photonic interconnection further comprises a two-dimensional photonic crystal with waveguides, a number of independent optical fibers, or free space.

15. A method for transmitting data between nodes in a compute cluster, the method comprising:

providing a photonic interconnection having one or more branching transmission paths that lead to each node in the computer cluster;

generating a first set of independent frequency channels and a second set of independent frequency channels, both sets of independent frequency channels are transmitted in the optical transmission paths to the nodes in opposite directions;

encoding information generated by the nodes into the frequency channels in the first set of independent frequency channels;

transmitting the encoded information to a switch fabric that partitions and assembles the information encoded in the first set of frequency channels and encodes the assembled information into the second set of independent frequency channels; and

directing the second set of independent frequency channels to the nodes so that the assembled information can be processed.

16. The method of claim 15 further comprising providing a clock signal that synchronizes transmission of information between nodes in the compute cluster.

17. The method of claim 16 wherein providing the switch fabric further comprises providing one or more cyclic permutation networks for distributing continuous bit streams of data and one or more virtual output queues for queuing packets of data into one or more buffers.

18. The method of claim 15 wherein the node further comprises any one of a processor, memory, computer server, storage server, an external network connection, a data transmitting device or any electrical circuit or mosaic of electrical circuits.

19. The method of claim 18 where the electrical circuits have microscale dimensions or nanoscale dimensions.

20. The method of claim 15 wherein the photonic interconnection further comprises a two-dimensional photonic crystal with waveguides, a number of independent optical fibers, or free space.

Assignments (2)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Nov 9, 2015
From: HEWLETT-PACKARD DEVELOPMENT COMPANY, L.P.
To: HEWLETT PACKARD ENTERPRISE DEVELOPMENT LP
Reel/Frame 037079/0001 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jan 23, 2006
From: SNIDER, GREGORY S.; BEAUSOLEIL, RAYMOND
To: HEWLETT-PACKARD DEVELOPMENT COMPANY, L.P.
Reel/Frame 017504/0654 →
Continuity (1)
Related Publication 20070172235A1 · Jul 26, 2007