Run length encoding aware direct memory access filtering engine for scratchpad enabled multicore processors

• US 10,055,358 B2
• Filed: 03/18/2016
• Issued: 08/21/2018
• Est. Priority Date: 03/18/2016
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Brown, David A., Jain, Rishabh, Idicula, Sam, Schlanger, Erik, Hawkins, David Joseph
• Primary Examiner(s): Sun, Michael
• Application Number:
  US15/074,248
• Publication Number:
  US 20170270053A1
• Time in Patent Office: 886 Days
• Field of Search: None
• CPC Class Codes: G06F 12/02 Addressing or allocation; Relocation program address sequencing G06F9/00; arrangements for selecting an address in a digital store G11C8/00
• CPC Class Codes: G06F 12/1081 for peripheral access to main memory, e.g. direct memory access [DMA]
• CPC Class Codes: G06F 16/221 Column-oriented storage; Management thereof
• CPC Class Codes: G06F 2212/656 Address space sharing
• CPC Class Codes: G06F 9/30036 Instructions to perform operations on packed data, e.g. vector, tile or matrix operations
• CPC Class Codes: G06F 9/3455 using stride

Abstract

Techniques are described herein for efficient movement of data from a source memory to a destination memory. In an embodiment, in response to a particular memory location being pushed into a first register within a first register space, the first set of electronic circuits accesses a descriptor stored at the particular memory location. The descriptor indicates a width of a column of tabular data, a number of rows of tabular data, and one or more tabular data manipulation operations to perform on the column of tabular data. The descriptor also indicates a source memory location for accessing the tabular data and a destination memory location for storing data manipulation result from performing the one or more data manipulation operations on the tabular data. Based on the descriptor, the first set of electronic circuits determines control information indicating that the one or more data manipulation operations are to be performed on the tabular data and transmits the control information, using a hardware data channel, to a second set of electronic circuits to perform the one or more operations. Based on the control information, the second set of electronic circuits retrieve the tabular data from source memory location and apply the one or more data manipulation operations to generate the data manipulation result. The second set of electronic circuits cause the data manipulation result to be stored at the destination memory location.

Claims

1. A method comprising:

in response to a particular memory location being pushed into a first register within a first register space that is accessible by a first set of electronic circuits;

said first set of electronic circuits accessing a descriptor stored at the particular memory location, wherein the descriptor indicates;

a width of a column of tabular data, a number of rows of said column of tabular data, and one or more tabular data manipulation operations to perform on the column of tabular data;

a source memory location for said column of tabular data

a destination memory location for a data manipulation result of said one or more tabular data manipulation operations;

the first set of electronic circuits determining, based on the descriptor, control information of the descriptor;

the first set of electronic circuits transmitting the control information of the descriptor to a second set of electronic circuits;

the second set of electronic circuits determining, based on the control information of the descriptor, control information to retrieve the column of tabular data from the source memory location;

the second set of electronic circuits transmitting, the control information to retrieve the column of tabular data, to a third set of electronic circuits;

based on the control information to retrieve the column of tabular data, the third set of electronic circuits retrieving the column of tabular data from the source memory location;

the third set of electronic circuits, transmitting the column of tabular data to the second set of electronic circuits;

the second set of electronic circuits performing, based on the control information of the descriptor, the one or more tabular data manipulation operations on the column of tabular data to generate the data manipulation result;

the second set of electronic circuits transmitting, the data manipulation result to the first set of electronic circuits;

the first set of electronic circuits causing the data manipulation result to be stored at said destination memory location.

2. The method of claim 1, wherein the control information of the descriptor includes the source memory location, the destination memory location, the width of the column of tabular data, the number of rows of said column of tabular data, and the one or more tabular data manipulation operations.

3. The method of claim 1, wherein one of the one or more tabular data manipulation operations include an ARLE (aligned run-length) decompression operation.

4. The method of claim 3,

wherein the method further includes the second set of electronic circuits storing, based on control information generated for another descriptor, a run address specifying one or more run lengths;

and

wherein the second set of electronic circuits performing said ARLE decompression operation includes retrieving said run from said run address and using said run to decompress said column of tabular data.

5. The method of claim 1, wherein:

the descriptor specifies ARLE decompression operation and a second tabular data manipulation operation; and

wherein the second set of electronic circuits performing, based on the control information of the descriptor, the one or more tabular data manipulation operations includes;

the second set of electronic circuits performing said ARLE decompression operation to generate a first data manipulation result, and

the second set of electronic circuits performing said second tabular data manipulation operation on said first data manipulation result.

6. The method of claim 1, wherein one of the one or more tabular data manipulation operations includes a gather operation.

7. The method of claim 6, further comprising:

the second set of electronic circuits performing the gather operation by;

identifying, based on a bit vector, one or more rows within the column of tabular data that should be stored at the destination memory location.

8. The method of claim 7, wherein each bit within the bit vector corresponds to a particular row within the column of tabular data and wherein each bit indicates whether the corresponding row of the bit within the column of tabular data should be stored at the destination memory location.

9. The method of claim 8, further comprising:

the second set of electronic circuits identifying within the bit vector a consecutive number of bits that indicate that rows within the column of tabular data that correspond to their bits should not be stored at the destination memory location;

the second set of electronic circuits determining, based on the consecutive number of bits and the control information of the descriptor, a total size of tabular data of the corresponding rows of the consecutive number of bits;

the second set of electronic circuits determining whether the total size is greater than a threshold value;

in response to determining that the total size is greater than a threshold value;

the second set of electronic circuits does not determine instructions to retrieve tabular data from the source memory location for rows within the column of tabular data that correspond to the consecutive number of bits.

10. The method of claim 1, wherein one of the one or more tabular data manipulation operations is a scatter operation.

11. The method of claim 1, wherein one of the one or more tabular data manipulation operations is a stride operation.

12. The method of claim 11, further comprising:

the second set of electronic circuits determining, based on a stride value and destination memory location, a new destination memory location for each row of the column of tabular data;

wherein the stride value is included within the control information of the descriptor;

the second set of electronic circuits transmitting the column of tabular data including the new destination memory location for each row within the column of tabular data to the first set of electronic circuits;

the first set of electronic circuits causing each row of the column of tabular data to be stored at a new destination memory location of that row.

13. The method of claim 1, wherein the second set of electronic circuits comprise a first subset of electronic circuits and a second subset of electronic circuits, the method further including:

the first subset of electronic circuits performing first tabular data manipulation operations specified by a first descriptor; and

while the first subset of electronic circuits are performing first tabular data manipulation operations specified by a first descriptor, the second subset of electronic circuits are performing second tabular data manipulation operations specified by a second descriptor.

14. The method of claim 13, the method furthering including the first subset of electronic circuits performing tabular data manipulation operations that are also specified by the second descriptor.