# US 2019/0244325 A1 · Task execution in a simd processing unit with parallel groups of processing lanes

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2019/0244325 A1
- Document ID: US20190244325A1
- Status: Published application
- Kind: A1
- Application number: 16/386,840
- Filed: 2019-04-17
- Published: 2019-08-08
- Earliest claimed: 2013-12-18
- Assignees: Imagination Technologies Limited
- Applicants: Imagination Technologies Limited
- Inventors: John Howson; Jonathan Redshaw; Yoong Chert Foo
- Web page: https://patentgrep.com/patent/US20190244325A1

## CPC

- G06T 1/20: Processor architectures; Processor configuration, e.g. pipelining
- G06F 9/30036: {Instructions to perform operations on packed data, e.g. vector, tile or matrix operations}
- G06F 9/30038: {using a mask}
- G06F 9/3822: {Parallel decoding, e.g. parallel decode units}
- G06F 9/3836: {Instruction issuing, e.g. dynamic instruction scheduling or out of order instruction execution}
- G06F 9/3887: {controlled by a single instruction for multiple data lanes [SIMD]}
- G06F 9/38885: {Divergence aspects}
- G06F 15/8007: {single instruction multiple data [SIMD] multiprocessors}
- G06F 9/4881: {Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues}
- G06F 2209/507: Low-level
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A SIMD processing unit processes a plurality of tasks which each include up to a predetermined maximum number of work items. The work items of a task are arranged for executing a common sequence of instructions on respective data items. The data items are arranged into blocks, with some of the blocks including at least one invalid data item. Work items which relate to invalid data items are invalid work items. The SIMD processing unit comprises a group of processing lanes configured to execute instructions of work items of a particular task over a plurality of processing cycles. A control module assembles work items into the tasks based on the validity of the work items, so that invalid work items of the particular task are temporally aligned across the processing lanes. In this way the number of wasted processing slots due to invalid work items may be reduced.

## Claims

### Claim 1

A processing unit configured to process a plurality of tasks which each include one or more work items, wherein the work items of a task are arranged for executing instructions on respective data items, the data items being arranged into blocks of data items, wherein blocks of work items within a task relate to respective blocks of data items, wherein if one or more of the data items within a block of data items is to be processed then all of the data items within that block of data items are scheduled for processing by the processing unit, and wherein one or more of the blocks of work items include at least one invalid work item, the processing unit comprising:
a group of processing lanes configured to execute instructions of work items of a particular task over a plurality of processing cycles, wherein each of the processing lanes of the group is configured to execute instructions of a respective block of work items over a plurality of consecutive processing cycles; and
a control module configured to assemble the work items into the tasks based on the validity of the work items.

### Claim 2 (depends on 1)

The processing unit of claim 1, further comprising a logic module coupled to the group of processing lanes configured to cause the group of processing lanes to skip the execution of a set of invalid work items if the set of invalid work items are the only work items scheduled for execution over the group of processing lanes in a processing cycle.

### Claim 3 (depends on 1)

The processing unit of claim 1, further comprising a logic module coupled to the group of processing lanes configured to cause the group of processing lanes to skip a processing cycle if there are no valid work items scheduled for execution over the group of processing lanes in the processing cycle.

### Claim 4 (depends on 1)

The processing unit of claim 1, wherein the control module is configured to set indicators to indicate how the work items have been assembled into the tasks.

### Claim 5 (depends on 1)

The processing unit of claim 1, wherein the control module is configured to assemble the work items into the tasks such that work items of a block of work items are grouped together into the same task.

### Claim 6 (depends on 1)

The processing unit of claim 1, wherein the control module is configured to assemble the work items into the tasks so that blocks of work items are grouped together into tasks based on the number of invalid work items in the respective blocks of work items.

### Claim 7 (depends on 1)

The processing unit of claim 1, wherein the control module is configured to assemble the work items into the tasks so that work items within a block of work items are re-ordered to thereby align the invalid work items from different blocks of work items within a task.

### Claim 8 (depends on 7)

The processing unit of claim 7, wherein the control module is configured to re-order work items within a block of work items by performing at least one of a rotation operation and a swapping operation of the work items within the block of work items.

### Claim 9 (depends on 8)

The processing unit of claim 8, wherein the control module is configured to set indicators to indicate how the work items have been assembled into the tasks, wherein the control module is configured to set a respective indicator for a plurality of the blocks of work items to indicate the order of the work items within the plurality of blocks of work items.

### Claim 10 (depends on 9)

The processing unit of claim 9, wherein each block of work items comprises four work items, and wherein each indicator comprises three bits: two bits to define a rotation of the work items within the block of work items, and one bit to define whether a swapping operation has been performed on two of the work items at predetermined positions within the block of work items.

### Claim 11 (depends on 1)

The processing unit of claim 1, wherein the control module is configured to assemble the work items into the tasks based on the validity of the work items such that invalid work items of the particular task are temporally aligned across the group of processing lanes.

### Claim 12 (depends on 1)

The processing unit of claim 1, wherein there are more than two levels of validity for the work items, and wherein the control module is configured to assemble the work items into the tasks, based on the validity of the work items, so that work items of the particular task which have the same level of validity are temporally aligned across the group of processing lanes.

### Claim 13 (depends on 1)

The processing unit of claim 1, wherein the data items are pixel values, and wherein the blocks of data items are pixel quads.

### Claim 14 (depends on 1)

The processing unit of claim 1, wherein some of the tasks comprise fewer than a predetermined maximum number of work items, said processing unit further comprising:
a plurality of groups of processing lanes, each group being configured to execute instructions of work items of a respective task in parallel over a plurality of processing cycles; and
a logic module coupled to the groups of processing lanes configured to cause a particular group of processing lanes to skip a particular processing cycle, independently of the other groups of processing lanes, if there are no work items scheduled for execution in any of the processing lanes of the particular group in the particular processing cycle.

### Claim 15 (depends on 1)

The processing unit of claim 1, wherein the processing unit is a single instruction multiple data (SIMD) processing unit, and wherein each of the plurality of tasks includes up to a predetermined maximum number of work items, and wherein the work items of a task are arranged for executing a common sequence of instructions on respective data items.

### Claim 16 (depends on 1)

The processing unit of claim 1, wherein invalid work items relate to invalid data items.

### Claim 17

A method of using a processing unit to process a plurality of tasks which each include one or more work items, wherein the work items of a task are arranged for executing instructions on respective data items, the data items being arranged into blocks of data items, wherein blocks of work items within a task relate to respective blocks of data items, wherein if one or more of the data items within a block of data items is to be processed then all of the data items within that block of data items are scheduled for processing by the processing unit, and wherein one or more of the blocks of work items include at least one invalid work item, and wherein the processing unit comprises a group of processing lanes configured to execute instructions of work items of a particular task over a plurality of processing cycles, the method comprising:
assembling the work items into the tasks based on the validity of the work items; and
executing instructions of work items of the particular task using the group of processing lanes, wherein each of the processing lanes of the group executes instructions of a respective block of work items over a plurality of consecutive processing cycles.

### Claim 18 (depends on 17)

The method of claim 17, further comprising causing the group of processing lanes to skip a processing cycle if there are no valid work items scheduled for execution over the group of processing lanes in the processing cycle.

### Claim 19 (depends on 17)

The method of claim 17, wherein the work items are assembled into the tasks based on the validity of the work items such that invalid work items of the particular task are temporally aligned across the group of processing lanes.

### Claim 20

A non-transitory computer readable storage medium having stored thereon processor executable instructions that when executed cause at least one processor to generate a processing unit which is configured to process a plurality of tasks which each include one or more work items, wherein the work items of a task are arranged for executing instructions on respective data items, the data items being arranged into blocks of data items, wherein blocks of work items within a task relate to respective blocks of data items, wherein if one or more of the data items within a block of data items is to be processed then all of the data items within that block of data items are scheduled for processing by the processing unit, and wherein one or more of the blocks of work items include at least one invalid work item, the processing unit comprising:
a group of processing lanes configured to execute instructions of work items of a particular task over a plurality of processing cycles, wherein each of the processing lanes of the group is configured to execute instructions of a respective block of work items over a plurality of consecutive processing cycles; and
a control module configured to assemble the work items into the tasks based on the validity of the work items.
