Spatial tiling of compute arrays with shared control

1. A system for machine learning acceleration, comprising:

a plurality of tensor processor clusters, each comprising:

a plurality of tensor processors; and

a cluster-level controller configured to:

receive a multi-cycle instruction, wherein each of the plurality of tensor processor clusters receives a respective multi-cycle instruction, and wherein the respective multi-cycle instructions are distributed across the plurality of tensor processor clusters in accordance with single-program-multiple-data (SPMD) parallelism such that at least two of the plurality of tensor processor clusters receive and execute different multi-cycle instructions while operating on an input feature map;

determine, based on the multi-cycle instruction, (1) a sequence of vector operations to be executed by the tensor processors and (2) address information usable to determine a respective spatial partition of an input tensor on which each tensor processor is to operate when performing each vector operation; and

for each vector operation in the sequence:

generate, based on the address information, a common address offset, relative to a respective base address associated with each tensor processor, at which each tensor processor is to retrieve the respective spatial partition of the input tensor on which the tensor processor is to operate;

multicast the common address offset to the tensor processors; and

control the tensor processors to execute the vector operation in lock step.

2. The system of claim 1 , wherein:

a first multi-cycle instruction received by a given one of the plurality of tensor processor clusters represents a portion of a machine-learning program;

the machine-learning program comprises a plurality of multi-cycle instructions, each of which is associated with one or more convolution operations to be performed in a respective layer in a convolutional neural network; and

a first cycle of the first multi-cycle instruction is associated with a first convolution operation.

3. The system of claim 2 , wherein:

the cluster-level controller in the given cluster is configured to:

determine, based on the first multi-cycle instruction, (1) a first sequence of vector operations to be executed by the tensor processors in the given cluster and (2) address information usable to determine the respective spatial partition of an input tensor on which each tensor processor in the given cluster is to operate when performing each vector operation in the first sequence of vector operations; and

for each vector operation in the first sequence of vector operations:

generate, based on the address information, a common address offset, relative to a respective base address associated with each tensor processor in the given cluster, at which each tensor processor in the given cluster is to retrieve the respective spatial partition of the input tensor on which the tensor processor is to operate;

multicast the common address offset to the tensor processors in the given cluster; and

control the tensor processors in the given cluster to execute the vector operation in lock step;

each of the vector operations in the first sequence of vector operations comprises one or more of: a vector read operation, a vector addition operation, and a vector multiply operation; and

each tensor processor in the given cluster comprises a hardware compute array of multiply-and-accumulate (MAC) computation units configured to execute vector operations on the respective spatial partition of the input tensor on which the tensor processor is to operate when performing each vector operation in the first sequence of vector operations.

4. The system of claim 3 , wherein:

the cluster-level controller in the given cluster is further configured to:

determine, based on the first multi-cycle instruction, (3) weight information usable to determine weights to be applied in the one or more convolution operations associated with the first multi-cycle instruction;

determine, for at least one vector operation in the first sequence of vector operations and dependent on the weight information, a first subset of the weights associated with the first multi-cycle instruction that are associated with the first convolution operation; and

provide the first subset of the weights that are associated with the first convolution operation to the hardware compute array of at least one of the tensor processors in the given cluster for execution of the first convolution operation in the first cycle of the first multi-cycle instruction.

5. The system of claim 4 , wherein:

each tensor processor in the given cluster is configured to generate a respective spatial partition of an output tensor based on the respective spatial partition of the input tensor on which each tensor processor in the given cluster is to operate using single-instruction-multiple-data (SIMD) parallelism;

to implement SIMD parallelism, each tensor processor in the given cluster is configured to implement data parallelism; and

the cluster-level controller is further configured to provide the first subset of the weights that are associated with the first convolution operation to the hardware compute arrays of two or more of the tensor processors in the given cluster for execution of the first convolution operation in the first cycle of the first multi-cycle instruction.

6. The system of claim 5 , wherein:

a second cycle of the first multi-cycle instruction is associated with a second convolution operation; and

the cluster-level controller in the given cluster is further configured to:

determine, for at least one vector operation in the first sequence of vector operations and dependent on the weight information, a second subset of the weights associated with the first multi-cycle instruction that are associated with the second convolution operation; and

provide the second subset of the weights that are associated with the second convolution operation to the hardware compute arrays of the two or more tensor processors in the given cluster for execution of the second convolution operation in the second cycle of the first multi-cycle instruction.

7. The system of claim 4 , wherein the cluster-level controller in the given cluster is further configured to:

determine, for at least one vector operation in the first sequence of vector operations and dependent on the weight information, a second subset of the weights associated with the first multi-cycle instruction that are associated with the first convolution operation; and

provide the second subset of the weights that are associated with the first convolution operation to the hardware compute array of one of the tensor processors in the given cluster other than the at least one of the one or more tensor processors in the given cluster for execution of the first convolution operation in the first cycle of the first multi-cycle instruction.

8. The system of claim 2 , wherein the cluster-level controller in the given cluster is further configured to:

receive a second multi-cycle instruction;

determine, based on the second multi-cycle instruction, (1) a second sequence of vector operations to be executed by the tensor processors in the given cluster and (2) address information usable to determine the respective spatial partition of an input tensor on which each tensor processor in the given cluster is to operate when performing each vector operation in the second sequence of vector operations; and

for each vector operation in the second sequence of vector operations:

generate, based on the address information, a common address offset, relative to a respective base address associated with each tensor processor in the given cluster, at which each tensor processor is to retrieve the respective spatial partition of the input tensor on which the tensor processor is to operate;

multicast the common address offset to the tensor processors in the given cluster; and

control the tensor processors in the given cluster to execute the vector operation in lock step.

9. The system of claim 1 , wherein:

the respective spatial partitions of the input tensor on which each tensor processor in each cluster are to operate are subsets of the input feature map;

each cluster receives a respective multi-cycle instruction;

the respective multi-cycle instructions are distributed across the clusters in accordance with single-instruction-multiple-data (SIMD) parallelism such that at least two of the clusters receive and execute a same multi-cycle instruction while operating on a different subset of the input feature map; and

the at least two clusters operate in parallel and in lock step to execute the same multi-cycle instruction.

10. A method for machine learning acceleration, comprising:

receiving, by a cluster-level controller of each of a plurality of tensor processor clusters that includes a plurality of tensor processors, a first multi-cycle instruction, wherein each of the plurality of tensor processor clusters receives a respective multi-cycle instruction, and wherein the respective multi-cycle instructions are distributed across the plurality of tensor processor clusters in accordance with single-program-multiple-data (SPMD) parallelism such that at least two of the plurality of tensor processor clusters receive and execute different multi-cycle instructions while operating on an input feature map;

determining, based on the first multi-cycle instruction, (1) a first sequence of vector operations to be executed by the tensor processors and (2) address information usable to determine a respective spatial partition of an input tensor on which each tensor processor is to operate when performing each vector operation in the first sequence; and

for each vector operation in the first sequence:

generating, based on the address information, a common address offset, relative to a respective base address associated with each tensor processor, at which each tensor processor is to retrieve the respective spatial partition on which the tensor processor is to operate;

multicasting the common address offset to the tensor processors; and

controlling the tensor processors to execute the vector operation in lock step.

11. The method of claim 10 , wherein:

the first multi-cycle instruction represents a portion of a machine-learning program;

the machine-learning program comprises a plurality of multi-cycle instructions, each of which is associated with one or more convolution operations to be performed in a respective layer in a convolutional neural network;

a first cycle of the first multi-cycle instruction is associated with a first convolution operation; and

each of the vector operations in the first sequence of vector operations comprises one or more of: a vector read operation, a vector addition operation, and a vector multiply operation.

12. The method of claim 11 , further comprising:

determining, based on the first multi-cycle instruction, (3) weight information usable to determine weights to be applied in the one or more convolution operations associated with the first multi-cycle instruction;

determining, for at least one vector operation in the first sequence of vector operations and dependent on the weight information, a first subset of the weights associated with the first multi-cycle instruction that are associated with the first convolution operation; and

providing the first subset of the weights that are associated with the first convolution operation to a hardware compute array of at least one of the tensor processors in the cluster for execution of the first convolution operation in the first cycle of the first multi-cycle instruction.

13. The method of claim 12 , wherein:

each tensor processor in the cluster is configured to generate a respective spatial partition of an output tensor based on the respective spatial partition of the input tensor on which each tensor processor in the cluster is to operate using single-instruction-multiple-data (SIMD) parallelism;

a second cycle of the first multi-cycle instruction is associated with a second convolution operation; and

the method further comprises:

providing the first subset of the weights that are associated with the first convolution operation to the hardware compute arrays of two or more of the tensor processors in the cluster for execution of the first convolution operation in the first cycle of the first multi-cycle instruction;

determining, for at least one vector operation in the first sequence of vector operations and dependent on the weight information, a second subset of the weights associated with the first multi-cycle instruction that are associated with the second convolution operation; and

providing the second subset of the weights that are associated with the second convolution operation to the hardware compute arrays of the two or more tensor processors in the cluster for execution of the second convolution operation in the second cycle of the first multi-cycle instruction.

14. The method of claim 12 , further comprising:

determining, for at least one vector operation in the first sequence of vector operations and dependent on the weight information, a second subset of the weights associated with the first multi-cycle instruction that are associated with the first convolution operation; and

providing the second subset of the weights that are associated with the first convolution operation to the hardware compute array of one of the tensor processors in the cluster other than the at least one of the one or more tensor processors in the cluster for execution of the first convolution operation in the first cycle of the first multi-cycle instruction.

15. One or more computer-readable non-transitory storage media embodying software that is operable when executed to:

receive, at a cluster-level controller of each of a plurality of tensor processor clusters that includes a plurality of tensor processors, a first multi-cycle instruction, wherein each of the plurality of tensor processor clusters receives a respective multi-cycle instruction, and wherein the respective multi-cycle instructions are distributed across the plurality of tensor processor clusters in accordance with single-program-multiple-data (SPMD) parallelism such that at least two of the plurality of tensor processor clusters receive and execute different multi-cycle instructions while operating on an input feature map;

determine, based on the first multi-cycle instruction, (1) a first sequence of vector operations to be executed by the tensor processors and (2) address information usable to determine a respective spatial partition of an input tensor on which each tensor processor is to operate when performing each vector operation in the first sequence; and

for each vector operation in the first sequence:

generate, based on the address information, a common address offset, relative to a respective base address associated with each tensor processor, at which each tensor processor is to retrieve the respective spatial partition on which the tensor processor is to operate;

multicast the common address offset to the tensor processors; and

control the tensor processors to execute the vector operation in lock step.

16. The media of claim 15 , wherein:

the first multi-cycle instruction represents a portion of a machine-learning program;

the machine-learning program comprises a plurality of multi-cycle instructions, each of which is associated with one or more convolution operations to be performed in a respective layer in a convolutional neural network;

a first cycle of the first multi-cycle instruction is associated with a first convolution operation;

each of the vector operations in the first sequence of vector operations comprises one or more of: a vector read operation, a vector addition operation, and a vector multiply operation; and

the software is further operable when executed to:

determine, based on the first multi-cycle instruction, (3) weight information usable to determine weights to be applied in the one or more convolution operations associated with the first multi-cycle instruction;

determine, for at least one vector operation in the first sequence of vector operations and dependent on the weight information, a first subset of the weights associated with the first multi-cycle instruction that are associated with the first convolution operation; and

provide the first subset of the weights that are associated with the first convolution operation to a hardware compute array of at least one of the tensor processors in the cluster for execution of the first convolution operation in the first cycle of the first multi-cycle instruction.

17. The media of claim 16 , wherein:

each tensor processor in the cluster is configured to generate a respective spatial partition of an output tensor based on the respective spatial partition of the input tensor on which each tensor processor in the cluster is to operate using single-instruction-multiple-data (SIMD) parallelism;

a second cycle of the first multi-cycle instruction is associated with a second convolution operation; and

the software is further operable when executed to:

provide the first subset of the weights that are associated with the first convolution operation to the hardware compute arrays of two or more of the tensor processors in the cluster for execution of the first convolution operation in the first cycle of the first multi-cycle instruction;

determine, for at least one vector operation in the first sequence of vector operations and dependent on the weight information, a second subset of the weights associated with the first multi-cycle instruction that are associated with the second convolution operation; and

provide the second subset of the weights that are associated with the second convolution operation to the hardware compute arrays of the two or more tensor processors in the cluster for execution of the second convolution operation in the second cycle of the first multi-cycle instruction.

18. The media of claim 17 , wherein the software is further operable when executed to:

determine, for at least one vector operation in the first sequence of vector operations and dependent on the weight information, a second subset of the weights associated with the first multi-cycle instruction that are associated with the first convolution operation; and

provide the second subset of the weights that are associated with the first convolution operation to the hardware compute array of one of the tensor processors in the cluster other than the at least one of the one or more tensor processors in the cluster for execution of the first convolution operation in the first cycle of the first multi-cycle instruction.