Synchronization of computation engines with non-blocking instructions

A computer-implemented method, comprising:

• receiving, at a computing device, an input data set including operations to be performed by an integrated circuit device and dependencies between the operations;
• identifying an operation in the input data set that includes a non-blocking instruction that is to be performed by an execution engine of the integrated circuit device, wherein the execution engine initiates any instruction subsequent to the non-blocking instruction without waiting for the non-blocking instruction to complete;
• identifying an operation in the input data set that requires that the non-blocking instruction be completed;
• generating instructions for performing the operation including a particular instruction to wait for a value to be set in a register of the integrated circuit device, wherein the particular instruction occurs first in the instructions for performing the operation;
• generating program code including the non-blocking instruction and the instructions for performing the operation, wherein the non-blocking instruction is configured to set the value in the register;
• and
• setting a particular index in a fixed-size array, each index of the fixed-size array corresponding to a register of the integrated circuit device, wherein the particular index corresponds to the register set by the non-blocking instruction, and wherein values stored in the fixed-size array indicate whether a respective register is available for use by a wait-on-event instruction.

A computing system, comprising:

• one or more processors;
• and
• a non-transitory computer-readable medium including instructions that, when executed by the one or more processors, cause the one or more processors to perform operations including:
• receiving an input data set including operations to be performed by an integrated circuit device and dependencies between the operations;
• identifying an operation in the input data set that includes a non-blocking instruction is to be performed by an execution engine of the integrated circuit device, wherein the execution engine initiates any instruction subsequent to the non-blocking instruction without waiting for the non-blocking instruction to complete;
• identifying an operation in the input data set that is to be performed by the execution engine, wherein the operation requires that the non-blocking instruction be completed;
• generating instructions for performing the operation including a wait-on-event instruction, wherein the wait-on-event instruction waits for a value to be set in a register of the integrated circuit device;
• generating program code including the non-blocking instruction followed by the instructions for performing the operation, wherein the non-blocking instruction is configured to set the value in the register;
• and
• setting a particular index in a fixed-size array, each index of the fixed-size array corresponding to a register of the integrated circuit device, wherein the particular index corresponds to the register set by the non-blocking instruction, and wherein values stored in the fixed-size array indicate whether a respective register is available for use by a wait-on-event instruction.