Device with data processing engine array that enables partial reconfiguration
A device may include a processor system and an array of data processing engines (DPEs) communicatively coupled to the processor system. Each of the DPEs includes a core and a DPE interconnect. The processor system is configured to transmit configuration data to the array of DPEs, and each of the DPEs is independently configurable based on the configuration data received at the respective DPE via the DPE interconnect of the respective DPE. The array of DPEs enable, without modifying operation of a first kernel of a first subset of the DPEs of the array of DPEs, reconfiguration of a second subset of the DPEs of the array of DPEs.
1 . An integrated circuit (IC) comprising:
a graphics processing unit (GPU);
an array of data processing engines (DPEs) communicatively coupled to the GPU, each of the DPEs comprising a data processing circuit and a DPE interconnect for communicating with other DPEs in the array; and
a network on chip (NoC), wherein the GPU and the array of DPEs are configured to communicate through the NoC.
2 . The IC of claim 1 , wherein the DPE interconnect of each of the DPEs further comprises:
a memory mapped switch configured to communicate via memory mapped transactions with a neighboring DPE in the array.
3 . The IC of claim 2 , wherein the DPE interconnect of each of the DPEs comprises:
a stream switch operable to be configured to communicate with a neighboring DPE in the array.
4 . The IC of claim 1 further comprising:
a system interface circuit coupled to the NoC and to the array of DPEs.
5 . The IC of claim 4 , wherein the system interface circuit comprises tiles, each tile being coupled to at least one column of DPEs of the array of DPEs.
6 . The IC of claim 5 , wherein the array of DPEs communicate to the NoC through the tiles of the system interface circuit.
7 . The IC of claim 4 , wherein the GPU and the array of DPEs are configured to communicate through the system interface circuit.
8 . The IC of claim 1 , wherein each of the DPEs further comprises a core, which includes the data processing circuit, and a memory module.
9 . The IC of claim 1 , wherein the array of DPEs are configured to perform a machine learning operation.
10 . An IC comprising:
a graphics processing unit (GPU);
an array of data processing engines (DPEs) communicatively coupled to the GPU, each of the DPEs comprising a core and a DPE interconnect for communicating with other DPEs in the array; and
a network on chip (NoC), wherein the GPU and the array of DPEs are configured to communicate through the NoC.
11 . The IC of claim 10 , wherein the DPE interconnect of each of the DPEs further comprises:
a memory mapped switch configured to communicate via memory mapped transactions with a neighboring DPE in the array.
12 . The IC of claim 11 , wherein the DPE interconnect of each of the DPEs comprises:
a stream switch operable to be configured to communicate with a neighboring DPE in the array.
13 . The IC of claim 10 further comprising:
a network on chip (NoC) coupled to a processor system; and
a system interface circuit coupled to the NoC and to the array of DPEs.
14 . The IC of claim 13 , wherein the system interface circuit comprises tiles, each tile being coupled to at least one column of DPEs of the array of DPEs.
15 . The IC of claim 14 , wherein the array of DPEs communicate to the NoC through the tiles of the system interface circuit.
16 . The IC of claim 13 , wherein the GPU and the array of DPEs are configured to communicate through the NoC and the system interface circuit.
17 . The IC of claim 10 , wherein each of the DPEs further comprises a core, which includes data processing circuitry, and a memory module.
18 . An IC comprising:
a network on chip (NoC);
a graphics processing unit (GPU); and
an array of data processing engines (DPEs) communicatively coupled to the GPU through the NoC, each of the DPEs comprising a data processing circuit and a memory mapped switch coupled to memory mapped switches in other DPEs in the array.
19 . The IC of claim 18 , wherein each of the DPEs further comprises:
a stream switch configured to stream data with stream switches in other DPEs in the array.
20 . The IC of claim 18 further comprising:
a system interface circuit coupled to the NoC and to the array of DPEs.