System and method for efficiently mapping heterogeneous objects onto an array of heterogeneous programmable logic resources
A system and method of mapping heterogeneous objects onto an array of heterogeneous programmable logic resources. The method comprises clustering to identify datapath modules from a netlist. The method further comprises aggregating the datapath modules into higher level modules. The method also comprises clustering random logic into structures.
1. A method of mapping heterogeneous objects onto an array of heterogeneous programmable logic resources, the method comprising:
clustering the heterogeneous objects from a mapped netlist to identify datapath modules;
aggregating the datapath modules into higher level modules of the heterogeneous objects and clusters of the heterogeneous objects;
clustering random logic into structures for the array of heterogeneous programmable logic resources; and
determining placement of the higher level modules and the structures in the array of heterogeneous programmable logic resources.
2. The method of claim 1 , wherein the array is a multi-scale programmable array.
3. The method of claim 2 , wherein clustering creates transitive processing unit (TPU) clusters.
4. The method of claim 2 , wherein clustering creates function cell/arithmetic logic unit (FC/ALU) clusters.
5. The method of claim 2 , wherein the array comprises a transitive processing unit (TPU) plane and a function cell/arithmetic logic unit (FC/ALU) plane.
6. The method of claim 1 , wherein the method comprises a pre-processing phase, a clustering phase, an initial placement phase and an optimization phase.
7. The method of claim 1 , wherein constraints affect the placement of objects.
8. A system for mapping heterogeneous objects onto an array of heterogeneous programmable logic resources, the system comprising:
a first clustering module for clustering the heterogeneous objects from a mapped netlist to identify datapath modules;
an aggregating module for aggregating the datapath modules into higher level modules of the heterogeneous objects and clusters of the heterogeneous objects;
a second clustering module for clustering random logic into structures for the array of heterogeneous programmable logic resources; and
a placement module for determining placement of the higher level modules and the structures in the array of heterogeneous programmable logic resources.
9. The system of claim 8 , wherein the array is a multi-scale programmable array.
10. The system of claim 9 , wherein clustering creates transitive processing unit (TPU) clusters.
11. The system of claim 9 , wherein clustering creates function cell/arithmetic logic unit (FC/ALU) clusters.
12. The system of claim 9 , wherein the array comprises a transitive processing unit (TPU) plane and a function cell/arithmetic logic unit (FC/ALU) plane.
13. The system of claim 8 , wherein the system comprises a pre-processing phase, a clustering phase, an initial placement phase and an optimization phase.
14. The system of claim 8 , wherein constraints affect the placement of objects.
15. An electronically-readable medium having embodied thereon a program, the program being executable by a machine to perform a method of mapping heterogeneous objects onto an array of heterogeneous programmable logic resources, the method comprising:
clustering the heterogeneous objects from a mapped netlist to identify datapath modules;
aggregating the datapath modules into higher level modules of the heterogeneous objects and clusters of the heterogeneous objects;
clustering random logic into structures for the array of heterogeneous programmable logic resources; and
determining placement of the higher level modules and the structures in the array of heterogeneous programmable logic resources.
16. The electronically-readable medium of claim 15 , wherein the array is a multi-scale programmable array.
17. The electronically-readable medium of claim 16 , wherein clustering creates transitive processing unit (TPU) clusters.
18. The electronically-readable medium of claim 16 , wherein clustering creates function cell/arithmetic logic unit (FC/ALU) clusters.
19. The electronically-readable medium of claim 16 , wherein the array comprises a transitive processing unit (TPU) plane and a function cell/arithmetic logic unit (FC/ALU) plane.
20. The electronically-readable medium of claim 15 , wherein the method comprises a pre-processing phase, a clustering phase, an initial placement phase and an optimization phase.
21. The electronically-readable medium of claim 15 , wherein constraints affect the placement of objects.
22. A system for mapping heterogeneous objects onto an array of heterogeneous programmable logic resources, the system comprising:
first clustering means for clustering the heterogeneous objects from a mapped netlist to identify datapath modules;
aggregating means for aggregating the datapath modules into higher level modules of the heterogeneous objects and clusters of the heterogeneous objects; and
second clustering means for clustering random logic into structures for the array of heterogeneous programmable logic resources; and
placement means for determining placement of the higher level modules and the structures in the array of heterogeneous programmable logic resources.
23. The system of claim 22 , wherein the array is a multi-scale programmable array.
24. The system of claim 23 , wherein clustering creates transitive processing unit (TPU) clusters.
25. The system of claim 23 , wherein clustering creates function cell/arithmetic logic unit (FC/ALU) clusters.
26. The system of claim 23 , wherein the array comprises a transitive processing unit (TPU) plane and a function cell/arithmetic logic unit (FC/ALU) plane.
27. The system of claim 22 , wherein the system comprises a pre-processing phase, a clustering phase, an initial placement phase and an optimization phase.
28. The system of claim 22 , wherein constraints affect the placement of objects.