Context load mechanism in a coarse-grained reconfigurable array processor
Devices and techniques for loading contexts in a coarse-grained reconfigurable array processor are described herein. A system or apparatus may include context load circuitry operable to load context for a coarse-grained reconfigurable array processor, where the context load circuitry is configured to: (a) receive a kernel identifier; (b) access a first registry to obtain a context mask base address; (c) determine a context mask address from the context mask base address and the kernel identifier; (d) access a second registry to obtain a context state base address; (e) determine a context state address from the context state base address and the kernel identifier; (f) use a context mask at the context mask address to determine corresponding active context state; and (g) load the corresponding active context state into the coarse-grained reconfigurable array processor.
1 . A context load circuitry configured to communicate with a memory, the context load circuitry operable to load context state data for a coarse-grained reconfigurable array processor, the context load circuitry configured to:
determine a context mask address from a context mask base address and a kernel identifier, the context mask base address stored in a first registry, the context mask base address for a context mask region in the memory;
determine a context state address from a context state base address and the kernel identifier, the context state base address for a context state region in the memory, and the context state base address stored in a second registry;
use a context mask at the context mask address to determine a corresponding active context state; and
load the corresponding active context state into the coarse-grained reconfigurable array processor.
2 . The context load circuitry of claim 1 , wherein the kernel identifier is an address offset, wherein to determine the context mask address, the context load circuitry is configured to add the address offset to the context mask base address, and wherein to determine the context state address, the context load circuitry is configured to add the address offset to the context state base address.
3 . The context load circuitry of claim 1 , wherein the kernel identifier is an alphanumeric identification associated with an address offset, wherein to determine the context mask address, the context load circuitry is configured to add the address offset to the context mask base address, and wherein to determine the context state address, the context load circuitry is configured to add the address offset to the context state base address.
4 . The context load circuitry of claim 3 , wherein the context load circuitry is configured to perform a lookup to determine the address offset associated with the alphanumeric identification.
5 . The context load circuitry of claim 1 , wherein the context mask is an 8-bit mask.
6 . The context load circuitry of claim 1 , wherein each bit of the context mask corresponds to a context state data field.
7 . The context load circuitry of claim 6 , wherein the context state data field is an 8-byte field.
8 . The context load circuitry of claim 1 , wherein the context load circuitry is configured to store the kernel identifier.
9 . The context load circuitry of claim 1 , wherein the context load circuitry is integrated in a hybrid threading fabric.
10 . A method for loading context state data for a coarse-grained reconfigurable array processor, comprising:
determining a context mask address from a context mask base address and a kernel identifier, the context mask base address stored in a first registry, the context mask base address for a context mask region in a memory;
determining a context state address from a context state base address and the kernel identifier, the context state base address for a context state region in the memory, and the context state base address stored in a second registry;
using a context mask at the context mask address to determine a corresponding active context state; and
loading, by context load circuitry, the corresponding active context state into the coarse-grained reconfigurable array processor.
11 . The method of claim 10 , wherein the kernel identifier is an address offset, wherein determining the context mask address comprises adding the address offset to the context mask base address, and wherein determining the context state address comprises adding the address offset to the context state base address.
12 . The method of claim 10 , wherein the kernel identifier is an alphanumeric identification associated with an address offset, wherein determining the context mask address comprises adding the address offset to the context mask base address, and wherein determining the context state address comprises adding the address offset to the context state base address.
13 . The method of claim 12 , comprising performing a lookup to determine the address offset associated with the alphanumeric identification.
14 . The method of claim 10 , wherein the context mask is an 8-bit mask.
15 . The method of claim 10 , wherein each bit of the context mask corresponds to an 8-byte context state data field.
16 . The method of claim 10 , wherein loading the corresponding active context state into the coarse-grained reconfigurable array processor comprises storing the corresponding active context state in a register of a processing element in the coarse-grained reconfigurable array processor.
17 . The method of claim 10 , wherein the context load circuitry is integrated in a hybrid threading fabric.
18 . The method of claim 17 , wherein the hybrid threading fabric is included in a memory device.
19 . The method of claim 18 , wherein the context load circuitry is integrated in the coarse-grained reconfigurable array processor.