Combined system responses in a chip multiprocessor

1. A node comprising, integrated onto a single integrated circuit chip:

a plurality of processor cores; and

a node controller coupled to the plurality of processor cores, wherein the node controller is coupled to receive an external request transmitted to the node, and wherein the node controller is configured to transmit a corresponding request to at least a subset of the plurality of processor cores responsive to the external request, and wherein the node controller is configured to receive respective responses from each processor core of the subset, each processor core transmitting the respective response independently in response to servicing the corresponding request and capable of transmitting the response on different clock cycles than other processor cores, and wherein the node controller is configured to transmit an external response to the external request responsive to receiving each of the respective responses, and wherein the external request is a system management message requesting that a power management state be entered by each of the plurality of processor cores.

2. The node as recited in claim 1 wherein the at least a subset of the plurality of processor cores comprises each of the plurality of processor cores.

3. The node as recited in claim 1 wherein the external response is a composite of the respective responses.

4. The node as recited in claim 1 wherein the node controller is configured to receive an external probe identifying a cache block.

5. The node as recited in claim 4 wherein each of the plurality of processor cores comprises at least one cache, and wherein the node controller is configured to transmit a request for the processor core to probe the cache for the cache block in response to the external probe.

6. The node as recited in claim 5 wherein a probe response from each processor core indicates a state of the cache block in the processor core, and wherein each processor core is configured to transmit the probe response to the node controller.

7. The node as recited in claim 6 wherein the node controller is configured to record the state provided in each probe response.

8. The node as recited in claim 7 wherein the node controller is configured to generate an external probe response for the external probe, wherein the external probe response comprises the highest reported state.

9. The node as recited in claim 1 wherein the responses from the processor cores indicate that the power management state has been entered.

10. The node as recited in claim 9 wherein the external response indicates that the power management state has been entered by each of the plurality of processor cores.

11. The node as recited in claim 1 wherein the node controller comprises a system request queue configured to store the external request, and wherein the node controller is configured to record receipt of respective responses in the system request queue.

12. A method comprising:

receiving an external request in a node comprising a plurality of processor cores;

transmitting a corresponding request to at least a subset of the plurality of processor cores responsive to the external request;

each processor core transmitting a respective response independently in response to servicing the corresponding request and capable of transmitting the response on different clock cycles than other processor cores;

receiving the respective responses from each processor core of the subset;

transmitting an external response to the external request responsive to receiving each of the respective responses, wherein the external request is a system management message requesting that a power management state be entered by each of the plurality of processor cores;

receiving an external probe in the node;

transmitting a corresponding probe request to at least a subset of the plurality of processor cores responsive to the external probe;

each processor core transmitting a respective probe response independently in response to servicing the corresponding probe request and capable of transmitting the response on different clock cycles than other processor cores;

receiving the respective probe responses from each processor core of the subset; and

transmitting an external probe response to the external probe responsive to receiving each of the respective probe responses.

13. The method as recited in claim 12 wherein each of the plurality of processor cores comprises at least one cache, and wherein the corresponding probe request is a request for the processor core to probe the cache for the cache block.

14. The method as recited in claim 13 wherein the respective probe response from each processor core indicates a state of the cache block in the processor core, the method further comprising recording the state provided in each respective probe response.

15. The method as recited in claim 12 wherein the responses from the processor cores indicate that the power management state has been entered, and wherein the external response indicates that the power management state has been entered by each of the plurality of processor cores.