HBM RAS cache architecture

1. An apparatus comprising:

stacked integrated circuit dies comprising:

a memory cell die configured to store data based, at least in part, on a memory address; and

a logic die comprising:

an interface to the stacked integrated circuit dies and configured to communicate memory accesses between the memory cell die and an external device, and

a reliability circuit comprising:

a spare memory configured to store data, and

an address table configured to map a memory address associated with an error to a portion of the spare memory,

wherein the reliability circuit is configured to, based on a detected error command from a processor external to the logic die, edit a memory address entry of the address table;

wherein the reliability circuit is configured to, based on a memory access to the stacked integrated circuit dies, determine that the memory access is associated with an error, and complete the memory access with, at least in part, the spare memory.

2. The apparatus of claim 1 , wherein the address table comprises:

an error correction code portion to protect against errors in the address table, and

a valid flag configured to indicate that the memory address entry is active.

3. The apparatus of claim 1 , wherein the reliability circuit is configured to:

receive a write memory access, from an external device, for data at a memory address;

determine that the memory address is associated with an error; and

write the data to a portion of the spare memory.

4. The apparatus of claim 3 , wherein the reliability circuit is configured to determine that the memory address is associated with an error by:

determining that the memory address is stored within the address table as a memory address entry, and

determining that a valid flag indicates that the memory address entry is active.

5. The apparatus of claim 1 , wherein the reliability circuit is configured to:

receive a read memory access, from an external device, for data at a memory address;

determine that the memory address is associated with an error; and

retrieve the data from a portion of the spare memory.

6. The apparatus of claim 1 , wherein the reliability circuit is configured to:

receive a message from an external device indicating that a memory address is associated with an error;

place the memory address in the address table; and

associate the memory address with the portion of the spare memory.

7. The apparatus of claim 6 , wherein the reliability circuit is configured to receive a message from an external device indicating that a memory address is associated with an error by:

receiving a write memory access to a predefined memory address, wherein the predefined memory address is associated with the address table.

8. The apparatus of claim 1 , wherein the reliability circuit is configured to:

monitor a usage level of the spare memory, and

store an indicator of the usage level at a predefined memory address; and

wherein the interface is configured to:

receive a read memory access to the predefined memory address, and

based on the read memory access, provide the indicator of the usage level of the spare memory.

9. A system comprising:

a processor comprising:

a memory controller configured to manage a flow of data to and from the processor, and

an error correction circuit configured to detect an error associated with stored data; and

a high-bandwidth memory stack of integrated circuits, separate from the processor, comprising:

high-bandwidth memory cells configured to store data based, at least in part, on a memory address, and

a reliability circuit comprising:

a spare memory configured to store data, and

an address table configured to map a memory address associated with the error to a portion of the spare memory

wherein the reliability circuit is configured to, based on a memory access to the high-bandwidth memory stack of integrated circuits, determine that the memory access is associated with an error detected by the error correction circuit, and complete the memory access with, at least in part, the spare memory.

10. The system of claim 9 , wherein the processor is configured to:

determine that the spare memory includes sufficient free capacity to correct, at least partially, the error, and

cause the reliability circuit to map the memory address associated with the error to a portion of the spare memory.

11. The system of claim 9 , wherein the reliability circuit is configured to:

receive a write memory access, from the processor, for data at a memory address;

determine that the memory address is associated with an error; and

write the data to a mapped portion of the spare memory.

12. The system of claim 9 , wherein the reliability circuit is configured to:

receive a read memory access, from the processor, for data at a memory address;

determine that the memory address is associated with an error; and

retrieve the data from a mapped portion of the spare memory.

13. The system of claim 9 , wherein the processor is configured to issue a write memory access to a predefined memory address, wherein the predefined memory address is associated with the address table and wherein the write memory access includes the memory address associated with the error as a piece of data

wherein the reliability circuit is configured to:

receive a message, from the processor, indicating that a memory address is associated with an error;

place the memory address in the address table; and

associate the memory address with the portion of the spare memory.

14. The system of claim 9 , wherein the reliability circuit is configured to:

monitor a usage level of the spare memory, and

store an indicator of the usage level at a predefined memory address, and

wherein the processor is configured to:

issue a read memory access to the predefined memory address, and

receive, based on the read memory access, an indicator of the usage level of the spare memory.

15. The system of claim 9 , wherein the processor includes a non-volatile memory configured to store a list of memory addresses associated with respective errors, and

wherein the error correction circuit is configured to, based on detecting an error to stored data cause the memory address associated with the error to be stored by both the non-volatile memory and the address table.

16. The system of claim 15 , wherein non-volatile memory is configured to add one or more of the list of memory addresses associated with respective errors to the address table; and

wherein the reliability circuit is configured to map one or more of the memory addresses to respective positions of the spare memory.

17. The system of claim 9 , wherein the reliability circuit is configured to remap data errors within the high-bandwidth memory cells without resetting the processor.

18. An apparatus comprising:

a logic die comprising:

an internal interface to a high-bandwidth memory die, wherein the high-bandwidth memory die configured to store data based upon a memory address;

an external interface configured to communicate memory accesses between the high-bandwidth memory die and at least one external device; and

a reliability circuit comprising:

a spare memory configured to store data, and

an address table configured to map a memory address associated with an error to a portion of the spare memory, and

wherein the reliability circuit is configured to, based on a detected error command from a processor external to the logic die, edit a memory address entry of the address table;

wherein the reliability circuit is configured to, based on a memory access to the high-bandwidth memory die, determine that the memory access is associated with an error, and complete the memory access with, at least in part, the spare memory.

19. The apparatus of claim 18 , wherein the reliability circuit is configured to:

receive a write memory access, from the external device, for data at a memory address;

determine that the memory address is associated with an error; and

write the data to a portion of the spare memory.

20. The apparatus of claim 18 , wherein the reliability circuit is configured to:

receive a read memory access, from the external device, for data at a memory address;

determine that the memory address is associated with an error; and

retrieve the data from a mapped portion of the spare memory.