Integrated semi-inclusive hierarchical metadata predictor
Embodiments are provided for an integrated semi-inclusive hierarchical metadata predictor. A hit in a second-level structure is determined, the hit being associated with a line of metadata in the second-level structure. Responsive to determining that a victim line of metadata in a first-level structure meets at least one condition, the victim line of metadata is stored in the second-level structure. The line of metadata from the second-level structure is stored in a first-level structure to be utilized to facilitate performance of a processor, the line of metadata from the second-level structure including entries for a plurality of instructions.
1. A computer-implemented method comprising:
determining, by a metadata predictor, a hit in a second-level structure, the hit being associated with a line of metadata in the second-level structure;
responsive to determining that a victim line of metadata in a first-level structure meets at least one condition, storing, by the metadata predictor, the victim line of metadata in the second-level structure; and
storing, by the metadata predictor, the line of metadata from the second-level structure in the first-level structure to be utilized to facilitate performance of a processor, the line of metadata from the second-level structure comprising entries for a plurality of instructions, wherein the metadata predictor runs asynchronously to a prediction pipeline of the processor.
2. The computer-implemented method of claim 1 , wherein the at least one condition comprises an indication that the victim line of metadata has been updated.
3. The computer-implemented method of claim 1 , wherein the at least one condition comprises an indication that the victim line of metadata is absent from the second-level structure.
4. The computer-implemented method of claim 1 , wherein the determining the hit in the second-level structure is performed responsive to determining that the line of metadata is a miss in the first-level structure.
5. The computer-implemented method of claim 1 , wherein the line of metadata from the second-level structure is provided to a prediction pipeline to make metadata predictions.
6. The computer-implemented method of claim 1 , wherein:
the second-level structure is configured to be searched for a preload line of metadata associated with a preload victim line of metadata, the first-level structure being searched to provide the preload victim line of metadata;
the preload line of metadata is written into the first-level structure; and
the preload line of metadata is provided to a prediction pipeline, responsive to being requested.
7. The computer-implemented method of claim 1 , wherein a prediction pipeline is stalled until the second-level structure makes a determination about whether there is the hit, the prediction pipeline being configured to resume responsive to the determination and configured to use the line of metadata associated with the hit.
8. The computer-implemented method of claim 1 , wherein the first-level structure a first branch target buffer and the second-level structure is a second branch target buffer coupled to the first branch target buffer.
9. The computer-implemented method of claim 1 , wherein a line index buffer is coupled to the first branch target buffer, the line index buffer being populated from restarts.
10. The computer-implemented method of claim 1 , wherein second branch target buffer indexing logic is coupled to the second branch target buffer, the second branch target buffer indexing logic being populated from restarts.
11. A system comprising:
a metadata predictor configured to perform operations comprising:
determining a hit in a second-level structure, the hit being associated with a line of metadata in the second-level structure;
responsive to determining that a victim line of metadata in a first-level structure meets at least one condition, storing the victim line of metadata in the second-level structure; and
storing the line of metadata from the second-level structure in the first-level structure to be utilized to facilitate performance of one or more processors, the line of metadata from the second-level structure comprising entries for a plurality of instructions, wherein the metadata predictor runs asynchronously to a prediction pipeline of the one or more processors.
12. The system of claim 11 , wherein the at least one condition comprises an indication that the victim line of metadata has been updated.
13. The system of claim 11 , wherein the at least one condition comprises an indication that the victim line of metadata is absent from the second-level structure.
14. The system of claim 11 , wherein the determining the hit in the second-level structure is performed responsive to determining that the line of metadata is a miss in the first-level structure.
15. The system of claim 11 , wherein the line of metadata from the second-level structure is provided to a prediction pipeline to make metadata predictions.
16. The system of claim 11 , wherein:
the second-level structure is configured to be searched for a preload line of metadata associated with a preload victim line of metadata, the first-level structure being searched to provide the preload victim line of metadata;
the preload line of metadata is written into the first-level structure; and
the preload line of metadata is provided to a prediction pipeline, responsive to being requested.
17. The system of claim 11 , wherein a prediction pipeline is stalled until the second-level structure makes a determination about whether there is the hit, the prediction pipeline being configured to resume responsive to the determination and configured to use the line of metadata associated with the hit.