Prediction unit with first predictor that provides a hashed fetch address of a current fetch block to its own input and to a second predictor that uses it to predict the fetch address of a next fetch block
A prediction unit includes a first predictor that provides an output comprising a hashed fetch address of a current fetch block in response to an input. The first predictor input comprises a hashed fetch address of a previous fetch block that immediately precedes the current fetch block in program execution order. A second predictor provides an output comprising a fetch address of a next fetch block that immediately succeeds the current fetch block in program execution order in response to an input. The second predictor input comprises the hashed fetch address of the current fetch block output by the first predictor.
1 . A microprocessor, comprising:
a first prediction circuit configured to make first predictions of information about fetch blocks of a program instruction stream, wherein a fetch block comprises a previously executed sequential run of instructions;
a second prediction circuit configured to make second predictions of information about the fetch blocks of the program instruction stream;
wherein the second prediction circuit is configured to use the first predictions to make the second predictions;
wherein the second prediction circuit is configured to make the second predictions with a higher prediction accuracy than the first prediction circuit is configured to make the first predictions; and
wherein a second misprediction penalty associated with a misprediction by the second prediction circuit is greater than a first misprediction penalty associated with a misprediction by the first prediction circuit.
2 . The microprocessor of claim 1 ,
wherein to detect a misprediction of a first prediction made by the first prediction circuit, the microprocessor is configured to compare the first prediction with a second prediction made by the second prediction circuit; and
wherein to detect a misprediction of the second prediction, the microprocessor is configured to compare the second prediction with results of execution of a fetch block associated with the first and second predictions.
3 . The microprocessor of claim 2 ,
wherein the misprediction of the first prediction is detected prior to execution of the fetch block.
4 . The microprocessor of claim 1 ,
wherein the first prediction circuit is a single-cycle prediction unit;
wherein the second prediction circuit is a fully pipelined multi-cycle prediction unit; and
wherein the second prediction circuit is configured to use the first predictions made at a rate of one per clock cycle to make the second predictions at a rate of one per clock cycle.
5 . The microprocessor of claim 1 ,
wherein the first prediction circuit comprises a set-associative structure.
6 . The microprocessor of claim 1 ,
wherein the information of the first predictions fully enables initiation of lookups in predictor structures of the second prediction circuit.
7 . The microprocessor of claim 1 ,
wherein the information of the first predictions comprises path history used in lookups in predictor structures of the second prediction circuit.
8 . The microprocessor of claim 1 ,
wherein the information of the first predictions comprises hashed tags used in lookups in predictor structures of the second prediction circuit.
9 . The microprocessor of claim 1 ,
wherein the information of the first predictions is insufficient for use by an instruction fetch unit to fetch a fetch block from an instruction cache, whereas the information of the second predictions is sufficient for use by the instruction fetch unit to fetch a fetch block from the instruction cache.
10 . The microprocessor of claim 1 ,
wherein the first prediction circuit comprises an input and an output; and
wherein the input of a current clock cycle comprises at least a portion of the output of the previous clock cycle.
11 . A method, comprising:
making, by a first prediction circuit, first predictions of information about fetch blocks of a program instruction stream, wherein a fetch block comprises a previously executed sequential run of instructions; and
making, by a second prediction circuit, second predictions of information about the fetch blocks of the program instruction stream;
wherein the first predictions are used by the second prediction circuit to make the second predictions;
wherein the second predictions are made by the second prediction circuit with a higher prediction accuracy than the first prediction circuit makes the first predictions; and
wherein a second misprediction penalty associated with a misprediction by the second prediction circuit is greater than a first misprediction penalty associated with a misprediction by the first prediction circuit.
12 . The method of claim 11 , further comprising:
comparing the first prediction with a second prediction made by the second prediction circuit to detect a misprediction of a first prediction made by the first prediction circuit; and
comparing the second prediction with results of execution of a fetch block associated with the first and second predictions to detect a misprediction of the second prediction.
13 . The method of claim 12 ,
wherein the misprediction of the first prediction is detected prior to execution of the fetch block.
14 . The method of claim 11 , further comprising:
wherein the first prediction circuit is a single-cycle prediction unit;
wherein the second prediction circuit is a fully pipelined multi-cycle prediction unit; and
using, by the second prediction circuit, the first predictions made at a rate of one per clock cycle to make the second predictions at a rate of one per clock cycle.
15 . The method of claim 11 ,
wherein the first prediction circuit comprises a set-associative structure.
16 . The method of claim 11 ,
wherein the information of the first predictions fully enables initiation of lookups in predictor structures of the second prediction circuit.
17 . The method of claim 11 ,
wherein the information of the first predictions comprises path history used in lookups in predictor structures of the second prediction circuit.
18 . The method of claim 11 ,
wherein the information of the first predictions comprises hashed tags used in lookups in predictor structures of the second prediction circuit.
19 . The method of claim 11 ,
wherein the information of the first predictions is insufficient for use by an instruction fetch unit to fetch a fetch block from an instruction cache, whereas the information of the second predictions is sufficient for use by the instruction fetch unit to fetch a fetch block from the instruction cache.
20 . The method of claim 11 ,
wherein the first prediction circuit comprises an input and an output; and
wherein the input of a current clock cycle comprises at least a portion of the output of the previous clock cycle.
21 . A non-transitory computer-readable medium having instructions stored thereon that are capable of causing or configuring a microprocessor comprising:
a first prediction circuit configured to make first predictions of information about fetch blocks of a program instruction stream, wherein a fetch block comprises a previously executed sequential run of instructions;
a second prediction circuit configured to make second predictions of information about the fetch blocks of the program instruction stream;
wherein the second prediction circuit is configured to use the first predictions to make the second predictions;
wherein the second prediction circuit is configured to make the second predictions with a higher prediction accuracy than the first prediction circuit is configured to make the first predictions; and
wherein a second misprediction penalty associated with a misprediction by the second prediction circuit is greater than a first misprediction penalty associated with a misprediction by the first prediction circuit.