[RISCV] Add tune info for mem* expansion #118439
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
So that CPUs can tune these options.
Member
|
@llvm/pr-subscribers-backend-risc-v Author: Pengcheng Wang (wangpc-pp) ChangesSo that CPUs can tune these options. Full diff: https://github.com/llvm/llvm-project/pull/118439.diff 3 Files Affected:
diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index b2e96b63a80953..8f6e0521b3c1b4 100644
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -1544,6 +1544,20 @@ RISCVTargetLowering::RISCVTargetLowering(const TargetMachine &TM,
// corresponding branch. This information is used in CGP/SelectOpt to decide
// when to convert selects into branches.
PredictableSelectIsExpensive = Subtarget.predictableSelectIsExpensive();
+
+ MaxStoresPerMemsetOptSize = Subtarget.getMaxStoresPerMemset(/*OptSize=*/true);
+ MaxStoresPerMemset = Subtarget.getMaxStoresPerMemset(/*OptSize=*/false);
+
+ MaxGluedStoresPerMemcpy = Subtarget.getMaxGluedStoresPerMemcpy();
+ MaxStoresPerMemcpyOptSize = Subtarget.getMaxStoresPerMemcpy(/*OptSize=*/true);
+ MaxStoresPerMemcpy = Subtarget.getMaxStoresPerMemcpy(/*OptSize=*/false);
+
+ MaxStoresPerMemmoveOptSize =
+ Subtarget.getMaxStoresPerMemmove(/*OptSize=*/true);
+ MaxStoresPerMemmove = Subtarget.getMaxStoresPerMemmove(/*OptSize=*/false);
+
+ MaxLoadsPerMemcmpOptSize = Subtarget.getMaxLoadsPerMemcmp(/*OptSize=*/true);
+ MaxLoadsPerMemcmp = Subtarget.getMaxLoadsPerMemcmp(/*OptSize=*/false);
}
EVT RISCVTargetLowering::getSetCCResultType(const DataLayout &DL,
diff --git a/llvm/lib/Target/RISCV/RISCVProcessors.td b/llvm/lib/Target/RISCV/RISCVProcessors.td
index 471f051728e99f..c4e19c515b155b 100644
--- a/llvm/lib/Target/RISCV/RISCVProcessors.td
+++ b/llvm/lib/Target/RISCV/RISCVProcessors.td
@@ -24,15 +24,32 @@ class RISCVTuneInfo {
// Tail duplication threshold at -O3.
bits<32> TailDupAggressiveThreshold = 6;
+
+ bits<32> MaxStoresPerMemsetOptSize = 4;
+ bits<32> MaxStoresPerMemset = 8;
+
+ bits<32> MaxGluedStoresPerMemcpy = 0;
+ bits<32> MaxStoresPerMemcpyOptSize = 4;
+ bits<32> MaxStoresPerMemcpy = 8;
+
+ bits<32> MaxStoresPerMemmoveOptSize = 4;
+ bits<32> MaxStoresPerMemmove = 8;
+
+ bits<32> MaxLoadsPerMemcmpOptSize = 4;
+ bits<32> MaxLoadsPerMemcmp = 8;
}
def RISCVTuneInfoTable : GenericTable {
let FilterClass = "RISCVTuneInfo";
let CppTypeName = "RISCVTuneInfo";
let Fields = ["Name", "PrefFunctionAlignment", "PrefLoopAlignment",
- "CacheLineSize", "PrefetchDistance",
- "MinPrefetchStride", "MaxPrefetchIterationsAhead",
- "MinimumJumpTableEntries", "TailDupAggressiveThreshold"];
+ "CacheLineSize", "PrefetchDistance", "MinPrefetchStride",
+ "MaxPrefetchIterationsAhead", "MinimumJumpTableEntries",
+ "TailDupAggressiveThreshold", "MaxStoresPerMemsetOptSize",
+ "MaxStoresPerMemset", "MaxGluedStoresPerMemcpy",
+ "MaxStoresPerMemcpyOptSize", "MaxStoresPerMemcpy",
+ "MaxStoresPerMemmoveOptSize", "MaxStoresPerMemmove",
+ "MaxLoadsPerMemcmpOptSize", "MaxLoadsPerMemcmp"];
}
def getRISCVTuneInfo : SearchIndex {
diff --git a/llvm/lib/Target/RISCV/RISCVSubtarget.h b/llvm/lib/Target/RISCV/RISCVSubtarget.h
index 043838e13b964d..feed84c68a7869 100644
--- a/llvm/lib/Target/RISCV/RISCVSubtarget.h
+++ b/llvm/lib/Target/RISCV/RISCVSubtarget.h
@@ -53,6 +53,19 @@ struct RISCVTuneInfo {
// Tail duplication threshold at -O3.
unsigned TailDupAggressiveThreshold;
+
+ unsigned MaxStoresPerMemsetOptSize;
+ unsigned MaxStoresPerMemset;
+
+ unsigned MaxGluedStoresPerMemcpy;
+ unsigned MaxStoresPerMemcpyOptSize;
+ unsigned MaxStoresPerMemcpy;
+
+ unsigned MaxStoresPerMemmoveOptSize;
+ unsigned MaxStoresPerMemmove;
+
+ unsigned MaxLoadsPerMemcmpOptSize;
+ unsigned MaxLoadsPerMemcmp;
};
#define GET_RISCVTuneInfoTable_DECL
@@ -325,6 +338,30 @@ class RISCVSubtarget : public RISCVGenSubtargetInfo {
return TuneInfo->TailDupAggressiveThreshold;
}
+ unsigned getMaxStoresPerMemset(bool OptSize) const {
+ return OptSize ? TuneInfo->MaxStoresPerMemsetOptSize
+ : TuneInfo->MaxStoresPerMemset;
+ }
+
+ unsigned getMaxGluedStoresPerMemcpy() const {
+ return TuneInfo->MaxGluedStoresPerMemcpy;
+ }
+
+ unsigned getMaxStoresPerMemcpy(bool OptSize) const {
+ return OptSize ? TuneInfo->MaxStoresPerMemcpyOptSize
+ : TuneInfo->MaxStoresPerMemcpy;
+ }
+
+ unsigned getMaxStoresPerMemmove(bool OptSize) const {
+ return OptSize ? TuneInfo->MaxStoresPerMemmoveOptSize
+ : TuneInfo->MaxStoresPerMemmove;
+ }
+
+ unsigned getMaxLoadsPerMemcmp(bool OptSize) const {
+ return OptSize ? TuneInfo->MaxLoadsPerMemcmpOptSize
+ : TuneInfo->MaxLoadsPerMemcmp;
+ }
+
void overrideSchedPolicy(MachineSchedPolicy &Policy,
unsigned NumRegionInstrs) const override;
};
|
Contributor
|
What's your plan for the getters in TargetLowering for the same properties? These are queried elsewhere in the codebase. |
Contributor
Author
I don't get what you mean, these getters should return the same values that we set in XXXTargetLowering? |
Contributor
Sorry, I'd somehow missed the part that assigned the underlying values used by those getters in RISCVISelLowering. All good. |
Collaborator
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/175/builds/9705 Here is the relevant piece of the build log for the reference |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
So that CPUs can tune these options.