Skip to content

[MCA][X86] Pretend To Have a Stack Engine #153348

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 4 commits into
base: main
Choose a base branch
from
+208 −2
Open

[MCA][X86] Pretend To Have a Stack Engine #153348

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
aa7076c
[MCA][X86] Pretend To Have a Stack Engine
boomanaiden154 Aug 13, 2025
2acc2b5
add todo
boomanaiden154 Aug 14, 2025
a5a592d
Feedback
boomanaiden154 Aug 14, 2025
98650ae
Merge branch 'main' into llvm-mca-x86-stack-engine
boomanaiden154 Aug 16, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
18 changes: 18 additions & 0 deletions llvm/lib/Target/X86/MCA/X86CustomBehaviour.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,11 +36,29 @@ void X86InstrPostProcess::setMemBarriers(std::unique_ptr<Instruction> &Inst,
}
}

void X86InstrPostProcess::useStackEngine(std::unique_ptr<Instruction> &Inst,
const MCInst &MCI) {
// TODO(boomanaiden154): We currently do not handle PUSHF/POPF because we
// have not done the necessary benchmarking to see if they are also
// optimized by the stack engine.
if (X86::isPOP(MCI.getOpcode()) || X86::isPUSH(MCI.getOpcode())) {
auto *StackRegisterDef =
llvm::find_if(Inst->getDefs(), [](const WriteState &State) {
return State.getRegisterID() == X86::RSP;
});
assert(
StackRegisterDef != Inst->getDefs().end() &&
"Expected push instruction to implicitly use stack pointer register.");
Inst->getDefs().erase(StackRegisterDef);
}
}

void X86InstrPostProcess::postProcessInstruction(
std::unique_ptr<Instruction> &Inst, const MCInst &MCI) {
// Currently, we only modify certain instructions' IsALoadBarrier and
// IsAStoreBarrier flags.
setMemBarriers(Inst, MCI);
useStackEngine(Inst, MCI);
}

} // namespace mca
Expand Down
5 changes: 5 additions & 0 deletions llvm/lib/Target/X86/MCA/X86CustomBehaviour.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,6 +28,11 @@ class X86InstrPostProcess : public InstrPostProcess {
/// as load and store barriers.
void setMemBarriers(std::unique_ptr<Instruction> &Inst, const MCInst &MCI);

/// Called within X86InstrPostPorcess to remove some rsp read operands
/// on stack instructions to better simulate the stack engine. We currently
/// do not model features of the stack engine like sync uops.
void useStackEngine(std::unique_ptr<Instruction> &Inst, const MCInst &MCI);

public:
X86InstrPostProcess(const MCSubtargetInfo &STI, const MCInstrInfo &MCII)
: InstrPostProcess(STI, MCII) {}
Expand Down
92 changes: 92 additions & 0 deletions llvm/test/tools/llvm-mca/X86/stack-engine-pop.s
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,92 @@
# NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
# RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=skylake -timeline -iterations=2 < %s | FileCheck %s

movq $0x80, %rsp
popq %rax
popq %rcx
popq %rdx
popq %rbx
popq %r12

# CHECK: Iterations: 2
# CHECK-NEXT: Instructions: 12
# CHECK-NEXT: Total Cycles: 14
# CHECK-NEXT: Total uOps: 22

# CHECK: Dispatch Width: 6
# CHECK-NEXT: uOps Per Cycle: 1.57
# CHECK-NEXT: IPC: 0.86
# CHECK-NEXT: Block RThroughput: 2.5

# CHECK: Instruction Info:
# CHECK-NEXT: [1]: #uOps
# CHECK-NEXT: [2]: Latency
# CHECK-NEXT: [3]: RThroughput
# CHECK-NEXT: [4]: MayLoad
# CHECK-NEXT: [5]: MayStore
# CHECK-NEXT: [6]: HasSideEffects (U)

# CHECK: [1] [2] [3] [4] [5] [6] Instructions:
# CHECK-NEXT: 1 1 0.25 movq $128, %rsp
# CHECK-NEXT: 2 6 0.50 * popq %rax
# CHECK-NEXT: 2 6 0.50 * popq %rcx
# CHECK-NEXT: 2 6 0.50 * popq %rdx
# CHECK-NEXT: 2 6 0.50 * popq %rbx
# CHECK-NEXT: 2 6 0.50 * popq %r12

# CHECK: Resources:
# CHECK-NEXT: [0] - SKLDivider
# CHECK-NEXT: [1] - SKLFPDivider
# CHECK-NEXT: [2] - SKLPort0
# CHECK-NEXT: [3] - SKLPort1
# CHECK-NEXT: [4] - SKLPort2
# CHECK-NEXT: [5] - SKLPort3
# CHECK-NEXT: [6] - SKLPort4
# CHECK-NEXT: [7] - SKLPort5
# CHECK-NEXT: [8] - SKLPort6
# CHECK-NEXT: [9] - SKLPort7

# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9]
# CHECK-NEXT: - - 1.50 1.50 2.50 2.50 - 1.50 1.50 -

# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] Instructions:
# CHECK-NEXT: - - - - - - - 0.50 0.50 - movq $128, %rsp
# CHECK-NEXT: - - 0.50 - 0.50 0.50 - 0.50 - - popq %rax
# CHECK-NEXT: - - - 0.50 0.50 0.50 - - 0.50 - popq %rcx
# CHECK-NEXT: - - 0.50 - 0.50 0.50 - 0.50 - - popq %rdx
# CHECK-NEXT: - - - 0.50 0.50 0.50 - - 0.50 - popq %rbx
# CHECK-NEXT: - - 0.50 0.50 0.50 0.50 - - - - popq %r12

# CHECK: Timeline view:
# CHECK-NEXT: 0123
# CHECK-NEXT: Index 0123456789

# CHECK: [0,0] DeER . . . movq $128, %rsp
# CHECK-NEXT: [0,1] D=eeeeeeER. . popq %rax
# CHECK-NEXT: [0,2] D=eeeeeeER. . popq %rcx
# CHECK-NEXT: [0,3] .D=eeeeeeER . popq %rdx
# CHECK-NEXT: [0,4] .D=eeeeeeER . popq %rbx
# CHECK-NEXT: [0,5] .D==eeeeeeER . popq %r12
# CHECK-NEXT: [1,0] . DeE------R . movq $128, %rsp
# CHECK-NEXT: [1,1] . D=eeeeeeER . popq %rax
# CHECK-NEXT: [1,2] . D==eeeeeeER. popq %rcx
# CHECK-NEXT: [1,3] . D=eeeeeeER. popq %rdx
# CHECK-NEXT: [1,4] . D==eeeeeeER popq %rbx
# CHECK-NEXT: [1,5] . D==eeeeeeER popq %r12

# CHECK: Average Wait times (based on the timeline view):
# CHECK-NEXT: [0]: Executions
# CHECK-NEXT: [1]: Average time spent waiting in a scheduler's queue
# CHECK-NEXT: [2]: Average time spent waiting in a scheduler's queue while ready
# CHECK-NEXT: [3]: Average time elapsed from WB until retire stage

# CHECK: [0] [1] [2] [3]
# CHECK-NEXT: 0. 2 1.0 1.0 3.0 movq $128, %rsp
# CHECK-NEXT: 1. 2 2.0 0.0 0.0 popq %rax
# CHECK-NEXT: 2. 2 2.5 0.5 0.0 popq %rcx
# CHECK-NEXT: 3. 2 2.0 1.0 0.0 popq %rdx
# CHECK-NEXT: 4. 2 2.5 1.5 0.0 popq %rbx
# CHECK-NEXT: 5. 2 3.0 2.0 0.0 popq %r12
# CHECK-NEXT: 2 2.2 1.0 0.5 <total>
92 changes: 92 additions & 0 deletions llvm/test/tools/llvm-mca/X86/stack-engine-push.s
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,92 @@
# NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
# RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=skylake -timeline -iterations=2 < %s | FileCheck %s

movq $0x80, %rsp
pushq %rax
pushq %rcx
pushq %rdx
pushq %rbx
pushq %r12

# CHECK: Iterations: 2
# CHECK-NEXT: Instructions: 12
# CHECK-NEXT: Total Cycles: 15
# CHECK-NEXT: Total uOps: 32

# CHECK: Dispatch Width: 6
# CHECK-NEXT: uOps Per Cycle: 2.13
# CHECK-NEXT: IPC: 0.80
# CHECK-NEXT: Block RThroughput: 5.0

# CHECK: Instruction Info:
# CHECK-NEXT: [1]: #uOps
# CHECK-NEXT: [2]: Latency
# CHECK-NEXT: [3]: RThroughput
# CHECK-NEXT: [4]: MayLoad
# CHECK-NEXT: [5]: MayStore
# CHECK-NEXT: [6]: HasSideEffects (U)

# CHECK: [1] [2] [3] [4] [5] [6] Instructions:
# CHECK-NEXT: 1 1 0.25 movq $128, %rsp
# CHECK-NEXT: 3 2 1.00 * pushq %rax
# CHECK-NEXT: 3 2 1.00 * pushq %rcx
# CHECK-NEXT: 3 2 1.00 * pushq %rdx
# CHECK-NEXT: 3 2 1.00 * pushq %rbx
# CHECK-NEXT: 3 2 1.00 * pushq %r12

# CHECK: Resources:
# CHECK-NEXT: [0] - SKLDivider
# CHECK-NEXT: [1] - SKLFPDivider
# CHECK-NEXT: [2] - SKLPort0
# CHECK-NEXT: [3] - SKLPort1
# CHECK-NEXT: [4] - SKLPort2
# CHECK-NEXT: [5] - SKLPort3
# CHECK-NEXT: [6] - SKLPort4
# CHECK-NEXT: [7] - SKLPort5
# CHECK-NEXT: [8] - SKLPort6
# CHECK-NEXT: [9] - SKLPort7

# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9]
# CHECK-NEXT: - - 1.50 1.50 1.50 1.50 5.00 1.50 1.50 2.00

# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] Instructions:
# CHECK-NEXT: - - - - - - - - 1.00 - movq $128, %rsp
# CHECK-NEXT: - - 0.50 - 0.50 - 1.00 0.50 - 0.50 pushq %rax
# CHECK-NEXT: - - - 0.50 - 0.50 1.00 - 0.50 0.50 pushq %rcx
# CHECK-NEXT: - - 0.50 - 0.50 0.50 1.00 0.50 - - pushq %rdx
# CHECK-NEXT: - - - 0.50 0.50 - 1.00 0.50 - 0.50 pushq %rbx
# CHECK-NEXT: - - 0.50 0.50 - 0.50 1.00 - - 0.50 pushq %r12

# CHECK: Timeline view:
# CHECK-NEXT: 01234
# CHECK-NEXT: Index 0123456789

# CHECK: [0,0] DeER . . . movq $128, %rsp
# CHECK-NEXT: [0,1] D=eeER . . pushq %rax
# CHECK-NEXT: [0,2] .D=eeER . . pushq %rcx
# CHECK-NEXT: [0,3] .D==eeER . . pushq %rdx
# CHECK-NEXT: [0,4] . D==eeER . . pushq %rbx
# CHECK-NEXT: [0,5] . D===eeER. . pushq %r12
# CHECK-NEXT: [1,0] . DeE---R. . movq $128, %rsp
# CHECK-NEXT: [1,1] . D===eeER . pushq %rax
# CHECK-NEXT: [1,2] . D===eeER . pushq %rcx
# CHECK-NEXT: [1,3] . D====eeER . pushq %rdx
# CHECK-NEXT: [1,4] . D====eeER. pushq %rbx
# CHECK-NEXT: [1,5] . D=====eeER pushq %r12

# CHECK: Average Wait times (based on the timeline view):
# CHECK-NEXT: [0]: Executions
# CHECK-NEXT: [1]: Average time spent waiting in a scheduler's queue
# CHECK-NEXT: [2]: Average time spent waiting in a scheduler's queue while ready
# CHECK-NEXT: [3]: Average time elapsed from WB until retire stage

# CHECK: [0] [1] [2] [3]
# CHECK-NEXT: 0. 2 1.0 1.0 1.5 movq $128, %rsp
# CHECK-NEXT: 1. 2 3.0 0.5 0.0 pushq %rax
# CHECK-NEXT: 2. 2 3.0 1.0 0.0 pushq %rcx
# CHECK-NEXT: 3. 2 4.0 1.0 0.0 pushq %rdx
# CHECK-NEXT: 4. 2 4.0 1.0 0.0 pushq %rbx
# CHECK-NEXT: 5. 2 5.0 1.0 0.0 pushq %r12
# CHECK-NEXT: 2 3.3 0.9 0.3 <total>