Skip to content

[RISCV][VLOPT] Add vector indexed loads and stores to getOperandInfo #119748

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

Merged
merged 9 commits into from
Dec 18, 2024
Merged

[RISCV][VLOPT] Add vector indexed loads and stores to getOperandInfo #119748

Show file tree
Hide file tree
Changes from 6 commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
8b59858
[RISCV][VLOPT] Add vector indexed instructions to getOperandInfo
michaelmaitland Dec 12, 2024
cf9c153
fixup! add load support
michaelmaitland Dec 12, 2024
d938a67
fixup! improve tests and variable naming
michaelmaitland Dec 17, 2024
9bbb4c2
fixup! fix tests
michaelmaitland Dec 18, 2024
eea71a9
fixup! move vmop_mm back down
michaelmaitland Dec 18, 2024
c775102
fixup! improve test case
michaelmaitland Dec 18, 2024
f695f14
fixup! clean tests
michaelmaitland Dec 18, 2024
1f01333
fixup! name of instr
michaelmaitland Dec 18, 2024
82d5178
fixup! update test checks after rebase
michaelmaitland Dec 18, 2024
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
37 changes: 37 additions & 0 deletions llvm/lib/Target/RISCV/RISCVVLOptimizer.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -270,6 +270,43 @@ static OperandInfo getOperandInfo(const MachineOperand &MO,
case RISCV::VSSE64_V:
return OperandInfo(RISCVVType::getEMULEqualsEEWDivSEWTimesLMUL(6, MI), 6);

// Vector Indexed Instructions
// vs(o|u)xei<eew>.v
// Dest/Data (operand 0) EEW=SEW, EMUL=LMUL. Source EEW=<eew> and
// EMUL=(EEW/SEW)*LMUL.
case RISCV::VLUXEI8_V:
case RISCV::VLOXEI8_V:
case RISCV::VSUXEI8_V:
case RISCV::VSOXEI8_V: {
if (MO.getOperandNo() == 0)
return OperandInfo(MIVLMul, MILog2SEW);
return OperandInfo(RISCVVType::getEMULEqualsEEWDivSEWTimesLMUL(3, MI), 3);
}
case RISCV::VLUXEI16_V:
case RISCV::VLOXEI16_V:
case RISCV::VSUXEI16_V:
case RISCV::VSOXEI16_V: {
if (MO.getOperandNo() == 0)
return OperandInfo(MIVLMul, MILog2SEW);
return OperandInfo(RISCVVType::getEMULEqualsEEWDivSEWTimesLMUL(4, MI), 4);
}
case RISCV::VLUXEI32_V:
case RISCV::VLOXEI32_V:
case RISCV::VSUXEI32_V:
case RISCV::VSOXEI32_V: {
if (MO.getOperandNo() == 0)
return OperandInfo(MIVLMul, MILog2SEW);
return OperandInfo(RISCVVType::getEMULEqualsEEWDivSEWTimesLMUL(5, MI), 5);
}
case RISCV::VLUXEI64_V:
case RISCV::VLOXEI64_V:
case RISCV::VSUXEI64_V:
case RISCV::VSOXEI64_V: {
if (MO.getOperandNo() == 0)
return OperandInfo(MIVLMul, MILog2SEW);
return OperandInfo(RISCVVType::getEMULEqualsEEWDivSEWTimesLMUL(6, MI), 6);
}

// Vector Integer Arithmetic Instructions
// Vector Single-Width Integer Add and Subtract
case RISCV::VADD_VI:
Expand Down
110 changes: 110 additions & 0 deletions llvm/test/CodeGen/RISCV/rvv/vl-opt-op-info.mir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -573,6 +573,116 @@ body: |
PseudoVSSE8_V_MF2 %x, $noreg, $noreg, 1, 3 /* e8 */
...
---
name: vsuxeiN_v_data
body: |
bb.0:
; CHECK-LABEL: name: vsuxeiN_v_data
; CHECK: %x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, 1, 3 /* e8 */, 0 /* tu, mu */
; CHECK-NEXT: PseudoVSUXEI8_V_M1_M1 %x, $noreg, $noreg, 1, 3 /* e8 */
%x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 3 /* e8 */, 0
PseudoVSUXEI8_V_M1_M1 %x, $noreg, $noreg, 1, 3 /* e8 */
...
---
name: vsuxeiN_v_data_incompatible_eew
body: |
bb.0:
; CHECK-LABEL: name: vsuxeiN_v_data_incompatible_eew
; CHECK: %x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 4 /* e16 */, 0 /* tu, mu */
; CHECK-NEXT: PseudoVSUXEI8_V_M1_M1 %x, $noreg, $noreg, 1, 3 /* e8 */
%x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 4 /* e16 */, 0
PseudoVSUXEI8_V_M1_M1 %x, $noreg, $noreg, 1, 3 /* e8 */
...
---
name: vsuxeiN_v_data_incompatible_emul
body: |
bb.0:
; CHECK-LABEL: name: vsuxeiN_v_data_incompatible_emul
; CHECK: %x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 3 /* e8 */, 0 /* tu, mu */
; CHECK-NEXT: PseudoVSUXEI8_V_MF2_MF2 %x, $noreg, $noreg, 1, 3 /* e8 */
%x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 3 /* e8 */, 0
PseudoVSUXEI8_V_MF2_MF2 %x, $noreg, $noreg, 1, 3 /* e8 */
...
---
name: vsuxeiN_v_idx
body: |
bb.0:
; CHECK-LABEL: name: vsuxeiN_v_idx
; CHECK: %x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, 1, 3 /* e8 */, 0 /* tu, mu */
; CHECK-NEXT: PseudoVSUXEI8_V_M1_M1 $noreg, $noreg, %x, 1, 3 /* e8 */
%x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 3 /* e8 */, 0
PseudoVSUXEI8_V_M1_M1 $noreg, $noreg, %x, 1, 3 /* e8 */
...
---
name: vsuxeiN_v_idx_incompatible_eew
body: |
bb.0:
; CHECK-LABEL: name: vsuxeiN_v_idx_incompatible_eew
; CHECK: %x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 4 /* e16 */, 0 /* tu, mu */
; CHECK-NEXT: %y:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 3 /* e8 */, 0 /* tu, mu */
; CHECK-NEXT: PseudoVSUXEI8_V_M1_M1 $noreg, $noreg, %x, 1, 3 /* e8 */
%x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 4 /* e16 */, 0
%y:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 3 /* e8 */, 0
PseudoVSUXEI8_V_M1_M1 $noreg, $noreg, %x, 1, 3 /* e8 */
...
---
name: vsuxeiN_v_idx_incompatible_emul
body: |
bb.0:
; CHECK-LABEL: name: vsuxeiN_v_idx_incompatible_emul
; CHECK: %x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 3 /* e8 */, 0 /* tu, mu */
; CHECK-NEXT: PseudoVSUXEI8_V_MF2_MF2 $noreg, $noreg, %x, 1, 3 /* e8 */
%x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 3 /* e8 */, 0
PseudoVSUXEI8_V_MF2_MF2 $noreg, $noreg, %x, 1, 3 /* e8 */
...
---
name: vluxeiN_v
body: |
bb.0:
; CHECK-LABEL: name: vluxeiN_v
; CHECK: %x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, 1, 3 /* e8 */, 0 /* tu, mu */
; CHECK-NEXT: %y:vr = PseudoVLUXEI8_V_M1_M1 $noreg, $noreg, %x, 1, 3 /* e8 */, 0 /* tu, mu */
; CHECK-NEXT: %z:vr = PseudoVADD_VV_M1 $noreg, %y, $noreg, 1, 3 /* e8 */, 0 /* tu, mu */
%x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 3 /* e8 */, 0
%y:vr = PseudoVLUXEI8_V_M1_M1 $noreg, $noreg, %x, 1, 3 /* e8 */, 0
%z:vr = PseudoVADD_VV_M1 $noreg, %y, $noreg, 1, 3 /* e8 */, 0
...
---
name: vluxeiN_v_incompatible_eew
body: |
bb.0:
; CHECK-LABEL: name: vluxeiN_v_incompatible_eew
; CHECK: %x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 4 /* e16 */, 0 /* tu, mu */
; CHECK-NEXT: %y:vr = PseudoVLUXEI8_V_M1_M1 $noreg, $noreg, %x, 1, 3 /* e8 */, 0 /* tu, mu */
; CHECK-NEXT: %z:vr = PseudoVADD_VV_M1 $noreg, %y, $noreg, 1, 3 /* e8 */, 0 /* tu, mu */
%x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 4 /* e16 */, 0
%y:vr = PseudoVLUXEI8_V_M1_M1 $noreg, $noreg, %x, 1, 3 /* e8 */, 0
%z:vr = PseudoVADD_VV_M1 $noreg, %y, $noreg, 1, 3 /* e8 */, 0
...
---
name: vluxeiN_v_data_incompatible_emul
body: |
bb.0:
; CHECK-LABEL: name: vluxeiN_v_data_incompatible_emul
; CHECK: %x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 3 /* e8 */, 0 /* tu, mu */
; CHECK-NEXT: %y:vr = PseudoVLUXEI8_V_MF2_MF2 $noreg, $noreg, %x, 1, 3 /* e8 */, 0 /* tu, mu */
; CHECK-NEXT: %z:vr = PseudoVADD_VV_MF2 $noreg, %y, $noreg, 1, 3 /* e8 */, 0 /* tu, mu */
%x:vr = PseudoVADD_VV_M1 $noreg, $noreg, $noreg, -1, 3 /* e8 */, 0
%y:vr = PseudoVLUXEI8_V_MF2_MF2 $noreg, $noreg, %x, 1, 3 /* e8 */, 0
%z:vr = PseudoVADD_VV_MF2 $noreg, %y, $noreg, 1, 3 /* e8 */, 0
...
---
name: vluxeiN_v_idx_incompatible_emul
Copy link
Collaborator

@topperc topperc Dec 18, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's no longer incompatible. It's being optimized.

Copy link
Contributor Author

@michaelmaitland michaelmaitland Dec 18, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

changed.

body: |
bb.0:
; CHECK-LABEL: name: vluxeiN_v_idx_incompatible_emul
; CHECK: %x:vr = PseudoVADD_VV_MF2 $noreg, $noreg, $noreg, 1, 3 /* e8 */, 0 /* tu, mu */
; CHECK-NEXT: %y:vr = PseudoVLUXEI8_V_M1_M1 $noreg, $noreg, %x, 1, 4 /* e16 */, 0 /* tu, mu */
Copy link
Collaborator

@topperc topperc Dec 18, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think this instruction is malformed. One of the M1 needs to be MF2 to match the VADD.

Copy link
Collaborator

@topperc topperc Dec 18, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think the index is first so it should be PseudoVLUXEI8_V_MF2_M1

Copy link
Contributor Author

@michaelmaitland michaelmaitland Dec 18, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

addressed.

; CHECK-NEXT: %z:vr = PseudoVADD_VV_M1 $noreg, %y, $noreg, 1, 5 /* e32 */, 0 /* tu, mu */
%x:vr = PseudoVADD_VV_MF2 $noreg, $noreg, $noreg, -1, 3 /* e8 */, 0
%y:vr = PseudoVLUXEI8_V_M1_M1 $noreg, $noreg, %x, 1, 4 /* e16 */, 0
%z:vr = PseudoVADD_VV_M1 $noreg, %y, $noreg, 1, 5 /* e32 */, 0
Copy link
Collaborator

@topperc topperc Dec 18, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What is the VADD for? It has the same LMUL but different EEW than the VLUX dest.

Copy link
Contributor Author

@michaelmaitland michaelmaitland Dec 18, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

removed.

...
---
name: vmop_mm
body: |
bb.0:
Expand Down
Loading