Skip to content

Hexagon: Add sincos intrinsic test #147474

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

Jump to bottom
Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Hexagon: Add sincos intrinsic test #147474

wants to merge 1 commit into from

Conversation

arsenm
Copy link
Contributor

@arsenm arsenm commented Jul 8, 2025

No description provided.

This was referenced Jul 8, 2025
Merged
Merged
Merged
@arsenm Graphite App
Copy link
Contributor Author

arsenm commented Jul 8, 2025 •
edited
Loading

This stack of pull requests is managed by Graphite. Learn more about stacking.

@arsenm arsenm mentioned this pull request Jul 8, 2025
Open
@llvmbot
Copy link
Member

llvmbot commented Jul 8, 2025

@llvm/pr-subscribers-backend-hexagon

Author: Matt Arsenault (arsenm)

Changes

Patch is 44.74 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/147474.diff

1 Files Affected:

  • (added) llvm/test/CodeGen/Hexagon/llvm.sincos.ll (+1487)
diff --git a/llvm/test/CodeGen/Hexagon/llvm.sincos.ll b/llvm/test/CodeGen/Hexagon/llvm.sincos.ll
new file mode 100644
index 0000000000000..f02ac2ca8480f
--- /dev/null
+++ b/llvm/test/CodeGen/Hexagon/llvm.sincos.ll
@@ -0,0 +1,1487 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -mtriple=hexagon < %s | FileCheck -check-prefix=BASE %s
+; RUN: llc -mtriple=hexagon-unknown-linux-gnu < %s | FileCheck -check-prefix=GNU %s
+; RUN: llc -mtriple=hexagon-unknown-linux-musl < %s | FileCheck -check-prefix=MUSL %s
+
+define { half, half } @test_sincos_f16(half %a) #0 {
+; BASE-LABEL: test_sincos_f16:
+; BASE:       // %bb.0:
+; BASE-NEXT:    {
+; BASE-NEXT:     call __extendhfsf2
+; BASE-NEXT:     memd(r29+#-16) = r17:16
+; BASE-NEXT:     allocframe(#8)
+; BASE-NEXT:    } // 8-byte Folded Spill
+; BASE-NEXT:    {
+; BASE-NEXT:     call sinf
+; BASE-NEXT:     r16 = r0
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call __truncsfhf2
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call cosf
+; BASE-NEXT:     r17 = r0
+; BASE-NEXT:     r0 = r16
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call __truncsfhf2
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     r1:0 = combine(r0,r17)
+; BASE-NEXT:     r17:16 = memd(r29+#0)
+; BASE-NEXT:     dealloc_return
+; BASE-NEXT:    } // 8-byte Folded Reload
+;
+; GNU-LABEL: test_sincos_f16:
+; GNU:       // %bb.0:
+; GNU-NEXT:    {
+; GNU-NEXT:     call __extendhfsf2
+; GNU-NEXT:     memd(r29+#-16) = r17:16
+; GNU-NEXT:     allocframe(#16)
+; GNU-NEXT:    } // 8-byte Folded Spill
+; GNU-NEXT:    {
+; GNU-NEXT:     call sincosf
+; GNU-NEXT:     r1 = add(r29,#0)
+; GNU-NEXT:     r2 = add(r29,#4)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call __truncsfhf2
+; GNU-NEXT:     r0 = memw(r29+#0)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call __truncsfhf2
+; GNU-NEXT:     r16 = r0
+; GNU-NEXT:     r0 = memw(r29+#4)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     r1:0 = combine(r0,r16)
+; GNU-NEXT:     r17:16 = memd(r29+#8)
+; GNU-NEXT:     dealloc_return
+; GNU-NEXT:    } // 8-byte Folded Reload
+;
+; MUSL-LABEL: test_sincos_f16:
+; MUSL:       // %bb.0:
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __extendhfsf2
+; MUSL-NEXT:     memd(r29+#-16) = r17:16
+; MUSL-NEXT:     allocframe(#8)
+; MUSL-NEXT:    } // 8-byte Folded Spill
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call sinf
+; MUSL-NEXT:     r16 = r0
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __truncsfhf2
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call cosf
+; MUSL-NEXT:     r17 = r0
+; MUSL-NEXT:     r0 = r16
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __truncsfhf2
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     r1:0 = combine(r0,r17)
+; MUSL-NEXT:     r17:16 = memd(r29+#0)
+; MUSL-NEXT:     dealloc_return
+; MUSL-NEXT:    } // 8-byte Folded Reload
+  %result = call { half, half } @llvm.sincos.f16(half %a)
+  ret { half, half } %result
+}
+
+define half @test_sincos_f16_only_use_sin(half %a) #0 {
+; BASE-LABEL: test_sincos_f16_only_use_sin:
+; BASE:       // %bb.0:
+; BASE-NEXT:    {
+; BASE-NEXT:     call __extendhfsf2
+; BASE-NEXT:     allocframe(r29,#0):raw
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call sinf
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call __truncsfhf2
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     r31:30 = dealloc_return(r30):raw
+; BASE-NEXT:    }
+;
+; GNU-LABEL: test_sincos_f16_only_use_sin:
+; GNU:       // %bb.0:
+; GNU-NEXT:    {
+; GNU-NEXT:     call __extendhfsf2
+; GNU-NEXT:     allocframe(r29,#8):raw
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call sincosf
+; GNU-NEXT:     r1 = add(r29,#0)
+; GNU-NEXT:     r2 = add(r29,#4)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call __truncsfhf2
+; GNU-NEXT:     r0 = memw(r29+#0)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     r31:30 = dealloc_return(r30):raw
+; GNU-NEXT:    }
+;
+; MUSL-LABEL: test_sincos_f16_only_use_sin:
+; MUSL:       // %bb.0:
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __extendhfsf2
+; MUSL-NEXT:     allocframe(r29,#0):raw
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call sinf
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __truncsfhf2
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     r31:30 = dealloc_return(r30):raw
+; MUSL-NEXT:    }
+  %result = call { half, half } @llvm.sincos.f16(half %a)
+  %result.0 = extractvalue { half, half } %result, 0
+  ret half %result.0
+}
+
+define half @test_sincos_f16_only_use_cos(half %a) #0 {
+; BASE-LABEL: test_sincos_f16_only_use_cos:
+; BASE:       // %bb.0:
+; BASE-NEXT:    {
+; BASE-NEXT:     call __extendhfsf2
+; BASE-NEXT:     allocframe(r29,#0):raw
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call cosf
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call __truncsfhf2
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     r31:30 = dealloc_return(r30):raw
+; BASE-NEXT:    }
+;
+; GNU-LABEL: test_sincos_f16_only_use_cos:
+; GNU:       // %bb.0:
+; GNU-NEXT:    {
+; GNU-NEXT:     call __extendhfsf2
+; GNU-NEXT:     allocframe(r29,#8):raw
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call sincosf
+; GNU-NEXT:     r1 = add(r29,#0)
+; GNU-NEXT:     r2 = add(r29,#4)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call __truncsfhf2
+; GNU-NEXT:     r0 = memw(r29+#4)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     r31:30 = dealloc_return(r30):raw
+; GNU-NEXT:    }
+;
+; MUSL-LABEL: test_sincos_f16_only_use_cos:
+; MUSL:       // %bb.0:
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __extendhfsf2
+; MUSL-NEXT:     allocframe(r29,#0):raw
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call cosf
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __truncsfhf2
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     r31:30 = dealloc_return(r30):raw
+; MUSL-NEXT:    }
+  %result = call { half, half } @llvm.sincos.f16(half %a)
+  %result.1 = extractvalue { half, half } %result, 1
+  ret half %result.1
+}
+
+define { <2 x half>, <2 x half> } @test_sincos_v2f16(<2 x half> %a) #0 {
+; BASE-LABEL: test_sincos_v2f16:
+; BASE:       // %bb.0:
+; BASE-NEXT:    {
+; BASE-NEXT:     r17:16 = combine(r3,r0)
+; BASE-NEXT:     r0 = r2
+; BASE-NEXT:     memd(r29+#-16) = r17:16
+; BASE-NEXT:     allocframe(#24)
+; BASE-NEXT:    } // 8-byte Folded Spill
+; BASE-NEXT:    {
+; BASE-NEXT:     call __extendhfsf2
+; BASE-NEXT:     memd(r29+#8) = r19:18
+; BASE-NEXT:     memd(r29+#0) = r21:20
+; BASE-NEXT:    } // 8-byte Folded Spill
+; BASE-NEXT:    {
+; BASE-NEXT:     call sinf
+; BASE-NEXT:     r18 = r0
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call __truncsfhf2
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call __extendhfsf2
+; BASE-NEXT:     r19 = r0
+; BASE-NEXT:     r0 = r17
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call sinf
+; BASE-NEXT:     r17 = r0
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call __truncsfhf2
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call cosf
+; BASE-NEXT:     r20 = r0
+; BASE-NEXT:     r0 = r18
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call __truncsfhf2
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call cosf
+; BASE-NEXT:     r18 = r0
+; BASE-NEXT:     r0 = r17
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call __truncsfhf2
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     memh(r16+#6) = r0
+; BASE-NEXT:     memh(r16+#4) = r18
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     memh(r16+#2) = r20
+; BASE-NEXT:     memh(r16+#0) = r19
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     r17:16 = memd(r29+#16)
+; BASE-NEXT:     r19:18 = memd(r29+#8)
+; BASE-NEXT:    } // 8-byte Folded Reload
+; BASE-NEXT:    {
+; BASE-NEXT:     r21:20 = memd(r29+#0)
+; BASE-NEXT:     dealloc_return
+; BASE-NEXT:    } // 8-byte Folded Reload
+;
+; GNU-LABEL: test_sincos_v2f16:
+; GNU:       // %bb.0:
+; GNU-NEXT:    {
+; GNU-NEXT:     r17:16 = combine(r3,r0)
+; GNU-NEXT:     r0 = r2
+; GNU-NEXT:     memd(r29+#-16) = r17:16
+; GNU-NEXT:     allocframe(#32)
+; GNU-NEXT:    } // 8-byte Folded Spill
+; GNU-NEXT:    {
+; GNU-NEXT:     call __extendhfsf2
+; GNU-NEXT:     memd(r29+#16) = r19:18
+; GNU-NEXT:    } // 8-byte Folded Spill
+; GNU-NEXT:    {
+; GNU-NEXT:     call sincosf
+; GNU-NEXT:     r1 = add(r29,#8)
+; GNU-NEXT:     r2 = add(r29,#12)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call __extendhfsf2
+; GNU-NEXT:     r0 = r17
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call sincosf
+; GNU-NEXT:     r1 = add(r29,#0)
+; GNU-NEXT:     r2 = add(r29,#4)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call __truncsfhf2
+; GNU-NEXT:     r0 = memw(r29+#8)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call __truncsfhf2
+; GNU-NEXT:     r17 = r0
+; GNU-NEXT:     r0 = memw(r29+#0)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call __truncsfhf2
+; GNU-NEXT:     r18 = r0
+; GNU-NEXT:     r0 = memw(r29+#12)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call __truncsfhf2
+; GNU-NEXT:     r19 = r0
+; GNU-NEXT:     r0 = memw(r29+#4)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     memh(r16+#6) = r0
+; GNU-NEXT:     memh(r16+#4) = r19
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     memh(r16+#2) = r18
+; GNU-NEXT:     memh(r16+#0) = r17
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     r17:16 = memd(r29+#24)
+; GNU-NEXT:     r19:18 = memd(r29+#16)
+; GNU-NEXT:    } // 8-byte Folded Reload
+; GNU-NEXT:    {
+; GNU-NEXT:     r31:30 = dealloc_return(r30):raw
+; GNU-NEXT:    }
+;
+; MUSL-LABEL: test_sincos_v2f16:
+; MUSL:       // %bb.0:
+; MUSL-NEXT:    {
+; MUSL-NEXT:     r17:16 = combine(r3,r0)
+; MUSL-NEXT:     r0 = r2
+; MUSL-NEXT:     memd(r29+#-16) = r17:16
+; MUSL-NEXT:     allocframe(#24)
+; MUSL-NEXT:    } // 8-byte Folded Spill
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __extendhfsf2
+; MUSL-NEXT:     memd(r29+#8) = r19:18
+; MUSL-NEXT:     memd(r29+#0) = r21:20
+; MUSL-NEXT:    } // 8-byte Folded Spill
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call sinf
+; MUSL-NEXT:     r18 = r0
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __truncsfhf2
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __extendhfsf2
+; MUSL-NEXT:     r19 = r0
+; MUSL-NEXT:     r0 = r17
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call sinf
+; MUSL-NEXT:     r17 = r0
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __truncsfhf2
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call cosf
+; MUSL-NEXT:     r20 = r0
+; MUSL-NEXT:     r0 = r18
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __truncsfhf2
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call cosf
+; MUSL-NEXT:     r18 = r0
+; MUSL-NEXT:     r0 = r17
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call __truncsfhf2
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     memh(r16+#6) = r0
+; MUSL-NEXT:     memh(r16+#4) = r18
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     memh(r16+#2) = r20
+; MUSL-NEXT:     memh(r16+#0) = r19
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     r17:16 = memd(r29+#16)
+; MUSL-NEXT:     r19:18 = memd(r29+#8)
+; MUSL-NEXT:    } // 8-byte Folded Reload
+; MUSL-NEXT:    {
+; MUSL-NEXT:     r21:20 = memd(r29+#0)
+; MUSL-NEXT:     dealloc_return
+; MUSL-NEXT:    } // 8-byte Folded Reload
+  %result = call { <2 x half>, <2 x half> } @llvm.sincos.v2f16(<2 x half> %a)
+  ret { <2 x half>, <2 x half> } %result
+}
+
+define { float, float } @test_sincos_f32(float %a) #0 {
+; BASE-LABEL: test_sincos_f32:
+; BASE:       // %bb.0:
+; BASE-NEXT:    {
+; BASE-NEXT:     call sinf
+; BASE-NEXT:     r16 = r0
+; BASE-NEXT:     memd(r29+#-16) = r17:16
+; BASE-NEXT:     allocframe(#8)
+; BASE-NEXT:    } // 8-byte Folded Spill
+; BASE-NEXT:    {
+; BASE-NEXT:     call cosf
+; BASE-NEXT:     r17 = r0
+; BASE-NEXT:     r0 = r16
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     r1:0 = combine(r0,r17)
+; BASE-NEXT:     r17:16 = memd(r29+#0)
+; BASE-NEXT:     dealloc_return
+; BASE-NEXT:    } // 8-byte Folded Reload
+;
+; GNU-LABEL: test_sincos_f32:
+; GNU:       // %bb.0:
+; GNU-NEXT:    {
+; GNU-NEXT:     allocframe(r29,#8):raw
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call sincosf
+; GNU-NEXT:     r1 = add(r29,#0)
+; GNU-NEXT:     r2 = add(r29,#4)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     r0 = memw(r29+#0)
+; GNU-NEXT:     r1 = memw(r29+#4)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     r31:30 = dealloc_return(r30):raw
+; GNU-NEXT:    }
+;
+; MUSL-LABEL: test_sincos_f32:
+; MUSL:       // %bb.0:
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call sinf
+; MUSL-NEXT:     r16 = r0
+; MUSL-NEXT:     memd(r29+#-16) = r17:16
+; MUSL-NEXT:     allocframe(#8)
+; MUSL-NEXT:    } // 8-byte Folded Spill
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call cosf
+; MUSL-NEXT:     r17 = r0
+; MUSL-NEXT:     r0 = r16
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     r1:0 = combine(r0,r17)
+; MUSL-NEXT:     r17:16 = memd(r29+#0)
+; MUSL-NEXT:     dealloc_return
+; MUSL-NEXT:    } // 8-byte Folded Reload
+  %result = call { float, float } @llvm.sincos.f32(float %a)
+  ret { float, float } %result
+}
+
+define { <2 x float>, <2 x float> } @test_sincos_v2f32(<2 x float> %a) #0 {
+; CHECKR2-LABEL: test_sincos_v2f32:
+; CHECKR2:       # %bb.0:
+; CHECKR2-NEXT:    daddiu $sp, $sp, -32
+; CHECKR2-NEXT:    sd $ra, 24($sp) # 8-byte Folded Spill
+; CHECKR2-NEXT:    sd $16, 16($sp) # 8-byte Folded Spill
+; CHECKR2-NEXT:    move $16, $4
+; CHECKR2-NEXT:    dsrl $1, $4, 32
+; CHECKR2-NEXT:    sll $1, $1, 0
+; CHECKR2-NEXT:    mtc1 $1, $f12
+; CHECKR2-NEXT:    daddiu $5, $sp, 12
+; CHECKR2-NEXT:    jal sincosf
+; CHECKR2-NEXT:    daddiu $6, $sp, 8
+; CHECKR2-NEXT:    sll $1, $16, 0
+; CHECKR2-NEXT:    mtc1 $1, $f12
+; CHECKR2-NEXT:    daddiu $5, $sp, 4
+; CHECKR2-NEXT:    jal sincosf
+; CHECKR2-NEXT:    daddiu $6, $sp, 0
+; CHECKR2-NEXT:    lwc1 $f0, 12($sp)
+; CHECKR2-NEXT:    mfc1 $1, $f0
+; CHECKR2-NEXT:    dsll $1, $1, 32
+; CHECKR2-NEXT:    lwc1 $f0, 4($sp)
+; CHECKR2-NEXT:    mfc1 $2, $f0
+; CHECKR2-NEXT:    dext $2, $2, 0, 32
+; CHECKR2-NEXT:    lwc1 $f0, 8($sp)
+; CHECKR2-NEXT:    or $2, $2, $1
+; CHECKR2-NEXT:    mfc1 $1, $f0
+; CHECKR2-NEXT:    dsll $1, $1, 32
+; CHECKR2-NEXT:    lwc1 $f0, 0($sp)
+; CHECKR2-NEXT:    mfc1 $3, $f0
+; CHECKR2-NEXT:    dext $3, $3, 0, 32
+; CHECKR2-NEXT:    or $3, $3, $1
+; CHECKR2-NEXT:    ld $16, 16($sp) # 8-byte Folded Reload
+; CHECKR2-NEXT:    ld $ra, 24($sp) # 8-byte Folded Reload
+; CHECKR2-NEXT:    jr $ra
+; CHECKR2-NEXT:    daddiu $sp, $sp, 32
+; BASE-LABEL: test_sincos_v2f32:
+; BASE:       // %bb.0:
+; BASE-NEXT:    {
+; BASE-NEXT:     r17:16 = combine(r0,r2)
+; BASE-NEXT:     r0 = r2
+; BASE-NEXT:     memd(r29+#-16) = r17:16
+; BASE-NEXT:     allocframe(#24)
+; BASE-NEXT:    } // 8-byte Folded Spill
+; BASE-NEXT:    {
+; BASE-NEXT:     call sinf
+; BASE-NEXT:     r18 = r3
+; BASE-NEXT:     memd(r29+#8) = r19:18
+; BASE-NEXT:     memd(r29+#0) = r21:20
+; BASE-NEXT:    } // 8-byte Folded Spill
+; BASE-NEXT:    {
+; BASE-NEXT:     call sinf
+; BASE-NEXT:     r19 = r0
+; BASE-NEXT:     r0 = r18
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call cosf
+; BASE-NEXT:     r20 = r0
+; BASE-NEXT:     r0 = r16
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     call cosf
+; BASE-NEXT:     r16 = r0
+; BASE-NEXT:     r0 = r18
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     r1:0 = combine(r0,r16)
+; BASE-NEXT:     r3:2 = combine(r20,r19)
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     memd(r17+#8) = r1:0
+; BASE-NEXT:     memd(r17+#0) = r3:2
+; BASE-NEXT:    }
+; BASE-NEXT:    {
+; BASE-NEXT:     r17:16 = memd(r29+#16)
+; BASE-NEXT:     r19:18 = memd(r29+#8)
+; BASE-NEXT:    } // 8-byte Folded Reload
+; BASE-NEXT:    {
+; BASE-NEXT:     r21:20 = memd(r29+#0)
+; BASE-NEXT:     dealloc_return
+; BASE-NEXT:    } // 8-byte Folded Reload
+;
+; GNU-LABEL: test_sincos_v2f32:
+; GNU:       // %bb.0:
+; GNU-NEXT:    {
+; GNU-NEXT:     r17:16 = combine(r0,r2)
+; GNU-NEXT:     r1 = add(r0,#4)
+; GNU-NEXT:     memd(r29+#-16) = r17:16
+; GNU-NEXT:     allocframe(#8)
+; GNU-NEXT:    } // 8-byte Folded Spill
+; GNU-NEXT:    {
+; GNU-NEXT:     call sincosf
+; GNU-NEXT:     r2 = add(r0,#12)
+; GNU-NEXT:     r0 = r3
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     call sincosf
+; GNU-NEXT:     r2 = add(r17,#8)
+; GNU-NEXT:     r1:0 = combine(r17,r16)
+; GNU-NEXT:    }
+; GNU-NEXT:    {
+; GNU-NEXT:     r17:16 = memd(r29+#0)
+; GNU-NEXT:     dealloc_return
+; GNU-NEXT:    } // 8-byte Folded Reload
+;
+; MUSL-LABEL: test_sincos_v2f32:
+; MUSL:       // %bb.0:
+; MUSL-NEXT:    {
+; MUSL-NEXT:     r17:16 = combine(r0,r2)
+; MUSL-NEXT:     r0 = r2
+; MUSL-NEXT:     memd(r29+#-16) = r17:16
+; MUSL-NEXT:     allocframe(#24)
+; MUSL-NEXT:    } // 8-byte Folded Spill
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call sinf
+; MUSL-NEXT:     r18 = r3
+; MUSL-NEXT:     memd(r29+#8) = r19:18
+; MUSL-NEXT:     memd(r29+#0) = r21:20
+; MUSL-NEXT:    } // 8-byte Folded Spill
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call sinf
+; MUSL-NEXT:     r19 = r0
+; MUSL-NEXT:     r0 = r18
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call cosf
+; MUSL-NEXT:     r20 = r0
+; MUSL-NEXT:     r0 = r16
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     call cosf
+; MUSL-NEXT:     r16 = r0
+; MUSL-NEXT:     r0 = r18
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     r1:0 = combine(r0,r16)
+; MUSL-NEXT:     r3:2 = combine(r20,r19)
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     memd(r17+#8) = r1:0
+; MUSL-NEXT:     memd(r17+#0) = r3:2
+; MUSL-NEXT:    }
+; MUSL-NEXT:    {
+; MUSL-NEXT:     r17:16 = memd(r29+#16)
+; MUSL-NEXT:     r19:18 = memd(r29+#8)
+; MUSL-NEXT:    } // 8-byte Folded Reload
+; MUSL-NEXT:    {
+; MUSL-NEXT:     r21:20 = memd(r29+#0)
+; MUSL-NEXT:     dealloc_return
+; MUSL-NEXT:    } // 8-byte Folded Reload
+  %result = call { <2 x float>, <2 x float> } @llvm.sincos.v2f32(<2 x float> %a)
+  ret { <2 x float>, <2 x float> } %result
+}
+
+define { <3 x float>, <3 x float> } @test_sincos_v3f32(<3 x float> %a) #0 {
+; CHECKR2-LABEL: test_sincos_v3f32:
+; CHECKR2:       # %bb.0:
+; CHECKR2-NEXT:    daddiu $sp, $sp, -48
+; CHECKR2-NEXT:    sdc1 $f25, 40($sp) # 8-byte Folded Spill
+; CHECKR2-NEXT:    sdc1 $f24, 32($sp) # 8-byte Folded Spill
+; CHECKR2-NEXT:    sd $ra, 24($sp) # 8-byte Folded Spill
+; CHECKR2-NEXT:    sd $16, 16($sp) # 8-byte Folded Spill
+; CHECKR2-NEXT:    mov.s $f24, $f15
+; CHECKR2-NEXT:    mov.s $f25, $f14
+; CHECKR2-NEXT:    mov.s $f12, $f13
+; CHECKR2-NEXT:    move $16, $4
+; CHECKR2-NEXT:    daddiu $5, $sp, 4
+; CHECKR2-NEXT:    jal sincosf
+; CHECKR2-NEXT:    daddiu $6, $sp, 0
+; CHECKR2-NEXT:    daddiu $5, $sp, 12
+; CHECKR2-NEXT:    daddiu $6, $sp, 8
+; CHECKR2-NEXT:    jal sincosf
+; CHECKR2-NEXT:    mov.s $f12, $f25
+; CHECKR2-NEXT:    daddiu $5, $16, 8
+; CHECKR2-NEXT:    daddiu $6, $16, 24
+; CHECKR2-NEXT:    jal sincosf
+; CHECKR2-NEXT:    mov.s $f12, $f24
+; CHECKR2-NEXT:    lwc1 $f0, 0($sp)
+; CHECKR2-NEXT:    mfc1 $1, $f0
+; CHECKR2-NEXT:    dsll $1, $1, 32
+; CHECKR2-NEXT:    lwc1 $f0, 8($sp)
+; CHECKR2-NEXT:    mfc1 $2, $f0
+; CHECKR2-NEXT:    dext $2, $2, 0, 32
+; CHECKR2-NEXT:    or $1, $2, $1
+; CHECKR2-NEXT:    lwc1 $f0, 4($sp)
+; CHECKR2-NEXT:    mfc1 $2, $f0
+; CHECKR2-NEXT:    sd $1, 16($16)
+; CHECKR2-NEXT:    dsll $1, $2, 32
+; CHECKR2-NEXT:    lwc1 $f0, 12($sp)
+; CHECKR2-NEXT:    mfc1 $2, $f0
+; CHECKR2-NEXT:    dext $2, $2, 0, 32
+; CHECKR2-NEXT:    or $1, $2, $1
+; CHECKR2-NEXT:    sd $1, 0($16)
+; CHECKR2-NEXT:    ld $16, 16($sp) # 8-byte Folded Reload
+; CHECKR2-NEXT:    ld $ra, 24($sp) # 8-byte Folded Reload
+; CHECKR2-NEXT:    ldc1 $f24, 32($sp) # 8-byte Folded Reload
+; CHECKR2-NEXT:    ldc1 $f25, 40($sp) # 8-byte Folded Reload
+; CHECKR2-NEXT:    jr $ra
+; CHECKR2-NEXT:    daddiu $sp, $sp, 48
+; BASE-LABEL: test_sincos_v3f32:
+; BASE:       // %bb.0:
+; BASE-NEXT:    {
+; BASE-NEXT:     memd(r29+#-16) = r17:16
+; BASE-NEXT:     allocframe(#32)
+; BASE-NEXT:    } // 8-byte Folded Spill
+; BASE-NEXT:    {
+; BASE-NEXT:     r19:18 = combine(r0,r2)
+; BASE-NEXT:     r0 = r2
+; BASE-NEXT:     memd(r29+#16) = r19:18
+; BASE-NEXT:     memd(r29+#8) = r21:20
+; BASE-NEXT:    } // 8-byte Folded Spill
+; BASE-NEXT:    {
+; BASE-NEXT:     call sinf
+; BAS...
[truncated]

@arsenm arsenm marked this pull request as ready for review July 8, 2025 08:01
@arsenm arsenm force-pushed the users/arsenm/hexagon/add-sincos-test branch from 71feda6 to b154356 Compare July 8, 2025 15:25
@arsenm arsenm force-pushed the users/arsenm/dag/fix-assert-soften-sincos-no-libcall branch 2 times, most recently from 248eb92 to aceabe9 Compare July 8, 2025 16:49
Base automatically changed from users/arsenm/dag/fix-assert-soften-sincos-no-libcall to main July 8, 2025 16:52
@arsenm arsenm force-pushed the users/arsenm/hexagon/add-sincos-test branch from b154356 to 281125f Compare July 8, 2025 17:04
aankit-ca
aankit-ca reviewed Jul 8, 2025
View reviewed changes
Copy link
Contributor

@aankit-ca aankit-ca left a comment

Choose a reason for hiding this comment

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

The tests seem to be brittle. Can we check only what's needed to be checked?

@arsenm
Copy link
Contributor Author

arsenm commented Jul 9, 2025

The tests seem to be brittle. Can we check only what's needed to be checked?

They are generated. I will not be adding manual checks for this

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@aankit-ca aankit-ca aankit-ca left review comments

@jverma-quic jverma-quic Awaiting requested review from jverma-quic

@kparzysz kparzysz Awaiting requested review from kparzysz

@SundeepKushwaha SundeepKushwaha Awaiting requested review from SundeepKushwaha

Assignees
No one assigned
Labels
backend:Hexagon
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@arsenm @llvmbot @aankit-ca