Skip to content

[RISCV] Refine vector division latencies in SiFive P600's scheduling model #115038

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
Merged
merged 1 commit into from
Nov 6, 2024
Merged

[RISCV] Refine vector division latencies in SiFive P600's scheduling model #115038

merged 1 commit into from
Nov 6, 2024

Conversation

mshockwave
Copy link
Member

@mshockwave mshockwave commented Nov 5, 2024 •
edited
Loading

For both vector integer and floating point divisions.

@llvmbot
Copy link
Member

llvmbot commented Nov 5, 2024

@llvm/pr-subscribers-backend-risc-v

Author: Min-Yih Hsu (mshockwave)

Changes

For both integer and floating point divisions.

Patch is 88.63 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/115038.diff

2 Files Affected:

  • (modified) llvm/lib/Target/RISCV/RISCVSchedSiFiveP600.td (+11-4)
  • (added) llvm/test/tools/llvm-mca/RISCV/SiFiveP600/div.s (+1012)
diff --git a/llvm/lib/Target/RISCV/RISCVSchedSiFiveP600.td b/llvm/lib/Target/RISCV/RISCVSchedSiFiveP600.td
index 51aa003139fbad..a9e3c6766d1214 100644
--- a/llvm/lib/Target/RISCV/RISCVSchedSiFiveP600.td
+++ b/llvm/lib/Target/RISCV/RISCVSchedSiFiveP600.td
@@ -443,12 +443,16 @@ foreach mx = SchedMxListW in {
   }
 }
 
-// Worst case needs 64 cycles if SEW is equal to 64.
+// Worst case needs 51/45/42/72 * lmul cycles for i8/16/32/64.
 foreach mx = SchedMxList in {
   foreach sew = SchedSEWSet<mx>.val in {
     defvar LMulLat = SiFiveP600GetLMulCycles<mx>.c;
     defvar IsWorstCase = SiFiveP600IsWorstCaseMXSEW<mx, sew, SchedMxList>.c;
-    let Latency = 64, ReleaseAtCycles = [LMulLat, !mul(63, LMulLat)] in {
+    defvar DivMicroOpLat =
+      !cond(!eq(sew, 8): 51, !eq(sew, 16): 45, !eq(sew, 32): 42,
+      /* SEW=64 */ true: 72);
+    defvar DivLatency = !mul(DivMicroOpLat, LMulLat);
+    let Latency = DivLatency, ReleaseAtCycles = [LMulLat, DivLatency] in {
       defm "" : LMULSEWWriteResMXSEW<"WriteVIDivV", [SiFiveP600VEXQ1, SiFiveP600VDiv], mx, sew, IsWorstCase>;
       defm "" : LMULSEWWriteResMXSEW<"WriteVIDivX", [SiFiveP600VEXQ1, SiFiveP600VDiv], mx, sew, IsWorstCase>;
     }
@@ -583,12 +587,15 @@ foreach mx = SchedMxListFW in {
   }
 }
 
-// Worst case needs 76 cycles if SEW is equal to 64.
+// Worst case needs around 29/25/37 * LMUL cycles for f16/32/64.
 foreach mx = SchedMxListF in {
   foreach sew = SchedSEWSet<mx, 1>.val in {
     defvar LMulLat = SiFiveP600GetLMulCycles<mx>.c;
     defvar IsWorstCase = SiFiveP600IsWorstCaseMXSEW<mx, sew, SchedMxListF, 1>.c;
-    let Latency = 76, ReleaseAtCycles = [LMulLat, !mul(76, LMulLat)] in {
+    defvar DivMicroOpLat =
+      !cond(!eq(sew, 16): 29, !eq(sew, 32): 25, /* SEW=64 */ true: 37);
+    defvar DivLatency = !mul(DivMicroOpLat, LMulLat);
+    let Latency = DivLatency, ReleaseAtCycles = [LMulLat, DivLatency] in {
       defm "" : LMULSEWWriteResMXSEW<"WriteVFDivV",  [SiFiveP600VEXQ1, SiFiveP600VFloatDiv], mx, sew, IsWorstCase>;
       defm "" : LMULSEWWriteResMXSEW<"WriteVFDivF",  [SiFiveP600VEXQ1, SiFiveP600VFloatDiv], mx, sew, IsWorstCase>;
       defm "" : LMULSEWWriteResMXSEW<"WriteVFSqrtV", [SiFiveP600VEXQ1, SiFiveP600VFloatDiv], mx, sew, IsWorstCase>;
diff --git a/llvm/test/tools/llvm-mca/RISCV/SiFiveP600/div.s b/llvm/test/tools/llvm-mca/RISCV/SiFiveP600/div.s
new file mode 100644
index 00000000000000..0b7cc95dcb8d62
--- /dev/null
+++ b/llvm/test/tools/llvm-mca/RISCV/SiFiveP600/div.s
@@ -0,0 +1,1012 @@
+# NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
+# RUN: llvm-mca -mtriple=riscv64 -mcpu=sifive-p670 -iterations=1 < %s | FileCheck %s
+
+vsetvli zero, zero, e8, mf8, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, mf4, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, mf2, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, m1, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, m1, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, m2, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, m4, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, m8, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, mf8, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, mf4, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, mf2, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, m1, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, m1, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, m2, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, m4, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, m8, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, mf8, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, mf4, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, mf2, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, m1, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, m1, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, m2, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, m4, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, m8, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, mf8, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, mf4, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, mf2, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, m1, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, m1, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, m2, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, m4, tu, mu
+vdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, m8, tu, mu
+vdiv.vv v8, v16, v24
+
+vsetvli zero, zero, e8, mf8, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e8, mf4, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e8, mf2, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e8, m1, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e8, m1, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e8, m2, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e8, m4, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e8, m8, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e16, mf8, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e16, mf4, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e16, mf2, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e16, m1, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e16, m1, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e16, m2, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e16, m4, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e16, m8, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e32, mf8, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e32, mf4, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e32, mf2, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e32, m1, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e32, m1, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e32, m2, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e32, m4, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e32, m8, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e64, mf8, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e64, mf4, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e64, mf2, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e64, m1, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e64, m1, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e64, m2, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e64, m4, tu, mu
+vdiv.vx v8, v16, a0
+vsetvli zero, zero, e64, m8, tu, mu
+vdiv.vx v8, v16, a0
+
+vsetvli zero, zero, e8, mf8, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, mf4, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, mf2, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, m1, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, m1, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, m2, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, m4, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e8, m8, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, mf8, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, mf4, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, mf2, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, m1, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, m1, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, m2, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, m4, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e16, m8, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, mf8, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, mf4, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, mf2, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, m1, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, m1, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, m2, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, m4, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e32, m8, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, mf8, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, mf4, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, mf2, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, m1, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, m1, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, m2, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, m4, tu, mu
+vfdiv.vv v8, v16, v24
+vsetvli zero, zero, e64, m8, tu, mu
+vfdiv.vv v8, v16, v24
+
+vsetvli zero, zero, e8, mf8, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e8, mf4, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e8, mf2, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e8, m1, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e8, m1, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e8, m2, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e8, m4, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e8, m8, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e16, mf8, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e16, mf4, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e16, mf2, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e16, m1, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e16, m1, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e16, m2, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e16, m4, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e16, m8, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e32, mf8, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e32, mf4, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e32, mf2, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e32, m1, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e32, m1, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e32, m2, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e32, m4, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e32, m8, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e64, mf8, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e64, mf4, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e64, mf2, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e64, m1, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e64, m1, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e64, m2, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e64, m4, tu, mu
+vfdiv.vf v8, v16, fa0
+vsetvli zero, zero, e64, m8, tu, mu
+vfdiv.vf v8, v16, fa0
+
+vsetvli zero, zero, e8, mf8, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e8, mf4, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e8, mf2, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e8, m1, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e8, m1, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e8, m2, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e8, m4, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e8, m8, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e16, mf8, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e16, mf4, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e16, mf2, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e16, m1, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e16, m1, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e16, m2, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e16, m4, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e16, m8, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e32, mf8, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e32, mf4, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e32, mf2, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e32, m1, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e32, m1, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e32, m2, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e32, m4, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e32, m8, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e64, mf8, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e64, mf4, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e64, mf2, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e64, m1, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e64, m1, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e64, m2, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e64, m4, tu, mu
+vfsqrt.v v8, v16
+vsetvli zero, zero, e64, m8, tu, mu
+vfsqrt.v v8, v16
+
+# CHECK:      Iterations:        1
+# CHECK-NEXT: Instructions:      320
+# CHECK-NEXT: Total Cycles:      14613
+# CHECK-NEXT: Total uOps:        320
+
+# CHECK:      Dispatch Width:    4
+# CHECK-NEXT: uOps Per Cycle:    0.02
+# CHECK-NEXT: IPC:               0.02
+# CHECK-NEXT: Block RThroughput: 14361.0
+
+# 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     1.00                  U     vsetvli	zero, zero, e8, mf8, tu, mu
+# CHECK-NEXT:  1      51    51.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, mf4, tu, mu
+# CHECK-NEXT:  1      51    51.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, mf2, tu, mu
+# CHECK-NEXT:  1      51    51.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, m1, tu, mu
+# CHECK-NEXT:  1      51    51.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, m1, tu, mu
+# CHECK-NEXT:  1      51    51.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, m2, tu, mu
+# CHECK-NEXT:  1      102   102.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, m4, tu, mu
+# CHECK-NEXT:  1      204   204.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, m8, tu, mu
+# CHECK-NEXT:  1      408   408.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, mf8, tu, mu
+# CHECK-NEXT:  1      408   408.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, mf4, tu, mu
+# CHECK-NEXT:  1      45    45.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, mf2, tu, mu
+# CHECK-NEXT:  1      45    45.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, m1, tu, mu
+# CHECK-NEXT:  1      45    45.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, m1, tu, mu
+# CHECK-NEXT:  1      45    45.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, m2, tu, mu
+# CHECK-NEXT:  1      90    90.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, m4, tu, mu
+# CHECK-NEXT:  1      180   180.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, m8, tu, mu
+# CHECK-NEXT:  1      360   360.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e32, mf8, tu, mu
+# CHECK-NEXT:  1      408   408.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e32, mf4, tu, mu
+# CHECK-NEXT:  1      408   408.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e32, mf2, tu, mu
+# CHECK-NEXT:  1      42    42.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e32, m1, tu, mu
+# CHECK-NEXT:  1      42    42.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e32, m1, tu, mu
+# CHECK-NEXT:  1      42    42.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e32, m2, tu, mu
+# CHECK-NEXT:  1      84    84.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e32, m4, tu, mu
+# CHECK-NEXT:  1      168   168.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e32, m8, tu, mu
+# CHECK-NEXT:  1      336   336.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e64, mf8, tu, mu
+# CHECK-NEXT:  1      408   408.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e64, mf4, tu, mu
+# CHECK-NEXT:  1      408   408.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e64, mf2, tu, mu
+# CHECK-NEXT:  1      408   408.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e64, m1, tu, mu
+# CHECK-NEXT:  1      72    72.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e64, m1, tu, mu
+# CHECK-NEXT:  1      72    72.00                       vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e64, m2, tu, mu
+# CHECK-NEXT:  1      144   144.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e64, m4, tu, mu
+# CHECK-NEXT:  1      288   288.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e64, m8, tu, mu
+# CHECK-NEXT:  1      576   576.00                      vdiv.vv	v8, v16, v24
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, mf8, tu, mu
+# CHECK-NEXT:  1      51    51.00                       vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, mf4, tu, mu
+# CHECK-NEXT:  1      51    51.00                       vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, mf2, tu, mu
+# CHECK-NEXT:  1      51    51.00                       vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, m1, tu, mu
+# CHECK-NEXT:  1      51    51.00                       vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, m1, tu, mu
+# CHECK-NEXT:  1      51    51.00                       vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, m2, tu, mu
+# CHECK-NEXT:  1      102   102.00                      vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, m4, tu, mu
+# CHECK-NEXT:  1      204   204.00                      vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e8, m8, tu, mu
+# CHECK-NEXT:  1      408   408.00                      vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, mf8, tu, mu
+# CHECK-NEXT:  1      408   408.00                      vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, mf4, tu, mu
+# CHECK-NEXT:  1      45    45.00                       vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, mf2, tu, mu
+# CHECK-NEXT:  1      45    45.00                       vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, m1, tu, mu
+# CHECK-NEXT:  1      45    45.00                       vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, m1, tu, mu
+# CHECK-NEXT:  1      45    45.00                       vdiv.vx	v8, v16, a0
+# CHECK-NEXT:  1      1     1.00                  U     vsetvli	zero, zero, e16, m2, tu, mu
+# CHECK-NEXT:  1      90    90.00                  ...
[truncated]

@mshockwave mshockwave changed the title [RISCV] Refine division latencies in SiFive P600's scheduling model [RISCV] Refine vector division latencies in SiFive P600's scheduling model Nov 5, 2024
wangpc-pp
wangpc-pp approved these changes Nov 6, 2024
View reviewed changes
Copy link
Contributor

@wangpc-pp wangpc-pp left a comment

Choose a reason for hiding this comment

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

LGTM.

@mshockwave mshockwave merged commit 7ef7c0d into llvm:main Nov 6, 2024
8 of 10 checks passed
@mshockwave mshockwave deleted the patch/riscv/sifive-p600-div branch November 6, 2024 22:53
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@wangpc-pp wangpc-pp wangpc-pp approved these changes

@michaelmaitland michaelmaitland Awaiting requested review from michaelmaitland

@dtcxzyw dtcxzyw Awaiting requested review from dtcxzyw

@topperc topperc Awaiting requested review from topperc

Assignees
No one assigned
Labels
backend:RISC-V
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@mshockwave @llvmbot @wangpc-pp