[RISCV] Add sched model for XiangShan-NanHu

llvm/lib/Target/RISCV/RISCV.td

@@ -44,6 +44,7 @@ include "RISCVSchedRocket.td"
 include "RISCVSchedSiFive7.td"
 include "RISCVSchedSiFiveP400.td"
 include "RISCVSchedSyntacoreSCR1.td"
+include "RISCVSchedXiangShanNanHu.td"
 
 //===----------------------------------------------------------------------===//
 // RISC-V processors supported.

llvm/lib/Target/RISCV/RISCVProcessors.td

@@ -330,7 +330,7 @@ def VENTANA_VEYRON_V1 : RISCVProcessorModel<"veyron-v1",
                                               TuneLDADDFusion]>;
 
 def XIANGSHAN_NANHU : RISCVProcessorModel<"xiangshan-nanhu",
-                                          NoSchedModel,
+                                          XiangShanNanHuModel,
                                           [Feature64Bit,
                                            FeatureStdExtZicsr,
                                            FeatureStdExtZifencei,
@@ -348,4 +348,8 @@ def XIANGSHAN_NANHU : RISCVProcessorModel<"xiangshan-nanhu",
                                            FeatureStdExtZksh,
                                            FeatureStdExtSvinval,
                                            FeatureStdExtZicbom,
-                                           FeatureStdExtZicboz]>;
+                                           FeatureStdExtZicboz],
+                                           [TuneNoDefaultUnroll,
+                                            TuneZExtHFusion,
+                                            TuneZExtWFusion,
+                                            TuneShiftedZExtWFusion]>;

llvm/lib/Target/RISCV/RISCVSchedXiangShanNanHu.td

@@ -0,0 +1,308 @@
+//==- RISCVSchedXiangShanNanHu.td - XS-NanHu Scheduling Defs -*- tablegen -*-=//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+//===----------------------------------------------------------------------===//
+
+// XiangShan is a high-performance open-source RISC-V processor developed by
+// the Institute of Computing Technology (ICT), Chinese Academy of Sciences.
+// Source: https://github.com/OpenXiangShan/XiangShan
+// Documentation: https://github.com/OpenXiangShan/XiangShan-doc
+
+// XiangShan-NanHu is the second generation of XiangShan processor series.
+// Overview: https://xiangshan-doc.readthedocs.io/zh-cn/latest/integration/overview/
+
+def XiangShanNanHuModel : SchedMachineModel {
+  let MicroOpBufferSize = 256;
+  let LoopMicroOpBufferSize = 48;  // Instruction queue size
+  let IssueWidth = 6;  // 6-way decode and dispatch
+  let LoadLatency = 4;
+  let MispredictPenalty = 11; // Based on estimate of pipeline depth.
+  let CompleteModel = 0;
+  let UnsupportedFeatures = [HasStdExtZcmt, HasStdExtZkr, HasVInstructions,
+                             HasVInstructionsI64];
+}
+
+let SchedModel = XiangShanNanHuModel in {
+
+// The reservation stations are distributed and grouped as 32-entry or 16-entry smaller ones.
+let BufferSize = 16 in {
+  def XS2ALU : ProcResource<4>;
+  def XS2MDU : ProcResource<2>;
+  def XS2MISC : ProcResource<1>;
+
+  def XS2FMAC : ProcResource<4>;
+  def XS2FMISC : ProcResource<2>;
+
+  // Load/Store queues are ignored.
+  def XS2LD : ProcResource<2>;
+  def XS2ST : ProcResource<2>;
+}
+
+// Branching
+def : WriteRes<WriteJmp, [XS2MISC]>;
+def : WriteRes<WriteJal, [XS2MISC]>;
+def : WriteRes<WriteJalr, [XS2MISC]>;
+
+// Integer arithmetic and logic
+let Latency = 1 in {
+def : WriteRes<WriteIALU, [XS2ALU]>;
+def : WriteRes<WriteIALU32, [XS2ALU]>;
+def : WriteRes<WriteShiftImm, [XS2ALU]>;
+def : WriteRes<WriteShiftImm32, [XS2ALU]>;
+def : WriteRes<WriteShiftReg, [XS2ALU]>;
+def : WriteRes<WriteShiftReg32, [XS2ALU]>;
+}
+
+// Integer multiplication
+let Latency = 3 in {
+def : WriteRes<WriteIMul, [XS2MDU]>;
+def : WriteRes<WriteIMul32, [XS2MDU]>;
+}
+
+// Integer division
+// SRT16 algorithm
+let Latency = 20, ReleaseAtCycles = [20] in {
+def : WriteRes<WriteIDiv32, [XS2MDU]>;
+def : WriteRes<WriteIDiv, [XS2MDU]>;
+}
+
+// Zb*
+let Latency = 1 in {
+// Zba
+def : WriteRes<WriteSHXADD, [XS2ALU]>;
+def : WriteRes<WriteSHXADD32, [XS2ALU]>;
+
+// Zbb
+def : WriteRes<WriteRotateImm, [XS2ALU]>;
+def : WriteRes<WriteRotateImm32, [XS2ALU]>;
+def : WriteRes<WriteRotateReg, [XS2ALU]>;
+def : WriteRes<WriteRotateReg32, [XS2ALU]>;
+def : WriteRes<WriteORCB, [XS2ALU]>;
+def : WriteRes<WriteREV8, [XS2ALU]>;
+
+// Zbkb
+def : WriteRes<WriteBREV8, [XS2ALU]>;
+def : WriteRes<WritePACK, [XS2ALU]>;
+def : WriteRes<WritePACK32, [XS2ALU]>;
+def : WriteRes<WriteZIP, [XS2ALU]>;
+
+// Zbs
+def : WriteRes<WriteSingleBit, [XS2ALU]>;
+def : WriteRes<WriteSingleBitImm, [XS2ALU]>;
+def : WriteRes<WriteBEXT, [XS2ALU]>;
+def : WriteRes<WriteBEXTI, [XS2ALU]>;
+}
+
+let Latency = 3 in {
+// Zbb
+def : WriteRes<WriteCLZ, [XS2MDU]>;
+def : WriteRes<WriteCLZ32, [XS2MDU]>;
+def : WriteRes<WriteCTZ, [XS2MDU]>;
+def : WriteRes<WriteCTZ32, [XS2MDU]>;
+def : WriteRes<WriteCPOP, [XS2MDU]>;
+def : WriteRes<WriteCPOP32, [XS2MDU]>;
+
+// Zbkc
+def : WriteRes<WriteCLMUL, [XS2MDU]>;
+
+// Zbkx
+def : WriteRes<WriteXPERM, [XS2MDU]>;
+}
+
+// Memory
+def : WriteRes<WriteSTB, [XS2ST]>;
+def : WriteRes<WriteSTH, [XS2ST]>;
+def : WriteRes<WriteSTW, [XS2ST]>;
+def : WriteRes<WriteSTD, [XS2ST]>;
+def : WriteRes<WriteFST32, [XS2ST]>;
+def : WriteRes<WriteFST64, [XS2ST]>;
+def : WriteRes<WriteAtomicSTW, [XS2ST]>;
+def : WriteRes<WriteAtomicSTD, [XS2ST]>;
+
+let Latency = 5 in {
+def : WriteRes<WriteLDB, [XS2LD]>;
+def : WriteRes<WriteLDH, [XS2LD]>;
+def : WriteRes<WriteLDW, [XS2LD]>;
+def : WriteRes<WriteLDD, [XS2LD]>;
+
+def : WriteRes<WriteAtomicW, [XS2LD]>;
+def : WriteRes<WriteAtomicD, [XS2LD]>;
+def : WriteRes<WriteAtomicLDW, [XS2LD]>;
+def : WriteRes<WriteAtomicLDD, [XS2LD]>;
+
+def : WriteRes<WriteFLD32, [XS2LD]>;
+def : WriteRes<WriteFLD64, [XS2LD]>;
+}
+
+// XiangShan-NanHu uses FuDian FPU instead of Berkeley HardFloat.
+// Documentation: https://github.com/OpenXiangShan/fudian
+
+let Latency = 3 in {
+def : WriteRes<WriteFAdd32, [XS2FMAC]>;
+def : WriteRes<WriteFSGNJ32, [XS2FMAC]>;
+def : WriteRes<WriteFMinMax32, [XS2FMAC]>;
+def : WriteRes<WriteFAdd64, [XS2FMAC]>;
+def : WriteRes<WriteFSGNJ64, [XS2FMAC]>;
+def : WriteRes<WriteFMinMax64, [XS2FMAC]>;
+
+def : WriteRes<WriteFCvtI32ToF32, [XS2FMAC]>;
+def : WriteRes<WriteFCvtI32ToF64, [XS2FMAC]>;
+def : WriteRes<WriteFCvtI64ToF32, [XS2FMAC]>;
+def : WriteRes<WriteFCvtI64ToF64, [XS2FMAC]>;
+def : WriteRes<WriteFCvtF32ToI32, [XS2FMAC]>;
+def : WriteRes<WriteFCvtF32ToI64, [XS2FMAC]>;
+def : WriteRes<WriteFCvtF64ToI32, [XS2FMAC]>;
+def : WriteRes<WriteFCvtF64ToI64, [XS2FMAC]>;
+def : WriteRes<WriteFCvtF32ToF64, [XS2FMAC]>;
+def : WriteRes<WriteFCvtF64ToF32, [XS2FMAC]>;
+
+def : WriteRes<WriteFClass32, [XS2FMAC]>;
+def : WriteRes<WriteFClass64, [XS2FMAC]>;
+def : WriteRes<WriteFCmp32, [XS2FMAC]>;
+def : WriteRes<WriteFCmp64, [XS2FMAC]>;
+def : WriteRes<WriteFMovF32ToI32, [XS2FMAC]>;
+def : WriteRes<WriteFMovI32ToF32, [XS2FMAC]>;
+def : WriteRes<WriteFMovF64ToI64, [XS2FMAC]>;
+def : WriteRes<WriteFMovI64ToF64, [XS2FMAC]>;
+}
+
+// FP multiplication
+let Latency = 3 in {
+def : WriteRes<WriteFMul32, [XS2FMAC]>;
+def : WriteRes<WriteFMul64, [XS2FMAC]>;
+}
+
+let Latency = 5 in {
+def : WriteRes<WriteFMA32, [XS2FMAC]>;
+def : WriteRes<WriteFMA64, [XS2FMAC]>;
+}
+
+// FP division
+def : WriteRes<WriteFDiv32, [XS2FMISC]> {
+    let Latency = 11;
+}
+def : WriteRes<WriteFDiv64, [XS2FMISC]> {
+    let Latency = 18;
+}
+
+def : WriteRes<WriteFSqrt32, [XS2FMISC]> {
+    let Latency = 17;
+}
+def : WriteRes<WriteFSqrt64, [XS2FMISC]> {
+    let Latency = 31;
+}
+
+// Others
+def : WriteRes<WriteCSR, [XS2MISC]>;
+def : WriteRes<WriteNop, []>;
+
+def : InstRW<[WriteIALU], (instrs COPY)>;
+
+// Bypass and advance
+
+class XS2LoadToALUBypass<SchedRead read>
+    : ReadAdvance<read, 1, [WriteLDB, WriteLDH, WriteLDW, WriteLDD, WriteAtomicW, WriteAtomicD, WriteAtomicLDW, WriteAtomicLDD]>;
+
+def : ReadAdvance<ReadJmp, 0>;
+def : ReadAdvance<ReadJalr, 0>;
+def : ReadAdvance<ReadCSR, 0>;
+def : ReadAdvance<ReadStoreData, 0>;
+def : ReadAdvance<ReadMemBase, 0>;
+def : XS2LoadToALUBypass<ReadIALU>;
+def : XS2LoadToALUBypass<ReadIALU32>;
+def : XS2LoadToALUBypass<ReadShiftImm>;
+def : XS2LoadToALUBypass<ReadShiftImm32>;
+def : XS2LoadToALUBypass<ReadShiftReg>;
+def : XS2LoadToALUBypass<ReadShiftReg32>;
+def : ReadAdvance<ReadIDiv, 0>;
+def : ReadAdvance<ReadIDiv32, 0>;
+def : ReadAdvance<ReadIMul, 0>;
+def : ReadAdvance<ReadIMul32, 0>;
+def : ReadAdvance<ReadAtomicWA, 0>;
+def : ReadAdvance<ReadAtomicWD, 0>;
+def : ReadAdvance<ReadAtomicDA, 0>;
+def : ReadAdvance<ReadAtomicDD, 0>;
+def : ReadAdvance<ReadAtomicLDW, 0>;
+def : ReadAdvance<ReadAtomicLDD, 0>;
+def : ReadAdvance<ReadAtomicSTW, 0>;
+def : ReadAdvance<ReadAtomicSTD, 0>;
+def : ReadAdvance<ReadFStoreData, 0>;
+def : ReadAdvance<ReadFMemBase, 0>;
+def : ReadAdvance<ReadFAdd32, 0>;
+def : ReadAdvance<ReadFAdd64, 0>;
+def : ReadAdvance<ReadFMul32, 0>;
+def : ReadAdvance<ReadFMul64, 0>;
+def : ReadAdvance<ReadFMA32, 0>;
+def : ReadAdvance<ReadFMA32Addend, 2>; // Cascade FMA
+def : ReadAdvance<ReadFMA64, 0>;
+def : ReadAdvance<ReadFMA64Addend, 2>; // Cascade FMA
+def : ReadAdvance<ReadFDiv32, 0>;
+def : ReadAdvance<ReadFDiv64, 0>;
+def : ReadAdvance<ReadFSqrt32, 0>;
+def : ReadAdvance<ReadFSqrt64, 0>;
+def : ReadAdvance<ReadFCmp32, 0>;
+def : ReadAdvance<ReadFCmp64, 0>;
+def : ReadAdvance<ReadFSGNJ32, 0>;
+def : ReadAdvance<ReadFSGNJ64, 0>;
+def : ReadAdvance<ReadFMinMax32, 0>;
+def : ReadAdvance<ReadFMinMax64, 0>;
+def : ReadAdvance<ReadFCvtF32ToI32, 0>;
+def : ReadAdvance<ReadFCvtF32ToI64, 0>;
+def : ReadAdvance<ReadFCvtF64ToI32, 0>;
+def : ReadAdvance<ReadFCvtF64ToI64, 0>;
+def : ReadAdvance<ReadFCvtI32ToF32, 0>;
+def : ReadAdvance<ReadFCvtI32ToF64, 0>;
+def : ReadAdvance<ReadFCvtI64ToF32, 0>;
+def : ReadAdvance<ReadFCvtI64ToF64, 0>;
+def : ReadAdvance<ReadFCvtF32ToF64, 0>;
+def : ReadAdvance<ReadFCvtF64ToF32, 0>;
+def : ReadAdvance<ReadFMovF32ToI32, 0>;
+def : ReadAdvance<ReadFMovI32ToF32, 0>;
+def : ReadAdvance<ReadFMovF64ToI64, 0>;
+def : ReadAdvance<ReadFMovI64ToF64, 0>;
+def : ReadAdvance<ReadFClass32, 0>;
+def : ReadAdvance<ReadFClass64, 0>;
+
+// Zb*
+// Zba
+def : XS2LoadToALUBypass<ReadSHXADD>;
+def : XS2LoadToALUBypass<ReadSHXADD32>;
+// Zbb
+def : XS2LoadToALUBypass<ReadRotateImm>;
+def : XS2LoadToALUBypass<ReadRotateImm32>;
+def : XS2LoadToALUBypass<ReadRotateReg>;
+def : XS2LoadToALUBypass<ReadRotateReg32>;
+def : ReadAdvance<ReadCLZ, 0>;
+def : ReadAdvance<ReadCLZ32, 0>;
+def : ReadAdvance<ReadCTZ, 0>;
+def : ReadAdvance<ReadCTZ32, 0>;
+def : ReadAdvance<ReadCPOP, 0>;
+def : ReadAdvance<ReadCPOP32, 0>;
+def : XS2LoadToALUBypass<ReadORCB>;
+def : XS2LoadToALUBypass<ReadREV8>;
+// Zbkc
+def : ReadAdvance<ReadCLMUL, 0>;
+// Zbs
+def : XS2LoadToALUBypass<ReadSingleBit>;
+def : XS2LoadToALUBypass<ReadSingleBitImm>;
+// Zbkb
+def : XS2LoadToALUBypass<ReadBREV8>;
+def : XS2LoadToALUBypass<ReadPACK>;
+def : XS2LoadToALUBypass<ReadPACK32>;
+def : XS2LoadToALUBypass<ReadZIP>;
+// Zbkx
+def : ReadAdvance<ReadXPERM, 0>;
+
+//===----------------------------------------------------------------------===//
+// Unsupported extensions
+defm : UnsupportedSchedV;
+defm : UnsupportedSchedZfa;
+defm : UnsupportedSchedZfh;
+defm : UnsupportedSchedSFB;
+defm : UnsupportedSchedZabha;
+}

llvm/test/tools/llvm-mca/RISCV/XiangShan/cascade-fma.s

@@ -0,0 +1,53 @@
+# NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
+# RUN: llvm-mca -mtriple=riscv64 -mcpu=xiangshan-nanhu < %s | FileCheck %s
+
+# Test XiangShan FuDian's cascade FMA, CPI = 3
+fmadd.s fa0, fa1, fa2, fa0
+
+# CHECK:      Iterations:        100
+# CHECK-NEXT: Instructions:      100
+# CHECK-NEXT: Total Cycles:      305
+# CHECK-NEXT: Total uOps:        100
+
+# CHECK:      Dispatch Width:    6
+# CHECK-NEXT: uOps Per Cycle:    0.33
+# CHECK-NEXT: IPC:               0.33
+# CHECK-NEXT: Block RThroughput: 0.3
+
+# 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      5     0.25                        fmadd.s	fa0, fa1, fa2, fa0
+
+# CHECK:      Resources:
+# CHECK-NEXT: [0.0] - XS2ALU
+# CHECK-NEXT: [0.1] - XS2ALU
+# CHECK-NEXT: [0.2] - XS2ALU
+# CHECK-NEXT: [0.3] - XS2ALU
+# CHECK-NEXT: [1.0] - XS2FMAC
+# CHECK-NEXT: [1.1] - XS2FMAC
+# CHECK-NEXT: [1.2] - XS2FMAC
+# CHECK-NEXT: [1.3] - XS2FMAC
+# CHECK-NEXT: [2.0] - XS2FMISC
+# CHECK-NEXT: [2.1] - XS2FMISC
+# CHECK-NEXT: [3.0] - XS2LD
+# CHECK-NEXT: [3.1] - XS2LD
+# CHECK-NEXT: [4.0] - XS2MDU
+# CHECK-NEXT: [4.1] - XS2MDU
+# CHECK-NEXT: [5]   - XS2MISC
+# CHECK-NEXT: [6.0] - XS2ST
+# CHECK-NEXT: [6.1] - XS2ST
+
+# CHECK:      Resource pressure per iteration:
+# CHECK-NEXT: [0.0]  [0.1]  [0.2]  [0.3]  [1.0]  [1.1]  [1.2]  [1.3]  [2.0]  [2.1]  [3.0]  [3.1]  [4.0]  [4.1]  [5]    [6.0]  [6.1]
+# CHECK-NEXT:  -      -      -      -     0.25   0.25   0.25   0.25    -      -      -      -      -      -      -      -      -
+
+# CHECK:      Resource pressure by instruction:
+# CHECK-NEXT: [0.0]  [0.1]  [0.2]  [0.3]  [1.0]  [1.1]  [1.2]  [1.3]  [2.0]  [2.1]  [3.0]  [3.1]  [4.0]  [4.1]  [5]    [6.0]  [6.1]  Instructions:
+# CHECK-NEXT:  -      -      -      -     0.25   0.25   0.25   0.25    -      -      -      -      -      -      -      -      -     fmadd.s	fa0, fa1, fa2, fa0