[PATCH] D20839: AMDGPU: Add amdgpu-ps-wqm-outputs function attributes

[Nicolai Hähnle via llvm-commits]

nhaehnle created this revision.
nhaehnle added reviewers: arsenm, tstellarAMD, mareko.
nhaehnle added a subscriber: llvm-commits.
Herald added subscribers: kzhuravl, arsenm.

The presence of this attribute indicates that VGPR outputs should be computed
in whole quad mode. This will be used by Mesa for prolog pixel shaders, so
that derivatives can be taken of shader inputs computed by the prolog, fixing
a bug.

The generated code could certainly be improved: if a prolog pixel shader is
used (which isn't common in modern OpenGL - they're used for gl_Color, polygon
stipples, and forcing per-sample interpolation), Mesa will use this attribute
unconditionally, because it has to be conservative. So WQM may be used in the
prolog when it isn't really needed, and furthermore a silly back-and-forth
switch is likely to happen at the boundary between prolog and main shader
parts.

Fixing this is a bit involved: we'd first have to add a mechanism by which
LLVM writes the WQM-related input requirements to the main shader part binary,
and then Mesa specializes the prolog part accordingly. At that point, we may
as well just compile a monolithic shader...

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=95130

http://reviews.llvm.org/D20839

Files:
  lib/Target/AMDGPU/SIWholeQuadMode.cpp
  test/CodeGen/AMDGPU/wqm.ll

Index: test/CodeGen/AMDGPU/wqm.ll
===================================================================
--- test/CodeGen/AMDGPU/wqm.ll
+++ test/CodeGen/AMDGPU/wqm.ll
@@ -332,6 +332,19 @@
   ret <4 x float> %tex
 }
 
+; Check prolog shaders.
+;
+; CHECK-LABEL: {{^}}test_prolog_1:
+; CHECK: s_mov_b64 [[ORIG:s\[[0-9]+:[0-9]+\]]], exec
+; CHECK: s_wqm_b64 exec, exec
+; CHECK: v_add_f32_e32 v0,
+; CHECK: s_and_b64 exec, exec, [[ORIG]]
+define amdgpu_ps float @test_prolog_1(float %a, float %b) #4 {
+main_body:
+  %s = fadd float %a, %b
+  ret float %s
+}
+
 declare void @llvm.amdgcn.image.store.v4i32(<4 x float>, <4 x i32>, <8 x i32>, i32, i1, i1, i1, i1) #1
 
 declare <4 x float> @llvm.amdgcn.image.load.v4i32(<4 x i32>, <8 x i32>, i32, i1, i1, i1, i1) #2
@@ -345,3 +358,4 @@
 attributes #1 = { nounwind }
 attributes #2 = { nounwind readonly }
 attributes #3 = { nounwind readnone }
+attributes #4 = { "amdgpu-ps-wqm-outputs" }
Index: lib/Target/AMDGPU/SIWholeQuadMode.cpp
===================================================================
--- lib/Target/AMDGPU/SIWholeQuadMode.cpp
+++ lib/Target/AMDGPU/SIWholeQuadMode.cpp
@@ -154,6 +154,7 @@
 char SIWholeQuadMode::scanInstructions(MachineFunction &MF,
                                        std::vector<WorkItem> &Worklist) {
   char GlobalFlags = 0;
+  bool WQMOutputs = MF.getFunction()->hasFnAttribute("amdgpu-ps-wqm-outputs");
 
   for (auto BI = MF.begin(), BE = MF.end(); BI != BE; ++BI) {
     MachineBasicBlock &MBB = *BI;
@@ -168,6 +169,11 @@
       } else if (TII->get(Opcode).mayStore() &&
                  (MI.getDesc().TSFlags & SIInstrFlags::VM_CNT)) {
         Flags = StateExact;
+      } else if (WQMOutputs && Opcode == AMDGPU::COPY) {
+        if (TRI->hasVGPRs(TII->getOpRegClass(MI, 0)))
+          Flags = StateWQM;
+        else
+          continue;
       } else {
         // Handle export instructions with the exec mask valid flag set
         if (Opcode == AMDGPU::EXP) {
@@ -184,6 +190,13 @@
       Worklist.push_back(&MI);
       GlobalFlags |= Flags;
     }
+
+    if (WQMOutputs && MBB.succ_empty()) {
+      // This is a prolog shader. Make sure we go back to exact mode at the end.
+      Blocks[&MBB].OutNeeds = StateExact;
+      Worklist.push_back(&MBB);
+      GlobalFlags |= StateExact;
+    }
   }
 
   return GlobalFlags;


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D20839.59138.patch
Type: text/x-patch
Size: 2322 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160531/57c9064f/attachment.bin>