Support derivative functions in compute & capabilities adjustments (#4014)

* Support derivative functions in compute & capabilities adjustments

fixes #4000

PR implements derivative functions in compute shaders properly so we have the functionality for SPIR-V & GLSL. Tests reflect fragment and compute paths.

PR also adjusts capabilities to correct wrong SPRI-V target capabilities for when using textures.

Remarks:
1. __requireComputeDerivative(); is a intrinsic_op and not modifier since inlining will destroy the modifier.
2. Derivative mode is tied to an entry point decoration `[DerivativeGroupQuad]`/`[DerivativeGroupLinear]` or GLSL syntax ``derivative_group_linearNV`. Default is to set the mode to `[DerivativeGroupQuad]`

* remove -emit-spirv-directly

* fixes

1. fix minor issue fwidth change where I returned the wrong type
2. fix issue where glslang{glsl->spirv} is wrong, so we don't run that test and just run the glsl test & direct spir-v test for intrinsic-texture.slang

* adjust as per review and refine code

1. add test to ensure multi-diverging-in-logic entry points work -- 2 functions which may cause computeDerivatives + 1 that uses, 1 that does not.
2. naming
3. use entry point ref graph for c-like-targets
4. reordered some code to util's and removed `static linline` since that was just for ease of coding on my end (should not have been pushed).

* Grammer

* split up source file + issolate GLSL emit path change.

---------

Co-authored-by: Yong He <yonghe@outlook.com>

1 files changed, 69 insertions, 14 deletions

diff --git a/source/slang/slang-lower-to-ir.cpp b/source/slang/slang-lower-to-ir.cpp
index a600963a7..e4fc33e33 100644
--- a/source/slang/slang-lower-to-ir.cpp
+++ b/source/slang/slang-lower-to-ir.cpp
@@ -12,6 +12,7 @@
 #include "slang-ir-bit-field-accessors.h"
 #include "slang-ir-loop-inversion.h"
 #include "slang-ir.h"
+#include "slang-ir-util.h"
 #include "slang-ir-constexpr.h"
 #include "slang-ir-dce.h"
 #include "slang-ir-diff-call.h"
@@ -7264,26 +7265,61 @@ struct DeclLoweringVisitor : DeclVisitor<DeclLoweringVisitor, LoweredValInfo>
 
 #undef IGNORED_CASE
 
+    void getAllEntryPointsNoOverride(List<IRInst*>& entryPoints)
+    {
+        if(entryPoints.getCount() != 0 )
+            return;
+        for(const auto d : context->irBuilder->getModule()->getModuleInst()->getGlobalInsts())
+            if(d->findDecoration<IREntryPointDecoration>())
+                entryPoints.add(d);
+    }
+
     LoweredValInfo visitEmptyDecl(EmptyDecl* decl)
     {
+        bool verifyComputeDerivativeGroupModifier = false;
+        List<IRInst*> entryPoints {};
         for(const auto modifier : decl->modifiers)
         {
             if(const auto layoutLocalSizeAttr = as<GLSLLayoutLocalSizeAttribute>(modifier))
             {
-                for(const auto d : context->irBuilder->getModule()->getModuleInst()->getGlobalInsts())
-                {
-                    if(d->findDecoration<IREntryPointDecoration>())
-                    {
-                        getBuilder()->addNumThreadsDecoration(
+                verifyComputeDerivativeGroupModifier = true;
+                getAllEntryPointsNoOverride(entryPoints);
+                for(auto d : entryPoints)
+                    as<IRNumThreadsDecoration>(getBuilder()->addNumThreadsDecoration(
                             d,
                             getSimpleVal(context, lowerVal(context, layoutLocalSizeAttr->x)),
                             getSimpleVal(context, lowerVal(context, layoutLocalSizeAttr->y)),
                             getSimpleVal(context, lowerVal(context, layoutLocalSizeAttr->z))
-                        );
-                    }
-                }
+                        ));
+            }
+            else if(as<GLSLLayoutDerivativeGroupQuadAttribute>(modifier))
+            {
+                verifyComputeDerivativeGroupModifier = true;
+                getAllEntryPointsNoOverride(entryPoints);
+                for(auto d : entryPoints)
+                    getBuilder()->addSimpleDecoration<IRDerivativeGroupQuadDecoration>(d);
+            }
+            else if(as<GLSLLayoutDerivativeGroupLinearAttribute>(modifier))
+            {
+                verifyComputeDerivativeGroupModifier = true;
+                getAllEntryPointsNoOverride(entryPoints);
+                for(auto d : entryPoints)
+                    getBuilder()->addSimpleDecoration<IRDerivativeGroupLinearDecoration>(d);
             }
         }
+        
+        if(!verifyComputeDerivativeGroupModifier)
+            return LoweredValInfo();
+        for(auto d : entryPoints)
+        {
+            verifyComputeDerivativeGroupModifiers(
+                getSink(),
+                decl->loc,
+                d->findDecoration<IRDerivativeGroupQuadDecoration>(), 
+                d->findDecoration<IRDerivativeGroupLinearDecoration>(),
+                d->findDecoration<IRNumThreadsDecoration>());
+        }
+
         return LoweredValInfo();
     }
 
@@ -9677,6 +9713,9 @@ struct DeclLoweringVisitor : DeclVisitor<DeclLoweringVisitor, LoweredValInfo>
 
         addBitFieldAccessorDecorations(irFunc, decl);
 
+        IRNumThreadsDecoration* numThreadsDecor = nullptr;
+        IRDecoration* derivativeGroupQuadDecor = nullptr;
+        IRDecoration* derivativeGroupLinearDecor = nullptr;
         for (auto modifier : decl->modifiers)
         {
             if (as<RequiresNVAPIAttribute>(modifier))
@@ -9691,6 +9730,14 @@ struct DeclLoweringVisitor : DeclVisitor<DeclLoweringVisitor, LoweredValInfo>
             {
                 getBuilder()->addSimpleDecoration<IRNoInlineDecoration>(irFunc);
             }
+            else if (auto derivativeGroupQuadMod = as<DerivativeGroupQuadAttribute>(modifier))
+            {
+                derivativeGroupQuadDecor = getBuilder()->addSimpleDecoration<IRDerivativeGroupQuadDecoration>(irFunc);
+            }
+            else if (auto derivativeGroupLinearMod = as<DerivativeGroupLinearAttribute>(modifier))
+            {
+                derivativeGroupLinearDecor = getBuilder()->addSimpleDecoration<IRDerivativeGroupLinearDecoration>(irFunc);
+            }
             else if (auto instanceAttr = as<InstanceAttribute>(modifier))
             {
                 IRIntLit* intLit = _getIntLitFromAttribute(getBuilder(), instanceAttr);
@@ -9703,12 +9750,13 @@ struct DeclLoweringVisitor : DeclVisitor<DeclLoweringVisitor, LoweredValInfo>
             }
             else if (auto numThreadsAttr = as<NumThreadsAttribute>(modifier))
             {
-                getBuilder()->addNumThreadsDecoration(
-                    irFunc,
-                    getSimpleVal(context, lowerVal(context, numThreadsAttr->x)),
-                    getSimpleVal(context, lowerVal(context, numThreadsAttr->y)),
-                    getSimpleVal(context, lowerVal(context, numThreadsAttr->z))
-                );
+                numThreadsDecor = as<IRNumThreadsDecoration>(
+                    getBuilder()->addNumThreadsDecoration(
+                        irFunc,
+                        getSimpleVal(context, lowerVal(context, numThreadsAttr->x)),
+                        getSimpleVal(context, lowerVal(context, numThreadsAttr->y)),
+                        getSimpleVal(context, lowerVal(context, numThreadsAttr->z))
+                    ));
             }
             else if (auto waveSizeAttr = as<WaveSizeAttribute>(modifier))
             {
@@ -9862,6 +9910,13 @@ struct DeclLoweringVisitor : DeclVisitor<DeclLoweringVisitor, LoweredValInfo>
                 getBuilder()->addDecoration(irFunc, kIROp_NonDynamicUniformReturnDecoration);
         }
 
+        verifyComputeDerivativeGroupModifiers(
+            getSink(),
+            decl->loc,
+            derivativeGroupQuadDecor,
+            derivativeGroupLinearDecor,
+            numThreadsDecor);
+
         if (!isInline)
         {
             // If there are any constant expr rate parameters, we should inline this function.