Generalize precompilation support (#534)

* Generalize precompilation support

* CUDA precompilation

* Update ReactantCUDAExt.jl

* Update ReactantCUDAExt.jl

* fixup

* fix

* Update ReactantCUDAExt.jl

* Apply suggestions from code review

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

---------

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

Author: wsmoses

ext/ReactantCUDAExt.jl

@@ -328,7 +328,9 @@ function compile(job)
             # :llvm, job; optimize=false, cleanup=false, validate=false, libraries=false
         )
 
-        GPUCompiler.link_library!(mod, GPUCompiler.load_runtime(job))
+        if !Reactant.precompiling()
+            GPUCompiler.link_library!(mod, GPUCompiler.load_runtime(job))
+        end
         entryname = LLVM.name(meta.entry)
 
         GPUCompiler.optimize_module!(job, mod)
@@ -788,4 +790,31 @@ function __init__()
     return nothing
 end
 
+@static if !Sys.isapple() && Sys.ARCH != :aarch64
+    Reactant.PrecompileTools.@setup_workload begin
+        Reactant.initialize_dialect()
+        client = Reactant.XLA.CPUClient(; checkcount=false)
+        Reactant.PrecompileTools.@compile_workload begin
+            @static if Reactant.precompilation_supported()
+                function square_kernel!(x)
+                    i = CUDA.threadIdx().x
+                    x[i] *= x[i]
+                    return nothing
+                end
+
+                function square!(x)
+                    CUDA.@cuda blocks = 1 threads = length(x) square_kernel!(x)
+                    return nothing
+                end
+                y = Reactant.ConcreteRArray([2.0]; client)
+                Reactant.Compiler.compile_mlir(square!, (y,); optimize=false)
+            end
+        end
+        Reactant.XLA.free_client(client)
+        client.client = C_NULL
+        Reactant.deinitialize_dialect()
+        Reactant.clear_oc_cache()
+    end
+end
+
 end # module ReactantCUDAExt

src/Precompile.jl

@@ -1,3 +1,4 @@
+using PrecompileTools
 using PrecompileTools: @setup_workload, @compile_workload
 
 function infer_sig(sig)
@@ -34,15 +35,33 @@ function infer_sig(sig)
     end
 end
 
+function clear_oc_cache()
+    # Opaque closures capture the worldage of their compilation and thus are not relocatable
+    # Therefore we explicitly purge all OC's we have created here
+    for v in oc_capture_vec
+        if v isa Base.RefValue
+            p = Ptr{Ptr{Cvoid}}(pointer_from_objref(v))
+            Base.atomic_pointerset(p, C_NULL, :monotonic)
+        else
+            empty!(v)
+        end
+    end
+end
+
+# Precompilation on 1.10 hits an apparent bug: https://github.com/JuliaLang/julia/issues/56947
+function precompilation_supported()
+    return VERSION >= v"1.11" || VERSION >= v"1.10.8"
+end
+
+function precompiling()
+    return (@ccall jl_generating_output()::Cint) == 1
+end
+
 @setup_workload begin
     initialize_dialect()
     client = XLA.CPUClient(; checkcount=false)
     @compile_workload begin
-        # Precompilation on 1.10 hits an apparent bug: https://github.com/JuliaLang/julia/issues/56947
-        @static if VERSION < v"1.11"
-        else
-            # infer_sig(Tuple{typeof(Base.sum), Reactant.TracedRArray{Float64, 2}})
-            # infer_sig(Tuple{typeof(Base.sin), Reactant.TracedRNumber{Float64}})
+        @static if precompilation_supported()
             x = ConcreteRNumber(2.0; client)
             Reactant.compile(sin, (x,); client)
 
@@ -53,14 +72,5 @@ end
     XLA.free_client(client)
     client.client = C_NULL
     deinitialize_dialect()
-    # Opaque closures capture the worldage of their compilation and thus are not relocatable
-    # Therefore we explicitly purge all OC's we have created here
-    for v in oc_capture_vec
-        if v isa Base.RefValue
-            p = Ptr{Ptr{Cvoid}}(pointer_from_objref(v))
-            Base.atomic_pointerset(p, C_NULL, :monotonic)
-        else
-            empty!(v)
-        end
-    end
+    clear_oc_cache()
 end