Merge pull request #146 from JuliaGPU/jps/llvm4

Update to LLVM.jl 4

Author: jpsamaroo

Manifest.toml

@@ -8,9 +8,9 @@ version = "1.0.1"
 
 [[Adapt]]
 deps = ["LinearAlgebra"]
-git-tree-sha1 = "ffcfa2d345aaee0ef3d8346a073d5dd03c983ebe"
+git-tree-sha1 = "84918055d15b3114ede17ac6a7182f68870c16f7"
 uuid = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"
-version = "3.2.0"
+version = "3.3.1"
 
 [[ArgTools]]
 uuid = "0dad84c5-d112-42e6-8d28-ef12dabb789f"
@@ -40,15 +40,15 @@ version = "0.4.1"
 
 [[Compat]]
 deps = ["Base64", "Dates", "DelimitedFiles", "Distributed", "InteractiveUtils", "LibGit2", "Libdl", "LinearAlgebra", "Markdown", "Mmap", "Pkg", "Printf", "REPL", "Random", "SHA", "Serialization", "SharedArrays", "Sockets", "SparseArrays", "Statistics", "Test", "UUIDs", "Unicode"]
-git-tree-sha1 = "919c7f3151e79ff196add81d7f4e45d91bbf420b"
+git-tree-sha1 = "dc7dedc2c2aa9faf59a55c622760a25cbefbe941"
 uuid = "34da2185-b29b-5c13-b0c7-acf172513d20"
-version = "3.25.0"
+version = "3.31.0"
 
 [[ConstructionBase]]
 deps = ["LinearAlgebra"]
-git-tree-sha1 = "48920211c95a6da1914a06c44ec94be70e84ffff"
+git-tree-sha1 = "f74e9d5388b8620b4cee35d4c5a618dd4dc547f4"
 uuid = "187b0558-2788-49d3-abe0-74a17ed4e7c9"
-version = "1.1.0"
+version = "1.3.0"
 
 [[DataStructures]]
 deps = ["Compat", "InteractiveUtils", "OrderedCollections"]
@@ -79,40 +79,47 @@ uuid = "ab5a07f8-06af-567f-a878-e8bb879eba5a"
 version = "0.179.0+0"
 
 [[ExprTools]]
-git-tree-sha1 = "10407a39b87f29d47ebaca8edbc75d7c302ff93e"
+git-tree-sha1 = "b7e3d17636b348f005f11040025ae8c6f645fe92"
 uuid = "e2ba6199-217a-4e67-a87a-7c52f15ade04"
-version = "0.1.3"
+version = "0.1.6"
 
 [[Future]]
 deps = ["Random"]
 uuid = "9fa8497b-333b-5362-9e8d-4d0656e87820"
 
 [[GPUArrays]]
-deps = ["AbstractFFTs", "Adapt", "LinearAlgebra", "Printf", "Random", "Serialization"]
-git-tree-sha1 = "f99a25fe0313121f2f9627002734c7d63b4dd3bd"
+deps = ["AbstractFFTs", "Adapt", "LinearAlgebra", "Printf", "Random", "Serialization", "Statistics"]
+git-tree-sha1 = "df5b8569904c5c10e84c640984cfff054b18c086"
 uuid = "0c68f7d7-f131-5f86-a1c3-88cf8149b2d7"
-version = "6.2.0"
+version = "6.4.1"
 
 [[GPUCompiler]]
-deps = ["DataStructures", "ExprTools", "InteractiveUtils", "LLVM", "Libdl", "Logging", "Scratch", "Serialization", "TimerOutputs", "UUIDs"]
-git-tree-sha1 = "0bae2d5a40c54c6c680a1a223f83a3c415ec730b"
+deps = ["DataStructures", "ExprTools", "InteractiveUtils", "LLVM", "Libdl", "Logging", "TimerOutputs", "UUIDs"]
+git-tree-sha1 = "e8a09182a4440489e2e3dedff5ad3f6bbe555396"
 uuid = "61eb1bfa-7361-4325-ad38-22787b887f55"
-version = "0.11.0"
+version = "0.12.5"
 
 [[InteractiveUtils]]
 deps = ["Markdown"]
 uuid = "b77e0a4c-d291-57a0-90e8-8db25a27a240"
 
 [[JLLWrappers]]
-git-tree-sha1 = "a431f5f2ca3f4feef3bd7a5e94b8b8d4f2f647a0"
+deps = ["Preferences"]
+git-tree-sha1 = "642a199af8b68253517b80bd3bfd17eb4e84df6e"
 uuid = "692b3bcd-3c85-4b1f-b108-f13ce0eb3210"
-version = "1.2.0"
+version = "1.3.0"
 
 [[LLVM]]
-deps = ["CEnum", "Libdl", "Printf", "Unicode"]
-git-tree-sha1 = "b616937c31337576360cb9fb872ec7633af7b194"
+deps = ["CEnum", "LLVMExtra_jll", "Libdl", "Printf", "Unicode"]
+git-tree-sha1 = "1b7ba36ea7aa6fa2278118951bad114fbb8359f2"
 uuid = "929cbde3-209d-540e-8aea-75f648917ca0"
-version = "3.6.0"
+version = "4.1.0"
+
+[[LLVMExtra_jll]]
+deps = ["Artifacts", "JLLWrappers", "Libdl", "Pkg"]
+git-tree-sha1 = "b36c0677a0549c7d1dc8719899a4133abbfacf7d"
+uuid = "dad2f222-ce93-54a1-a47d-0025e8a3acab"
+version = "0.0.6+0"
 
 [[LibCURL]]
 deps = ["LibCURL_jll", "MozillaCACerts_jll"]
@@ -170,14 +177,20 @@ version = "2.0.13+1"
 uuid = "ca575930-c2e3-43a9-ace4-1e988b2c1908"
 
 [[OrderedCollections]]
-git-tree-sha1 = "4fa2ba51070ec13fcc7517db714445b4ab986bdf"
+git-tree-sha1 = "85f8e6578bf1f9ee0d11e7bb1b1456435479d47c"
 uuid = "bac558e1-5e72-5ebc-8fee-abe8a469f55d"
-version = "1.4.0"
+version = "1.4.1"
 
 [[Pkg]]
 deps = ["Artifacts", "Dates", "Downloads", "LibGit2", "Libdl", "Logging", "Markdown", "Printf", "REPL", "Random", "SHA", "Serialization", "TOML", "Tar", "UUIDs", "p7zip_jll"]
 uuid = "44cfe95a-1eb2-52ea-b672-e2afdf69b78f"
 
+[[Preferences]]
+deps = ["TOML"]
+git-tree-sha1 = "00cfd92944ca9c760982747e9a1d0d5d86ab1e5a"
+uuid = "21216c6a-2e73-6563-6e65-726566657250"
+version = "1.2.2"
+
 [[Printf]]
 deps = ["Unicode"]
 uuid = "de0858da-6303-5e67-8744-51eddeeeb8d7"
@@ -199,12 +212,6 @@ version = "1.1.3"
 [[SHA]]
 uuid = "ea8e919c-243c-51af-8825-aaa63cd721ce"
 
-[[Scratch]]
-deps = ["Dates"]
-git-tree-sha1 = "ad4b278adb62d185bbcb6864dc24959ab0627bf6"
-uuid = "6c6a2e73-6563-6170-7368-637461726353"
-version = "1.0.3"
-
 [[Serialization]]
 uuid = "9e88b42a-f829-5b0c-bbe9-9e923198166b"
 
@@ -242,10 +249,10 @@ deps = ["InteractiveUtils", "Logging", "Random", "Serialization"]
 uuid = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [[TimerOutputs]]
-deps = ["Printf"]
-git-tree-sha1 = "32cdbe6cd2d214c25a0b88f985c9e0092877c236"
+deps = ["ExprTools", "Printf"]
+git-tree-sha1 = "209a8326c4f955e2442c07b56029e88bb48299c7"
 uuid = "a759f4b9-e2f1-59dc-863e-4aeb61b1ea8f"
-version = "0.5.8"
+version = "0.5.12"
 
 [[UUIDs]]
 deps = ["Random", "SHA"]
@@ -272,9 +279,9 @@ version = "1.3.0+0"
 
 [[fts_jll]]
 deps = ["Artifacts", "JLLWrappers", "Libdl", "Pkg"]
-git-tree-sha1 = "fa3d234bbbad588979c5409eee29e7967b6ed8a6"
+git-tree-sha1 = "78732b942383d2cb521df8a1a0814911144e663d"
 uuid = "d65627f6-89bd-53e8-8ab5-8b75ff535eee"
-version = "1.2.7+0"
+version = "1.2.7+1"
 
 [[hsa_rocr_jll]]
 deps = ["Artifacts", "Elfutils_jll", "JLLWrappers", "Libdl", "NUMA_jll", "Pkg", "Zlib_jll", "hsakmt_roct_jll"]
@@ -294,9 +301,9 @@ uuid = "8e850ede-7688-5339-a07c-302acd2aaf8d"
 
 [[obstack_jll]]
 deps = ["Artifacts", "JLLWrappers", "Libdl", "Pkg"]
-git-tree-sha1 = "c417cbbce3efb49a309b2d478fea9df7cfe024c0"
+git-tree-sha1 = "1c4a6b66e934fc6db4649cb2910c72f53bbfea7e"
 uuid = "c88a4935-d25e-5644-aacc-5db6f1b8ef79"
-version = "1.1.0+0"
+version = "1.2.2+0"
 
 [[p7zip_jll]]
 deps = ["Artifacts", "Libdl"]

Project.toml

@@ -1,7 +1,7 @@
 name = "AMDGPU"
 uuid = "21141c5a-9bdb-4563-92ae-f87d6854732e"
 authors = ["Julian P Samaroo <jpsamaroo@jpsamaroo.me>"]
-version = "0.2.8"
+version = "0.2.9"
 
 [deps]
 AbstractFFTs = "621f4979-c628-5d54-868e-fcf4e3e8185c"
@@ -28,7 +28,7 @@ BinaryProvider = "0.5"
 CEnum = "0.2, 0.3, 0.4"
 GPUArrays = "6"
 GPUCompiler = "0.12"
-LLVM = "3"
+LLVM = "4"
 MacroTools = "0.5"
 Requires = "1"
 Setfield = "0.5, 0.6, 0.7"

src/device/gcn/assertion.jl

@@ -38,7 +38,7 @@ end
 assert_counter = 0
 
 @generated function rocassert_fail(::Val{msg}, ::Val{file}, ::Val{line}) where {msg, file, line}
-    JuliaContext() do ctx
+    Context() do ctx
         T_void = LLVM.VoidType(ctx)
         T_int32 = LLVM.Int32Type(ctx)
         T_pint8 = LLVM.PointerType(LLVM.Int8Type(ctx))
@@ -49,15 +49,15 @@ assert_counter = 0
 
         # generate IR
         Builder(ctx) do builder
-            entry = BasicBlock(llvm_f, "entry", ctx)
+            entry = BasicBlock(llvm_f, "entry"; ctx)
             position!(builder, entry)
             global assert_counter
             assert_counter += 1
             message = globalstring_ptr!(builder, String(msg), "assert_message_$(assert_counter)")
             file = globalstring_ptr!(builder, String(file), "assert_file_$(assert_counter)")
             line = ConstantInt(T_int32, line)
             func = globalstring_ptr!(builder, "unknown", "assert_function_$(assert_counter)")
-            charSize = ConstantInt(Csize_t(1), ctx)
+            charSize = ConstantInt(Csize_t(1); ctx)
 
             # invoke __assertfail and return
             # TODO: mark noreturn since we don't use ptxas?

src/device/gcn/atomics.jl

@@ -21,16 +21,16 @@
 # >   that points to either the global address space or the shared address space.
 
 @generated function llvm_atomic_op(::Val{binop}, ptr::LLVMPtr{T,A}, val::T) where {binop, T, A}
-    JuliaContext() do ctx
-        T_val = convert(LLVMType, T, ctx)
-        T_ptr = convert(LLVMType, ptr, ctx)
+    Context() do ctx
+        T_val = convert(LLVMType, T; ctx)
+        T_ptr = convert(LLVMType, ptr; ctx)
 
         T_typed_ptr = LLVM.PointerType(T_val, A)
 
         llvm_f, _ = create_function(T_val, [T_ptr, T_val])
 
         Builder(ctx) do builder
-            entry = BasicBlock(llvm_f, "entry", ctx)
+            entry = BasicBlock(llvm_f, "entry"; ctx)
             position!(builder, entry)
 
             typed_ptr = bitcast!(builder, parameters(llvm_f)[1], T_typed_ptr)
@@ -42,7 +42,7 @@
             ret!(builder, rv)
         end
 
-        call_function(llvm_f, T, Tuple{LLVMPtr{T,A}, T}, :((ptr,val)))
+        call_function(llvm_f, T, Tuple{LLVMPtr{T,A}, T}, :ptr, :val)
     end
 end
 
@@ -89,16 +89,16 @@ for T in (Int32, Int64, UInt32, UInt64)
 end
 
 @generated function llvm_atomic_cas(ptr::LLVMPtr{T,A}, cmp::T, val::T) where {T, A}
-    JuliaContext() do ctx
-        T_val = convert(LLVMType, T, ctx)
-        T_ptr = convert(LLVMType, ptr, ctx)
+    Context() do ctx
+        T_val = convert(LLVMType, T; ctx)
+        T_ptr = convert(LLVMType, ptr; ctx)
 
         T_typed_ptr = LLVM.PointerType(T_val, A)
 
         llvm_f, _ = create_function(T_val, [T_ptr, T_val, T_val])
 
         Builder(ctx) do builder
-            entry = BasicBlock(llvm_f, "entry", ctx)
+            entry = BasicBlock(llvm_f, "entry"; ctx)
             position!(builder, entry)
 
             typed_ptr = bitcast!(builder, parameters(llvm_f)[1], T_typed_ptr)
@@ -112,7 +112,7 @@ end
             ret!(builder, rv)
         end
 
-        call_function(llvm_f, T, Tuple{LLVMPtr{T,A}, T, T}, :((ptr,cmp,val)))
+        call_function(llvm_f, T, Tuple{LLVMPtr{T,A}, T, T}, :ptr, :cmp, :val)
     end
 end
 

src/device/gcn/execution_control.jl

@@ -3,7 +3,7 @@
 const completion_signal_base = _packet_offsets[findfirst(x->x==:completion_signal,_packet_names)]
 
 @generated function _completion_signal()
-    JuliaContext() do ctx
+    Context() do ctx
         T_int8 = LLVM.Int8Type(ctx)
         T_int64 = LLVM.Int64Type(ctx)
         _as = convert(Int, AS.Constant)
@@ -16,7 +16,7 @@ const completion_signal_base = _packet_offsets[findfirst(x->x==:completion_signa
 
         # generate IR
         Builder(ctx) do builder
-            entry = BasicBlock(llvm_f, "entry", ctx)
+            entry = BasicBlock(llvm_f, "entry"; ctx)
             position!(builder, entry)
 
             # get the kernel dispatch pointer
@@ -25,7 +25,7 @@ const completion_signal_base = _packet_offsets[findfirst(x->x==:completion_signa
             ptr = call!(builder, intr)
 
             # load the index
-            signal_ptr_i8 = inbounds_gep!(builder, ptr, [ConstantInt(completion_signal_base, ctx)])
+            signal_ptr_i8 = inbounds_gep!(builder, ptr, [ConstantInt(completion_signal_base; ctx)])
             signal_ptr = bitcast!(builder, signal_ptr_i8, T_ptr_i64)
             signal = load!(builder, signal_ptr)
             ret!(builder, signal)

src/device/gcn/helpers.jl

@@ -3,21 +3,21 @@ _packet_names = fieldnames(HSA.KernelDispatchPacket)
 _packet_offsets = fieldoffset.(HSA.KernelDispatchPacket, 1:length(_packet_names))
 
 @generated function _intr(::Val{fname}, out_arg, inp_args...) where {fname,}
-    JuliaContext() do ctx
+    Context() do ctx
         inp_exprs = [:( inp_args[$i] ) for i in 1:length(inp_args)]
         inp_types = [inp_args...]
-        out_type = convert(LLVMType, out_arg.parameters[1], ctx)
+        out_type = convert(LLVMType, out_arg.parameters[1]; ctx)
 
         # create function
         bool_types = map(x->x===Bool, inp_types)
         T_bool = LLVM.Int1Type(ctx)
-        param_types = LLVMType[convert.(LLVMType, inp_types, Ref(ctx))...]
+        param_types = LLVMType[convert.(LLVMType, inp_types; ctx=ctx)...]
         llvm_f, _ = create_function(out_type, param_types)
         mod = LLVM.parent(llvm_f)
 
         # generate IR
         Builder(ctx) do builder
-            entry = BasicBlock(llvm_f, "entry", ctx)
+            entry = BasicBlock(llvm_f, "entry"; ctx)
             position!(builder, entry)
 
             # call the intrinsic
@@ -28,19 +28,19 @@ _packet_offsets = fieldoffset.(HSA.KernelDispatchPacket, 1:length(_packet_names)
             for idx in 1:length(param_types)
                 if bool_types[idx]
                     attrs = parameter_attributes(intr, idx)
-                    push!(attrs, EnumAttribute("zeroext", 0, ctx))
+                    push!(attrs, EnumAttribute("zeroext", 0; ctx))
                 end
             end
             params = map(x->bool_types[x[1]] ? trunc!(builder, x[2], T_bool) : x[2], enumerate(parameters(llvm_f)))
             value = call!(builder, intr, [params...])
             if out_arg === Type{Bool}
-                ret!(builder, zext!(builder, value, convert(LLVMType, Bool, ctx)))
+                ret!(builder, zext!(builder, value, convert(LLVMType, Bool; ctx)))
             else
                 ret!(builder, value)
             end
         end
 
-        call_function(llvm_f, out_arg.parameters[1], Tuple{inp_args...}, Expr(:tuple, inp_exprs...))
+        call_function(llvm_f, out_arg.parameters[1], Tuple{inp_args...}, inp_exprs...)
     end
 end