Merge branch 'develop' into dyn_resize_gather

test/gpu/codegen_literal.cpp

@@ -80,7 +80,7 @@ TEST_CASE(mul_literal_round_test)
     migraphx::target gpu_t = migraphx::make_target("gpu");
     run_prog(p, gpu_t, m, gpu_result);
 
-    EXPECT(migraphx::verify::verify_range(ref_result, gpu_result));
+    EXPECT(migraphx::verify::verify_rms_range(gpu_result, ref_result));
 }
 
 int main(int argc, const char* argv[]) { test::run(argc, argv); }

test/gpu/fuse_mlir.cpp

@@ -34,7 +34,8 @@
 
 void run_pass(migraphx::program& p)
 {
-    migraphx::run_passes(p, {migraphx::gpu::fuse_mlir{}, migraphx::dead_code_elimination{}});
+    migraphx::run_passes(
+        p, {migraphx::gpu::fuse_mlir{.enable_extra = true}, migraphx::dead_code_elimination{}});
 }
 
 template <class F>
@@ -151,7 +152,6 @@ TEST_CASE(int_quant_dot_tanh_fails)
 
 int main(int argc, const char* argv[])
 {
-    if(migraphx::gpu::mlir_enabled())
-        test::run(argc, argv);
+    test::run(argc, argv);
     return 0;
 }

test/gpu/fuse_ops.cpp

+/*
+ * The MIT License (MIT)
+ *
+ * Copyright (c) 2015-2023 Advanced Micro Devices, Inc. All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#include "make_precompile_op.hpp"
+#include <migraphx/dead_code_elimination.hpp>
+#include <migraphx/gpu/fuse_ops.hpp>
+#include <migraphx/pass_manager.hpp>
+#include <migraphx/instruction.hpp>
+#include <migraphx/program.hpp>
+#include <basic_ops.hpp>
+#include <migraphx/make_op.hpp>
+#include <test.hpp>
+#include <pointwise.hpp>
+
+void run_pass(migraphx::program& p)
+{
+    migraphx::run_passes(p, {migraphx::gpu::fuse_ops{}, migraphx::dead_code_elimination{}});
+}
+
+TEST_CASE(layernorm_pointwise)
+{
+    migraphx::shape s{migraphx::shape::float_type, {2, 3, 4}};
+    auto create_program = [=](bool first_arg_layernorm) {
+        migraphx::program p;
+        auto* mm       = p.get_main_module();
+        auto x         = mm->add_parameter("x", s);
+        auto y         = mm->add_parameter("y", s);
+        auto z         = mm->add_parameter("z", s);
+        auto alloc     = migraphx::make_op("allocate", {{"shape", to_value(s)}});
+        auto alloc_ins = mm->add_instruction(alloc);
+        auto* pw_add1 =
+            create_pointwise_module(p, "main:pointwise0", {x, y}, single_pointwise("add"));
+        auto add1 =
+            mm->add_instruction(make_precompile_op("pointwise"), {x, y, alloc_ins}, {pw_add1});
+        auto alloc_ins2 = mm->add_instruction(alloc);
+        auto layernorm_ins =
+            mm->add_instruction(make_precompile_op("gpu::prelayernorm"), add1, alloc_ins2);
+        std::vector<migraphx::instruction_ref> pw_inputs = {layernorm_ins, z};
+        if(not first_arg_layernorm)
+        {
+            pw_inputs = {z, layernorm_ins};
+        }
+        auto* pw_add2 =
+            create_pointwise_module(p, "main:pointwise1", pw_inputs, single_pointwise("add"));
+        auto alloc_ins3 = mm->add_instruction(alloc);
+        pw_inputs.push_back(alloc_ins3);
+        auto add2 = mm->add_instruction(make_precompile_op("pointwise"), pw_inputs, {pw_add2});
+        mm->add_return({add2});
+        return p;
+    };
+
+    auto create_fused_program = [=]() {
+        migraphx::program p;
+        auto* mm       = p.get_main_module();
+        auto x         = mm->add_parameter("x", s);
+        auto y         = mm->add_parameter("y", s);
+        auto z         = mm->add_parameter("z", s);
+        auto alloc     = migraphx::make_op("allocate", {{"shape", to_value(s)}});
+        auto alloc_ins = mm->add_instruction(alloc);
+        auto* pw_add1 =
+            create_pointwise_module(p, "main:pointwise0", {x, y}, single_pointwise("add"));
+        auto add1 =
+            mm->add_instruction(make_precompile_op("pointwise"), {x, y, alloc_ins}, {pw_add1});
+        auto alloc_ins2 = mm->add_instruction(alloc);
+        auto* pw_add2 =
+            create_pointwise_module(p, "main:pointwise1", {x, z}, single_pointwise("add"));
+        auto layernorm_ins = mm->add_instruction(
+            make_precompile_op("gpu::prelayernorm"), {add1, z, alloc_ins2}, {pw_add2});
+        mm->add_return({layernorm_ins});
+        return p;
+    };
+
+    {
+        migraphx::program p1 = create_program(true);
+        run_pass(p1);
+        migraphx::program p2 = create_fused_program();
+        EXPECT(p1 == p2);
+    }
+    {
+        migraphx::program p1 = create_program(false);
+        run_pass(p1);
+        migraphx::program p2 = create_fused_program();
+        EXPECT(p1 == p2);
+    }
+}
+
+int main(int argc, const char* argv[]) { test::run(argc, argv); }

test/gpu/jit.cpp

@@ -155,7 +155,7 @@ int main() {}
 
 migraphx::src_file make_src_file(const std::string& name, const std::string& content)
 {
-    return {name, std::make_pair(content.data(), content.data() + content.size())};
+    return {name, content};
 }
 
 TEST_CASE(simple_compile_hip)
@@ -218,6 +218,15 @@ TEST_CASE(compile_warnings)
 #endif
 }
 
+TEST_CASE(has_flags)
+{
+    EXPECT(migraphx::gpu::hip_has_flags({"--std=c++17"}));
+    EXPECT(not migraphx::gpu::hip_has_flags({"--non-existent-flag-to-test-in-migraphx"}));
+    EXPECT(migraphx::gpu::hip_has_flags({"-Wunused-parameter"}));
+    EXPECT(not migraphx::gpu::hip_has_flags(
+        {"-Wnon-existent-warnings-flag-to-test-in-migraphx", "-Werror"}));
+}
+
 TEST_CASE(code_object_hip)
 {
     auto binaries = migraphx::gpu::compile_hip_src(

test/gpu/manage_host_buffer.cpp

@@ -53,7 +53,6 @@ TEST_CASE(host_same_buffer_copy)
     migraphx::parameter_map pp;
     std::vector<float> a_vec(ss.elements(), -1);
     std::vector<float> b_vec(ss.elements(), 2);
-    std::vector<float> c_vec(ss.elements(), 0);
     pp["a"] = migraphx::argument(ss, a_vec.data());
     pp["b"] = migraphx::argument(ss, b_vec.data());
     std::vector<float> gpu_result;
@@ -64,7 +63,8 @@ TEST_CASE(host_same_buffer_copy)
     auto result = p.eval(pp).back();
     std::vector<float> results_vector(ss.elements(), -1);
     result.visit([&](auto output) { results_vector.assign(output.begin(), output.end()); });
-    EXPECT(migraphx::verify::verify_range(c_vec, results_vector));
+    std::vector<float> gold_vec(ss.elements(), 0);
+    EXPECT(migraphx::verify::verify_rms_range(results_vector, gold_vec));
 }
 
 TEST_CASE(arguments_lifetime)

test/gpu/mlir.cpp

@@ -133,7 +133,8 @@ bool verify_mlir(const migraphx::module& mmlir)
     auto inputs = generate_params(ref);
 
     auto mlir = create_program_from_mlir(mmlir);
-    return migraphx::verify_args("mlir", run_ref(ref, inputs), run_gpu(mlir, inputs));
+    return migraphx::verify_args_with_tolerance(
+        "mlir", run_gpu(mlir, inputs), migraphx::verify::expected{run_ref(ref, inputs)});
 }
 
 TEST_CASE(conv)

test/gpu/quantization.cpp

@@ -40,7 +40,6 @@
 TEST_CASE(gpu_target_copy)
 {
     migraphx::target gpu_t = migraphx::make_target("gpu");
-    migraphx::target ref_t = migraphx::make_target("ref");
     migraphx::shape s{migraphx::shape::int8_type, {2, 3, 4, 5}};
 
     auto ref_arg_orig  = migraphx::generate_argument(s, 0x123456L);
@@ -52,7 +51,7 @@ TEST_CASE(gpu_target_copy)
     std::vector<int8_t> val_final;
     ref_arg_final.visit([&](auto v) { val_final.assign(v.begin(), v.end()); });
 
-    EXPECT(migraphx::verify::verify_range(val_orig, val_final));
+    EXPECT(migraphx::verify::verify_rms_range(val_orig, val_final));
 }
 
 TEST_CASE(int8_quantization)
@@ -118,9 +117,12 @@ TEST_CASE(int8_quantization)
         // the regular pipeline uses the rewrite_quantization in the much
         // earlier stage.
         if(migraphx::gpu::mlir_enabled())
-            EXPECT(migraphx::verify::verify_range(ref_result, gpu_result, 1e5));
+            EXPECT(migraphx::verify::verify_range_with_tolerance(
+                gpu_result,
+                migraphx::verify::expected{ref_result},
+                migraphx::verify::tolerance{0.01}));
         else
-            EXPECT(migraphx::verify::verify_range(ref_result, gpu_result));
+            EXPECT(migraphx::verify::verify_rms_range(gpu_result, ref_result));
     }
 }
 

test/gpu/stream_sync.cpp

@@ -64,7 +64,7 @@ int main() {}
 
 migraphx::src_file make_src_file(const std::string& name, const std::string& content)
 {
-    return {name, std::make_pair(content.data(), content.data() + content.size())};
+    return {name, content};
 }
 
 hip_stream_ptr get_stream()

test/include/pointwise.hpp

@@ -24,16 +24,16 @@
 #ifndef MIGRAPHX_GUARD_TEST_INCLUDE_POINTWISE_HPP
 #define MIGRAPHX_GUARD_TEST_INCLUDE_POINTWISE_HPP
 
+#include <migraphx/instruction_ref.hpp>
 #include <migraphx/program.hpp>
 #include <migraphx/module.hpp>
 #include <migraphx/make_op.hpp>
 
 template <class F>
-migraphx::instruction_ref add_pointwise(migraphx::program& p,
-                                        migraphx::module_ref mm,
-                                        const std::string& name,
-                                        std::vector<migraphx::instruction_ref> inputs,
-                                        F f)
+migraphx::module_ref create_pointwise_module(migraphx::program& p,
+                                             const std::string& name,
+                                             std::vector<migraphx::instruction_ref> inputs,
+                                             F f)
 {
     auto* pm = p.create_module(name);
     pm->set_bypass();
@@ -44,6 +44,17 @@ migraphx::instruction_ref add_pointwise(migraphx::program& p,
     });
     auto r = f(pm, params);
     pm->add_return({r});
+    return pm;
+}
+
+template <class F>
+migraphx::instruction_ref add_pointwise(migraphx::program& p,
+                                        migraphx::module_ref mm,
+                                        const std::string& name,
+                                        std::vector<migraphx::instruction_ref> inputs,
+                                        F f)
+{
+    auto* pm = create_pointwise_module(p, name, inputs, f);
     return mm->add_instruction(migraphx::make_op("pointwise"), inputs, {pm});
 }
 

test/include/test.hpp

@@ -339,6 +339,8 @@ inline std::ostream& operator<<(std::ostream& os, const color& c)
     static const bool use_color = isatty(STDOUT_FILENO) != 0;
     if(use_color)
         return os << "\033[" << static_cast<std::size_t>(c) << "m";
+#else
+    (void)c;
 #endif
     return os;
 }

test/jit.cpp

@@ -48,9 +48,7 @@ compile_function(const std::string& src, const std::string& flags, const std::st
     migraphx::src_compiler compiler;
     compiler.flags  = flags + "-std=c++14 -fPIC -shared";
     compiler.output = "libsimple.so";
-    migraphx::src_file f;
-    f.path     = "main.cpp";
-    f.content  = std::make_pair(src.data(), src.data() + src.size());
+    migraphx::src_file f{"main.cpp", src};
     auto image = compiler.compile({f});
     return migraphx::dynamic_loader{image}.get_function<F>(fname);
 }

test/memory_coloring_test.cpp

@@ -55,7 +55,7 @@ struct allocate
                                const migraphx::shape& output_shape,
                                const std::vector<migraphx::argument>&) const
     {
-        return {output_shape};
+        return migraphx::argument{output_shape};
     }
 };
 

test/msgpack.cpp

@@ -25,13 +25,37 @@
 #include <migraphx/value.hpp>
 #include <msgpack.hpp>
 #include <map>
+#include <numeric>
 #include "test.hpp"
 
+template <class T, MIGRAPHX_REQUIRES(not std::is_base_of<std::vector<std::uint8_t>, T>{})>
+void write_msgpack(std::ostream& os, const T& src)
+{
+    msgpack::pack(os, src);
+}
+void write_msgpack(std::ostream& os, const std::vector<std::uint8_t>& src)
+{
+    const auto limit = std::numeric_limits<uint32_t>::max() - 1;
+    std::vector<std::vector<std::uint8_t>> chunks;
+    if(src.size() > limit)
+    {
+        // Only test two chunks
+        assert(std::distance(src.begin() + limit, src.end()) < limit);
+        chunks.emplace_back(src.begin(), src.begin() + limit);
+        chunks.emplace_back(src.begin() + limit, src.end());
+    }
+    else
+    {
+        chunks = {src};
+    }
+    write_msgpack(os, chunks);
+}
+
 template <class T>
 std::vector<char> msgpack_buffer(const T& src)
 {
     std::stringstream buffer;
-    msgpack::pack(buffer, src);
+    write_msgpack(buffer, src);
     buffer.seekg(0);
     std::string str = buffer.str();
     return std::vector<char>(str.data(), str.data() + str.size()); // NOLINT
@@ -73,9 +97,12 @@ TEST_CASE(test_msgpack_bool)
 
 TEST_CASE(test_msgpack_float)
 {
-    migraphx::value v = 3.0;
+    // changed all double values in this code to not end with .0 because on msgpack for Windows if
+    // input type is double and ends with .0 it could be converted to uint64_t or int64_t and the
+    // goal of these functions is to test double without conversions
+    migraphx::value v = 3.01;
     auto buffer       = migraphx::to_msgpack(v);
-    EXPECT(buffer == msgpack_buffer(3.0));
+    EXPECT(buffer == msgpack_buffer(3.01));
     EXPECT(migraphx::from_msgpack(buffer) == v);
 }
 
@@ -105,10 +132,10 @@ TEST_CASE(test_msgpack_empty_array)
 
 TEST_CASE(test_msgpack_object)
 {
-    migraphx::value v = {{"one", 1.0}, {"three", 3.0}, {"two", 2.0}};
+    migraphx::value v = {{"one", 1.01}, {"three", 3.01}, {"two", 2.01}};
     auto buffer       = migraphx::to_msgpack(v);
     EXPECT(buffer == msgpack_buffer(std::map<std::string, double>{
-                         {"one", 1.0}, {"three", 3.0}, {"two", 2.0}}));
+                         {"one", 1.01}, {"three", 3.01}, {"two", 2.01}}));
     EXPECT(migraphx::from_msgpack(buffer) == v);
 }
 
@@ -133,18 +160,65 @@ struct foo
 
 TEST_CASE(test_msgpack_object_class)
 {
-    migraphx::value v = {{"a", 1.0}, {"b", "abc"}};
+    migraphx::value v = {{"a", 1.01}, {"b", "abc"}};
     auto buffer       = migraphx::to_msgpack(v);
-    EXPECT(buffer == msgpack_buffer(foo{1.0, "abc"}));
+    EXPECT(buffer == msgpack_buffer(foo{1.01, "abc"}));
     EXPECT(migraphx::from_msgpack(buffer) == v);
 }
 
 TEST_CASE(test_msgpack_array_class)
 {
-    migraphx::value v = {{{"a", 1.0}, {"b", "abc"}}, {{"a", 3.0}, {"b", "xyz"}}};
+    migraphx::value v = {{{"a", 1.01}, {"b", "abc"}}, {{"a", 3.01}, {"b", "xyz"}}};
     auto buffer       = migraphx::to_msgpack(v);
-    EXPECT(buffer == msgpack_buffer(std::vector<foo>{foo{1.0, "abc"}, foo{3.0, "xyz"}}));
+    EXPECT(buffer == msgpack_buffer(std::vector<foo>{foo{1.01, "abc"}, foo{3.01, "xyz"}}));
     EXPECT(migraphx::from_msgpack(buffer) == v);
 }
 
+TEST_CASE(test_msgpack_binary)
+{
+    migraphx::value::binary bin{64};
+    std::iota(bin.begin(), bin.end(), 1);
+    auto buffer = migraphx::to_msgpack(bin);
+    EXPECT(buffer == msgpack_buffer(bin));
+    EXPECT(migraphx::from_msgpack(buffer) == bin);
+}
+
+#ifndef MIGRAPHX_DISABLE_LARGE_BUFFER_TESTS
+TEST_CASE(test_msgpack_large_binary1)
+{
+    const std::size_t n   = 4LL * 1024 * 1024 * 1024 + 2;
+    const char fill_value = 2;
+    migraphx::value v;
+    {
+        std::vector<char> buffer;
+        {
+            migraphx::value::binary bin{n};
+            std::fill(bin.begin(), bin.begin() + n / 2, fill_value);
+            std::fill(bin.begin() + n / 2, bin.end(), fill_value + 1);
+            buffer = migraphx::to_msgpack(std::move(bin));
+        }
+        v = migraphx::from_msgpack(buffer);
+    }
+    EXPECT(v.is_binary());
+    EXPECT(v.get_binary().size() == n);
+    EXPECT(std::all_of(v.get_binary().begin(), v.get_binary().begin() + n / 2, [](auto c) {
+        return c == fill_value;
+    }));
+    EXPECT(std::all_of(v.get_binary().begin() + n / 2, v.get_binary().end(), [](auto c) {
+        return c == fill_value + 1;
+    }));
+}
+
+TEST_CASE(test_msgpack_binary2)
+{
+    const std::size_t n = 4LL * 1024 * 1024 * 1024 + 2;
+    migraphx::value::binary bin{n};
+    std::size_t i = 0;
+    std::generate(bin.begin(), bin.end(), [&] {
+        i++;
+        return i % 256;
+    });
+    EXPECT(migraphx::to_msgpack(bin) == msgpack_buffer(bin));
+}
+#endif
 int main(int argc, const char* argv[]) { test::run(argc, argv); }

test/multi_target/multitarget_test.cpp

@@ -37,7 +37,6 @@
 #include <migraphx/make_op.hpp>
 #include <migraphx/check_shapes.hpp>
 #include <migraphx/functional.hpp>
-#include <basic_ops.hpp>
 #include <migraphx/compile_options.hpp>
 #include <migraphx/register_target.hpp>
 #include <migraphx/generate.hpp>

test/normalize_ops_test.cpp

@@ -57,7 +57,7 @@ struct normalize_test_op
                                const migraphx::shape& output_shape,
                                const std::vector<migraphx::argument>&) const
     {
-        return {output_shape};
+        return migraphx::argument{output_shape};
     }
 };
 

test/onnx/.onnxrt-commit

-ae74a517b62baa6d973e46b5b51ac9a640512c46
+635d3faa3b3908d2806d009dc6872152cfcfcdda

test/onnx/castlike_error_test.onnx

+	castlike_error_test:M
+
+
0out"CastLikecastlike_error_testZ
+
0
+
+
+
+
+b
+out
+
+

+
+B
\ No newline at end of file

test/onnx/castlike_test.onnx

+	
castlike_test:[
+
+
0
+
1out"CastLike
castlike_testZ
+
0
+
+
+
+
+Z
+
1
+
+

+
+b
+out
+
+

+
+B
\ No newline at end of file