Merge branch 'develop' into uif2-initial

docs/.sphinx/requirements.txt

@@ -89,7 +89,7 @@ requests==2.28.2
     # via
     #   pygithub
     #   sphinx
-rocm-docs-core==0.30.0
+rocm-docs-core==0.30.1
     # via -r requirements.in
 smmap==5.0.0
     # via gitdb

src/include/migraphx/op/dequantizelinear.hpp

@@ -72,8 +72,8 @@ struct dequantizelinear
         visit_all(x, x_zero_point)([&](auto input, auto zero_pts) {
             visit_all(result, x_scale)([&](auto output, auto scales) {
                 par_for(output_shape.elements(), [&](auto i) {
-                    output[i] = static_cast<double>(static_cast<int64_t>(input[i]) -
-                                                    static_cast<int64_t>(zero_pts[i])) *
+                    output[i] = static_cast<double>(static_cast<double>(input[i]) -
+                                                    static_cast<double>(zero_pts[i])) *
                                 scales[i];
                 });
             });

src/include/migraphx/op/quant_convolution.hpp

@@ -27,6 +27,7 @@
 #include <migraphx/op/common.hpp>
 #include <migraphx/argument.hpp>
 #include <migraphx/check_shapes.hpp>
+#include <migraphx/shape.hpp>
 #include <migraphx/config.hpp>
 #include <migraphx/convolution.hpp>
 #include <migraphx/value.hpp>
@@ -87,11 +88,13 @@ struct quant_convolution
         }
 
         // all input type must be int8_type and output is float_type
-        if(t != shape::int8_type)
+        std::set<migraphx::shape::type_t> supported_types = {shape::int8_type,
+                                                             shape::fp8e4m3fnuz_type};
+        if(not contains(supported_types, t))
         {
-            MIGRAPHX_THROW("QUANT_CONVOLUTION: only accept input and weights of type int8_t");
+            MIGRAPHX_THROW("QUANT_CONVOLUTION: only accept input and weights of type int8_t or "
+                           "fp8e4m3fnuz_type");
         }
-        t = shape::int32_type;
 
         std::vector<size_t> output_lens{input.lens()[0], weights.lens()[0]};
         auto padding_size = padding.size();
@@ -107,8 +110,11 @@ struct quant_convolution
                         stride[i] +
                     1)));
         }
-
-        return inputs[0].with_lens(t, output_lens);
+        if(t == shape::int8_type)
+        {
+            return inputs[0].with_lens(shape::int32_type, output_lens);
+        } // else fp8 conv
+        return inputs[0].with_lens(shape::float_type, output_lens);
     }
 
     size_t kdims() const

src/include/migraphx/op/quantizelinear.hpp

@@ -80,10 +80,10 @@ struct quantizelinear
                 auto min_value   = std::numeric_limits<quant_type>::min();
                 auto max_value   = std::numeric_limits<quant_type>::max();
                 par_for(output_shape.elements(), [&](auto i) {
-                    int64_t quantized = static_cast<int64_t>(std::nearbyint(input[i] / scales[i])) +
-                                        static_cast<int64_t>(zero_pts[i]);
-                    output[i] = std::max(static_cast<int64_t>(min_value),
-                                         std::min(static_cast<int64_t>(max_value), quantized));
+                    double quantized = static_cast<double>(std::nearbyint(input[i] / scales[i])) +
+                                       static_cast<double>(zero_pts[i]);
+                    output[i] = std::max(static_cast<double>(min_value),
+                                         std::min(static_cast<double>(max_value), quantized));
                 });
             });
         });

src/module.cpp

@@ -669,6 +669,15 @@ void module::finalize(std::vector<context>& contexts)
             smod->finalize(contexts);
         }
     }
+#ifndef BUILD_DEV
+    if(std::any_of(this->begin(), this->end(), [](const auto i) {
+           return i.get_shape().type() == migraphx::shape::fp8e4m3fnuz_type;
+       }))
+    {
+        std::cout << "[Warning] : MIGraphX has BETA support for FP8. Using FP8 may result in "
+                     "incorrect final outputs\n";
+    }
+#endif
 
     // Warn when an instruction is not normalized
     auto ins = std::find_if(begin(), end(), [](auto& i) { return i.need_normalization(); });

src/onnx/onnx_parser.cpp

@@ -625,7 +625,11 @@ shape::type_t get_type(int dtype)
     case 11: return shape::double_type;
     case 12: return shape::uint32_type;
     case 13: return shape::uint64_type;
-    case 18: return shape::fp8e4m3fnuz_type;
+    case 18: {
+        std::cout << "[Warning] : MIGraphX has BETA support for FP8. Using FP8 may result in "
+                     "incorrect final outputs\n";
+        return shape::fp8e4m3fnuz_type;
+    }
     case 14:
     case 15:
     case 16:

src/rewrite_quantization.cpp

@@ -58,8 +58,8 @@ void apply_quantizelinear(module& m, instruction_ref ins)
         add_zero_point = m.insert_instruction(ins, make_op("add"), add_zero_point, zero_point);
     }
 
-    int64_t max_quant = 0;
-    int64_t min_quant = 0;
+    double max_quant = 0;
+    double min_quant = 0;
     ins->get_shape().visit_type([&](auto qt) {
         max_quant = qt.max();
         min_quant = qt.min();
@@ -70,8 +70,8 @@ void apply_quantizelinear(module& m, instruction_ref ins)
 
     if(enabled(MIGRAPHX_ENABLE_CK_WORKAROUNDS{}))
     {
-        std::vector<int> min_data(s.elements(), min_quant);
-        std::vector<int> max_data(s.elements(), max_quant);
+        std::vector<double> min_data(s.elements(), min_quant);
+        std::vector<double> max_data(s.elements(), max_quant);
         min_arg = m.add_literal(literal(s, min_data));
         max_arg = m.add_literal(literal(s, max_data));
     }

src/simplify_qdq.cpp

@@ -82,18 +82,21 @@ struct match_find_quantizable_ops
     // Helper function to insert quantized versions of any broadcasts and transpose ops that
     // occur between dequantizelinear and the quantized op
     static auto
-    propagate_quantized_ins(module& m, const instruction_ref dqins, const instruction_ref qop)
+    propagate_quantized_ins(module& m, const instruction_ref dqins, const instruction_ref qop_arg)
     {
-        auto qinp     = dqins->inputs().front();
-        auto next_ins = dqins;
-
-        while(next_ins != qop)
-        {
-            if(next_ins->name() != "dequantizelinear")
+        auto prev_ins = qop_arg;
+        std::vector<instruction_ref> ins_inbetween;
+        // matcher skips continguous, multi/broadcasts and transposes, collect all those
+        // instructions
+        while(prev_ins != dqins)
         {
-                qinp = m.insert_instruction(qop, next_ins->get_operator(), qinp);
+            ins_inbetween.push_back(prev_ins);
+            prev_ins = prev_ins->inputs().front();
         }
-            next_ins = next_ins->outputs().front();
+        auto qinp = dqins->inputs().front();
+        for(auto ins : reverse_iterator_for(ins_inbetween))
+        {
+            qinp = m.insert_instruction(dqins, (*ins)->get_operator(), {qinp});
         }
         return qinp;
     }
@@ -124,10 +127,11 @@ struct match_find_quantizable_ops
         auto scale2 = r.instructions["scale2"];
         auto zp1    = r.instructions["zp1"];
         auto zp2    = r.instructions["zp2"];
-
-        // Only INT8 type currently supported
-        if(dq1->inputs().front()->get_shape().type() != migraphx::shape::int8_type or
-           dq2->inputs().front()->get_shape().type() != migraphx::shape::int8_type)
+        // Only INT8 or FP8 type currently supported
+        std::set<migraphx::shape::type_t> supported_types = {migraphx::shape::fp8e4m3fnuz_type,
+                                                             migraphx::shape::int8_type};
+        if(not contains(supported_types, dq1->inputs().front()->get_shape().type()) or
+           not contains(supported_types, dq2->inputs().front()->get_shape().type()))
             return;
 
         // Only symmetric quantization supported (ie. non-zero zero_points not allowed)
@@ -140,8 +144,8 @@ struct match_find_quantizable_ops
 
         // Propagate q1 and q2 through any broadcasts and transposes before qop
         auto qop_args  = qop->inputs();
-        qop_args.at(0) = propagate_quantized_ins(m, dq1, qop);
-        qop_args.at(1) = propagate_quantized_ins(m, dq2, qop);
+        qop_args.at(0) = propagate_quantized_ins(m, dq1, qop_args[0]);
+        qop_args.at(1) = propagate_quantized_ins(m, dq2, qop_args[1]);
         instruction_ref dq;
         instruction_ref out_scale;
         instruction_ref zero_point;

src/targets/gpu/device_name.cpp

@@ -49,6 +49,12 @@ std::string get_device_name()
     return props.gcnArchName;
 }
 
+bool gfx_has_fp8_intrinsics()
+{
+    const auto device_name = trim(split_string(get_device_name(), ':').front());
+    return (starts_with(device_name, "gfx9") and device_name >= "gfx940");
+}
+
 } // namespace gpu
 } // namespace MIGRAPHX_INLINE_NS
 } // namespace migraphx

src/targets/gpu/fuse_mlir.cpp

@@ -218,6 +218,7 @@ auto is_mlir_conv(mlir_mode mode)
             return false;
         if(ins->name() != "convolution" and ins->name() != "quant_convolution")
             return false;
+        auto input_arg_t = ins->inputs().front()->get_shape().type();
         value v    = ins->get_operator().to_value();
         auto group = v.at("group").to<int>();
         if(group != 1)
@@ -225,6 +226,10 @@ auto is_mlir_conv(mlir_mode mode)
         // Avoid MLIR assertion: Index < Length && "Invalid index!"
         if(ins->get_shape().lens().size() != 4)
             return false;
+        if(ins->get_shape().type() == shape::fp8e4m3fnuz_type)
+            return true;
+        if(ins->get_shape().type() == shape::float_type and input_arg_t == shape::fp8e4m3fnuz_type)
+            return true;
         if(ins->get_shape().type() == shape::int8_type)
             return true;
         if(mode == mlir_mode::int8)
@@ -292,6 +297,7 @@ bool is_pointwise_op_supported_by_mlir(const instruction& i)
     const auto result_type                            = i.get_shape().type();
     const std::initializer_list<type_t> allowed_types = {type_t::float_type,
                                                          type_t::half_type,
+                                                         type_t::fp8e4m3fnuz_type,
                                                          type_t::int8_type,
                                                          type_t::int32_type,
                                                          type_t::bool_type};
@@ -331,7 +337,8 @@ bool is_pointwise_op_supported_by_mlir(const instruction& i)
         "softmax",
         "tanh",
     };
-    bool is_float = contains({type_t::float_type, type_t::half_type}, result_type);
+    bool is_float =
+        contains({type_t::float_type, type_t::half_type, type_t::fp8e4m3fnuz_type}, result_type);
     if(contains(any_type_ops, name))
         return true;
     if(result_type != type_t::bool_type and contains(no_bool_ops, name))
@@ -342,6 +349,10 @@ bool is_pointwise_op_supported_by_mlir(const instruction& i)
     // supported.
     if(is_float and name == "convert")
     {
+        if(result_type == shape::fp8e4m3fnuz_type)
+        {
+            return false;
+        } // else
         return std::all_of(i.inputs().begin(), i.inputs().end(), [](const auto& arg) {
             return contains({type_t::float_type, type_t::half_type}, arg->get_shape().type());
         });
@@ -404,11 +415,12 @@ struct find_mlir_standalone_op
     void apply(module_pass_manager& mpm, const match::matcher_result& r) const
     {
         auto gemm_based_op = r.result;
-        //
-        // enable only for fp32/fp16/i8 types
+        // enable only for fp32/fp16/i8/fp8 types
         if(std::any_of(gemm_based_op->inputs().begin(), gemm_based_op->inputs().end(), [&](auto i) {
-               return not contains(
-                   {shape::type_t::float_type, shape::type_t::half_type, shape::type_t::int8_type},
+               return not contains({shape::type_t::float_type,
+                                    shape::type_t::half_type,
+                                    shape::type_t::int8_type,
+                                    shape::type_t::fp8e4m3fnuz_type},
                                    i->get_shape().type());
            }))
             return;

src/targets/gpu/include/migraphx/gpu/device_name.hpp

@@ -37,6 +37,8 @@ MIGRAPHX_GPU_EXPORT std::string get_device_name();
 
 MIGRAPHX_GPU_EXPORT int get_device_id();
 
+MIGRAPHX_GPU_EXPORT bool gfx_has_fp8_intrinsics();
+
 } // namespace gpu
 } // namespace MIGRAPHX_INLINE_NS
 } // namespace migraphx

src/targets/gpu/mlir.cpp

@@ -300,6 +300,8 @@ struct mlir_program
                 result = mlirF32TypeGet(ctx.get());
             else if(as.type_enum() == shape::half_type)
                 result = mlirF16TypeGet(ctx.get());
+            else if(as.type_enum() == shape::fp8e4m3fnuz_type)
+                result = mlirFloat8E4M3FNUZTypeGet(ctx.get());
             else if(as.type_enum() == shape::double_type)
                 result = mlirF64TypeGet(ctx.get());
             else if(as.is_integral())

src/targets/gpu/rocblas.cpp

@@ -58,8 +58,7 @@ bool rocblas_fp8_available()
 #ifndef MIGRAPHX_USE_ROCBLAS_FP8_API
     return false;
 #else
-    const auto device_name = trim(split_string(get_device_name(), ':').front());
-    return (starts_with(device_name, "gfx9") and device_name >= "gfx940");
+    return gfx_has_fp8_intrinsics();
 #endif
 }
 

src/targets/gpu/target.cpp

@@ -105,11 +105,19 @@ std::vector<pass> target::get_passes(migraphx::context& gctx, const compile_opti
     unsupported_types.erase(shape::type_t::uint8_type);
     unsupported_types.erase(shape::type_t::int32_type);
     unsupported_types.erase(shape::type_t::tuple_type);
+    // whiltelist supported Ops for the FP8
     std::set<std::string> unsupported_fp8_ops = {};
     if(not gpu::rocblas_fp8_available())
     {
         unsupported_fp8_ops.insert("dot");
     }
+    // MIOpen doesn't have support for fp8 pooling yet.
+    unsupported_fp8_ops.insert("pooling");
+    if(not gpu::gfx_has_fp8_intrinsics())
+    {
+        unsupported_fp8_ops.insert("convolution");
+        unsupported_fp8_ops.insert("quant_convolution");
+    }
     // add all device kernels
     unsupported_fp8_ops.insert("logsoftmax");
     unsupported_fp8_ops.insert("nonzero");

test/simplify_qdq_test.cpp

@@ -527,6 +527,62 @@ TEST_CASE(dot_add)
     EXPECT(m1 == m2);
 }
 
+TEST_CASE(dot_add_multiple_dq_use)
+{
+    migraphx::shape sh1{migraphx::shape::float_type, {32, 1}};
+    migraphx::shape sh2{migraphx::shape::float_type, {32, 32}};
+    migraphx::module m1;
+    {
+        auto t1    = m1.add_parameter("t1", sh1);
+        auto t2    = m1.add_parameter("t2", sh2);
+        auto scale = m1.add_literal(0.5f);
+        auto zero  = m1.add_literal(std::int8_t{0});
+
+        auto q1 = add_quantize_op(m1, "quantizelinear", t1, scale, zero);
+        auto d1 = add_quantize_op(m1, "dequantizelinear", q1, scale, zero);
+        auto d1_t =
+            m1.add_instruction(migraphx::make_op("transpose", {{"permutation", {1, 0}}}), d1);
+        auto d1_tmb =
+            m1.add_instruction(migraphx::make_op("multibroadcast", {{"out_lens", {32, 32}}}), d1_t);
+        auto d1_tmbc = m1.add_instruction(migraphx::make_op("contiguous"), d1_tmb);
+        auto q2      = add_quantize_op(m1, "quantizelinear", t2, scale, zero);
+        auto d2      = add_quantize_op(m1, "dequantizelinear", q2, scale, zero);
+        auto dot_1   = m1.add_instruction(migraphx::make_op("dot"), d1_tmbc, d2);
+        auto q3      = add_quantize_op(m1, "quantizelinear", dot_1, scale, zero);
+        auto d3      = add_quantize_op(m1, "dequantizelinear", q3, scale, zero);
+        auto dot_2   = m1.add_instruction(migraphx::make_op("dot"), d3, d1);
+        auto add     = m1.add_instruction(migraphx::make_op("add"), {dot_2, d1});
+        m1.add_return({add});
+    }
+
+    migraphx::module m2;
+    {
+        auto t1    = m2.add_parameter("t1", sh1);
+        auto t2    = m2.add_parameter("t2", sh2);
+        auto scale = m2.add_literal(0.5f);
+        auto zero  = m2.add_literal(std::int8_t{0});
+
+        auto q1 = add_quantize_op(m2, "quantizelinear", t1, scale, zero);
+        auto q1_t =
+            m2.add_instruction(migraphx::make_op("transpose", {{"permutation", {1, 0}}}), q1);
+        auto q1_tmb =
+            m2.add_instruction(migraphx::make_op("multibroadcast", {{"out_lens", {32, 32}}}), q1_t);
+        auto q1_tmbc     = m2.add_instruction(migraphx::make_op("contiguous"), q1_tmb);
+        auto q2          = add_quantize_op(m2, "quantizelinear", t2, scale, zero);
+        auto dot_1       = m2.add_instruction(migraphx::make_op("quant_dot"), q1_tmbc, q2);
+        auto out_scale   = add_scale_mul(m2, scale, scale, 1, 1, dot_1->get_shape().lens());
+        auto d3          = add_quantize_op(m2, "dequantizelinear", dot_1, out_scale);
+        auto d3_q        = add_quantize_op(m2, "quantizelinear", d3, scale, zero);
+        auto dot_2       = m2.add_instruction(migraphx::make_op("quant_dot"), d3_q, q1);
+        auto out_scale_2 = add_scale_mul(m2, scale, scale, 1, 1, dot_2->get_shape().lens());
+        auto d4          = add_quantize_op(m2, "dequantizelinear", dot_2, out_scale_2);
+        auto add         = m2.add_instruction(migraphx::make_op("add"), d4, t1);
+        m2.add_return({add});
+    }
+    run_pass(m1);
+    EXPECT(m1 == m2);
+}
+
 TEST_CASE(conv)
 {
     migraphx::shape s4{migraphx::shape::int8_type, {1280, 320, 1, 1}};
@@ -919,7 +975,6 @@ TEST_CASE(mobilenet_snippet)
 
     auto mod1 = create_module();
     auto mod2 = create_module();
-
     run_pass(mod2);
 
     auto match_qdq = migraphx::match::name("dequantizelinear")(

test/verify/main.cpp

@@ -77,6 +77,5 @@ int main(int argc, const char* argv[])
                          "test_split_single_dyn_dim",
                          "test_instancenorm_large_3d<migraphx::shape::float_type>",
                          "test_instancenorm_large_3d<migraphx::shape::half_type>"});
-    rv.disable_test_for("gpu", {"test_conv_bn_add"});
     rv.run(argc, argv);
 }

test/verify/quant_conv.cpp

@@ -27,17 +27,21 @@
 #include <migraphx/generate.hpp>
 #include <migraphx/make_op.hpp>
 
-struct quant_conv : verify_program<quant_conv>
+template <migraphx::shape::type_t DType>
+struct quant_conv : verify_program<quant_conv<DType>>
 {
     migraphx::program create_program() const
     {
         migraphx::program p;
         auto* mm = p.get_main_module();
-        migraphx::shape a_shape{migraphx::shape::int8_type, {2, 3, 4, 4}};
+        migraphx::shape a_shape{DType, {2, 3, 4, 4}};
         auto pa = mm->add_parameter("a", a_shape);
-        migraphx::shape c_shape{migraphx::shape::int8_type, {2, 3, 3, 3}};
+        migraphx::shape c_shape{DType, {2, 3, 3, 3}};
         auto pc = mm->add_parameter("c", c_shape);
         mm->add_instruction(migraphx::make_op("quant_convolution"), pa, pc);
         return p;
     }
 };
+
+template struct quant_conv<migraphx::shape::int8_type>;
+template struct quant_conv<migraphx::shape::fp8e4m3fnuz_type>;

test/verify/quant_conv_1.cpp

@@ -27,17 +27,21 @@
 #include <migraphx/generate.hpp>
 #include <migraphx/op/quant_convolution.hpp>
 
-struct quant_conv_1 : verify_program<quant_conv_1>
+template <migraphx::shape::type_t DType>
+struct quant_conv_1 : verify_program<quant_conv_1<DType>>
 {
     migraphx::program create_program() const
     {
         migraphx::program p;
         auto* mm = p.get_main_module();
-        migraphx::shape a_shape{migraphx::shape::int8_type, {2, 3, 4, 4}};
+        migraphx::shape a_shape{DType, {2, 3, 4, 4}};
         auto pa = mm->add_parameter("a", a_shape);
-        migraphx::shape c_shape{migraphx::shape::int8_type, {2, 3, 3, 3}};
+        migraphx::shape c_shape{DType, {2, 3, 3, 3}};
         auto pc = mm->add_parameter("c", c_shape);
         mm->add_instruction(migraphx::op::quant_convolution{{{0, 0}}, {{1, 1}}, {{1, 1}}}, pa, pc);
         return p;
     }
 };
+
+template struct quant_conv_1<migraphx::shape::int8_type>;
+template struct quant_conv_1<migraphx::shape::fp8e4m3fnuz_type>;

test/verify/quant_conv_1d.cpp

@@ -27,15 +27,16 @@
 #include <migraphx/generate.hpp>
 #include <migraphx/make_op.hpp>
 
-struct quant_conv_1d : verify_program<quant_conv_1d>
+template <migraphx::shape::type_t DType>
+struct quant_conv_1d : verify_program<quant_conv_1d<DType>>
 {
     migraphx::program create_program() const
     {
         migraphx::program p;
         auto* mm = p.get_main_module();
-        migraphx::shape a_shape{migraphx::shape::int8_type, {2, 3, 4}};
+        migraphx::shape a_shape{DType, {2, 3, 4}};
         auto pa = mm->add_parameter("a", a_shape);
-        migraphx::shape c_shape{migraphx::shape::int8_type, {2, 3, 3}};
+        migraphx::shape c_shape{DType, {2, 3, 3}};
         auto pc = mm->add_parameter("c", c_shape);
         mm->add_instruction(
             migraphx::make_op("quant_convolution",
@@ -45,3 +46,7 @@ struct quant_conv_1d : verify_program<quant_conv_1d>
         return p;
     }
 };
+
+template struct quant_conv_1d<migraphx::shape::int8_type>;
+// MLIR 1D convolution is not supported in MIGraphX yet. Enable this through MIOpen route later.
+// template struct quant_conv_1d<migraphx::shape::fp8e4m3fnuz_type>;

test/verify/quant_conv_2.cpp

@@ -27,17 +27,21 @@
 #include <migraphx/generate.hpp>
 #include <migraphx/op/quant_convolution.hpp>
 
-struct quant_conv_2 : verify_program<quant_conv_2>
+template <migraphx::shape::type_t DType>
+struct quant_conv_2 : verify_program<quant_conv_2<DType>>
 {
     migraphx::program create_program() const
     {
         migraphx::program p;
         auto* mm = p.get_main_module();
-        migraphx::shape a_shape{migraphx::shape::int8_type, {16, 16, 4, 4}};
+        migraphx::shape a_shape{DType, {16, 16, 4, 4}};
         auto pa = mm->add_parameter("a", a_shape);
-        migraphx::shape c_shape{migraphx::shape::int8_type, {16, 16, 3, 3}};
+        migraphx::shape c_shape{DType, {16, 16, 3, 3}};
         auto pc = mm->add_parameter("c", c_shape);
         mm->add_instruction(migraphx::op::quant_convolution{{{0, 0}}, {{1, 1}}, {{1, 1}}}, pa, pc);
         return p;
     }
 };
+
+template struct quant_conv_2<migraphx::shape::int8_type>;
+template struct quant_conv_2<migraphx::shape::fp8e4m3fnuz_type>;