[CPU][ARM64] Add JIT emitter for Eltwise Less operation

* Added a jit_less_emitter derived class in aarch64/jit_eltwise_emitters * Created entry Algorithm::EltwiseLess in the get_supported_precisions in nodes/kernels/aarch64 * Add the EltwiseLess entry in the aarch64 executors supported algorithms Closes: openvinotoolkit#24415 Signed-off-by: Nashez Zubair <[email protected]>
nashez · Nov 21, 2024 · 073d0ed · 073d0ed
1 parent 9d36703
commit 073d0ed
Show file tree

Hide file tree

Showing 4 changed files with 84 additions and 0 deletions.
diff --git a/src/plugins/intel_cpu/src/emitters/plugin/aarch64/jit_eltwise_emitters.cpp b/src/plugins/intel_cpu/src/emitters/plugin/aarch64/jit_eltwise_emitters.cpp
@@ -1135,6 +1135,59 @@ void jit_is_nan_emitter::register_table_entries() {
     push_arg_entry_of("zero", 0x00000000, true);
 }
 
+/// LESS ///
+jit_less_emitter::jit_less_emitter(dnnl::impl::cpu::aarch64::jit_generator* host,
+                                   dnnl::impl::cpu::aarch64::cpu_isa_t host_isa,
+                                   const std::shared_ptr<ov::Node>& node)
+        : jit_emitter(host, host_isa, node, get_arithmetic_binary_exec_precision(node)) {
+    prepare_table();
+}
+
+jit_less_emitter::jit_less_emitter(dnnl::impl::cpu::aarch64::jit_generator* host,
+                                   dnnl::impl::cpu::aarch64::cpu_isa_t host_isa,
+                                   const ov::element::Type exec_prc)
+    : jit_emitter(host, host_isa, exec_prc) {
+    prepare_table();
+}
+
+size_t jit_less_emitter::get_inputs_count() const { return 2; }
+
+size_t jit_less_emitter::get_aux_vecs_count() const { return 1; }
+
+size_t jit_less_emitter::get_aux_gprs_count() const { return 1; }
+
+void jit_less_emitter::emit_impl(const std::vector<size_t> &in_vec_idxs, const std::vector<size_t> &out_vec_idxs) const {
+    if (host_isa_ == dnnl::impl::cpu::aarch64::asimd) {
+        emit_isa<dnnl::impl::cpu::aarch64::asimd>(in_vec_idxs, out_vec_idxs);
+    } else {
+        OV_CPU_JIT_EMITTER_THROW("Can't create jit eltwise kernel");
+    }
+}
+
+template <dnnl::impl::cpu::aarch64::cpu_isa_t isa>
+void jit_less_emitter::emit_isa(const std::vector<size_t> &in_vec_idxs, const std::vector<size_t> &out_vec_idxs) const {
+    OV_CPU_JIT_EMITTER_ASSERT(exec_prc_ == ov::element::f32, "unsupported precision: " + exec_prc_.to_string());
+
+    using TReg = typename dnnl::impl::cpu::aarch64::cpu_isa_traits<isa>::TReg;
+    const TReg src1 = TReg(in_vec_idxs[0]);
+    const TReg src2 = TReg(in_vec_idxs[1]);
+    const TReg dst = TReg(out_vec_idxs[0]);
+    const TReg aux = TReg(aux_vec_idxs[0]);
+
+    h->fcmge(dst.s, src1.s, src2.s);
+    h->not_(dst.b16, dst.b16);
+    h->ld1r(aux.s, table_val2("one"));
+    h->and_(dst.b16, dst.b16, aux.b16);
+}
+
+void jit_less_emitter::register_table_entries() {
+    push_arg_entry_of("one", 0x3f800000, true);
+}
+
+std::set<std::vector<element::Type>> jit_less_emitter::get_supported_precisions(const std::shared_ptr<ov::Node>& node) {
+    return {{element::f32, element::f32}};
+}
+
 /// LESS_EQUAL ///
 jit_less_equal_emitter::jit_less_equal_emitter(dnnl::impl::cpu::aarch64::jit_generator* host,
                                                dnnl::impl::cpu::aarch64::cpu_isa_t host_isa,

diff --git a/src/plugins/intel_cpu/src/emitters/plugin/aarch64/jit_eltwise_emitters.hpp b/src/plugins/intel_cpu/src/emitters/plugin/aarch64/jit_eltwise_emitters.hpp
@@ -521,6 +521,34 @@ class jit_is_inf_emitter : public jit_emitter {
     bool detect_positive;
 };
 
+class jit_less_emitter : public jit_emitter {
+public:
+    jit_less_emitter(dnnl::impl::cpu::aarch64::jit_generator *host,
+                     dnnl::impl::cpu::aarch64::cpu_isa_t host_isa,
+                     const ov::element::Type exec_prc = ov::element::f32);
+
+    jit_less_emitter(dnnl::impl::cpu::aarch64::jit_generator *host,
+                     dnnl::impl::cpu::aarch64::cpu_isa_t host_isa,
+                     const std::shared_ptr<ov::Node>& n);
+
+    size_t get_inputs_count() const override;
+
+    size_t get_aux_vecs_count() const override;
+
+    size_t get_aux_gprs_count() const override;
+
+    static std::set<std::vector<element::Type>> get_supported_precisions(
+            const std::shared_ptr<ov::Node>& node = nullptr);
+
+private:
+    void emit_impl(const std::vector<size_t> &in_vec_idxs, const std::vector<size_t> &out_vec_idxs) const override;
+
+    template <dnnl::impl::cpu::aarch64::cpu_isa_t isa>
+    void emit_isa(const std::vector<size_t> &in_vec_idxs, const std::vector<size_t> &out_vec_idxs) const;
+
+    void register_table_entries() override;
+};
+
 class jit_less_equal_emitter : public jit_emitter {
 public:
     jit_less_equal_emitter(dnnl::impl::cpu::aarch64::jit_generator *host,

diff --git a/src/plugins/intel_cpu/src/nodes/executors/aarch64/jit_eltwise.cpp b/src/plugins/intel_cpu/src/nodes/executors/aarch64/jit_eltwise.cpp
@@ -34,6 +34,7 @@ bool JitEltwiseExecutor::isSupported(
                                      Algorithm::EltwiseIsFinite,
                                      Algorithm::EltwiseIsInf,
                                      Algorithm::EltwiseIsNaN,
+                                     Algorithm::EltwiseLess,
                                      Algorithm::EltwiseLessEqual,
                                      Algorithm::EltwiseLogicalNot,
                                      Algorithm::EltwiseLogicalXor,

diff --git a/src/plugins/intel_cpu/src/nodes/kernels/aarch64/jit_uni_eltwise_generic.cpp b/src/plugins/intel_cpu/src/nodes/kernels/aarch64/jit_uni_eltwise_generic.cpp
@@ -651,6 +651,7 @@ std::shared_ptr<jit_emitter> jit_uni_eltwise_generic<isa>::create_eltwise_emitte
     OV_CASE(Algorithm::EltwiseHswish, ov::intel_cpu::aarch64::jit_hswish_emitter),
     OV_CASE(Algorithm::EltwiseIsFinite, ov::intel_cpu::aarch64::jit_is_finite_emitter),
     OV_CASE(Algorithm::EltwiseIsInf, ov::intel_cpu::aarch64::jit_is_inf_emitter),
+    OV_CASE(Algorithm::EltwiseLess, ov::intel_cpu::aarch64::jit_less_emitter),
     OV_CASE(Algorithm::EltwiseLessEqual, ov::intel_cpu::aarch64::jit_less_equal_emitter),
     OV_CASE(Algorithm::EltwiseLogicalNot, ov::intel_cpu::aarch64::jit_logical_not_emitter),
     OV_CASE(Algorithm::EltwiseLogicalXor, ov::intel_cpu::aarch64::jit_logical_xor_emitter),
@@ -836,6 +837,7 @@ std::set<std::vector<element::Type>> eltwise_precision_helper::get_supported_pre
         OV_CASE(Algorithm::EltwiseIsFinite, jit_is_finite_emitter),
         OV_CASE(Algorithm::EltwiseIsInf, jit_is_inf_emitter),
         OV_CASE(Algorithm::EltwiseIsNaN, jit_is_nan_emitter),
+        OV_CASE(Algorithm::EltwiseLess, jit_less_emitter),
         OV_CASE(Algorithm::EltwiseLessEqual, jit_less_equal_emitter),
         OV_CASE(Algorithm::EltwiseLogicalNot, jit_logical_not_emitter),
         OV_CASE(Algorithm::EltwiseLogicalXor, jit_logical_xor_emitter),