fix: Treat equivalent pooling qinfo as non-requantized

arm_compute/core/QuantizationInfo.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2025 Arm Limited.
+ * Copyright (c) 2019-2026 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -203,6 +203,17 @@ class QuantizationInfo
  */
 inline bool operator==(const QuantizationInfo &lhs, const QuantizationInfo &rhs)
 {
+    const bool lhs_is_uniform = lhs.scale().size() < 2 && lhs.offset().size() < 2;
+    const bool rhs_is_uniform = rhs.scale().size() < 2 && rhs.offset().size() < 2;
+
+    if (lhs_is_uniform && rhs_is_uniform)
+    {
+        const auto lhs_qinfo = lhs.uniform();
+        const auto rhs_qinfo = rhs.uniform();
+
+        return (lhs_qinfo.scale == rhs_qinfo.scale) && (lhs_qinfo.offset == rhs_qinfo.offset);
+    }
+
     return (lhs.scale() == rhs.scale()) && (lhs.offset() == rhs.offset());
 }
 

src/cpu/kernels/internal/CpuPool2dAssemblyWrapperKernel.cpp

@@ -132,7 +132,7 @@ CpuPool2dAssemblyWrapperKernel::validate(const ITensorInfo *src, const ITensorIn
         const auto src_qinfo = src->quantization_info().uniform();
         const auto dst_qinfo = dst->quantization_info().uniform();
 
-        if (src_qinfo != dst_qinfo)
+        if (src->quantization_info() != dst->quantization_info())
         {
             ARM_COMPUTE_RETURN_ERROR_ON_MSG(info.pool_type != PoolingType::MAX,
                                             "Assembly kernels only support differing src/dst quantization info for "

tests/validation/UNIT/QuantizationInfo.cpp

@@ -0,0 +1,80 @@
+/*
+ * Copyright (c) 2026 Arm Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * 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 "arm_compute/core/QuantizationInfo.h"
+
+#include "tests/framework/Asserts.h"
+#include "tests/framework/Macros.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+TEST_SUITE(UNIT)
+TEST_SUITE(QuantizationInfo)
+
+TEST_CASE(EquivalentUniformImplicitAndExplicitZeroOffset, framework::DatasetMode::ALL)
+{
+    const arm_compute::QuantizationInfo implicit_zero(1.0f);
+    const arm_compute::QuantizationInfo explicit_zero(1.0f, 0);
+
+    ARM_COMPUTE_EXPECT(implicit_zero == explicit_zero, framework::LogLevel::ERRORS);
+    ARM_COMPUTE_EXPECT(!(implicit_zero != explicit_zero), framework::LogLevel::ERRORS);
+}
+
+TEST_CASE(DifferentUniformQInfoCompareDifferent, framework::DatasetMode::ALL)
+{
+    const arm_compute::QuantizationInfo unit_scale(1.0f);
+    const arm_compute::QuantizationInfo half_scale(0.5f, 0);
+    const arm_compute::QuantizationInfo offset_one(1.0f, 1);
+    const arm_compute::QuantizationInfo offset_zero(1.0f, 0);
+
+    ARM_COMPUTE_EXPECT(unit_scale != half_scale, framework::LogLevel::ERRORS);
+    ARM_COMPUTE_EXPECT(offset_one != offset_zero, framework::LogLevel::ERRORS);
+}
+
+TEST_CASE(PerChannelQInfoUsesStrictVectorEquality, framework::DatasetMode::ALL)
+{
+    const arm_compute::QuantizationInfo matching_per_channel(std::vector<float>{1.0f, 2.0f});
+    const arm_compute::QuantizationInfo same_per_channel(std::vector<float>{1.0f, 2.0f});
+    const arm_compute::QuantizationInfo different_per_channel(std::vector<float>{1.0f, 3.0f});
+
+    ARM_COMPUTE_EXPECT(matching_per_channel == same_per_channel, framework::LogLevel::ERRORS);
+    ARM_COMPUTE_EXPECT(matching_per_channel != different_per_channel, framework::LogLevel::ERRORS);
+}
+
+TEST_CASE(PerChannelQInfoDoesNotMatchUniformQInfo, framework::DatasetMode::ALL)
+{
+    const arm_compute::QuantizationInfo per_channel(std::vector<float>{1.0f, 2.0f});
+    const arm_compute::QuantizationInfo uniform(1.0f);
+
+    ARM_COMPUTE_EXPECT(per_channel != uniform, framework::LogLevel::ERRORS);
+}
+
+TEST_SUITE_END() // QuantizationInfo
+TEST_SUITE_END() // UNIT
+} // namespace validation
+} // namespace test
+} // namespace arm_compute