diff --git a/aten/src/ATen/cpu/vec/vec_base.h b/aten/src/ATen/cpu/vec/vec_base.h
index 472a8a6b7846..fca2b2c824e9 100644
--- a/aten/src/ATen/cpu/vec/vec_base.h
+++ b/aten/src/ATen/cpu/vec/vec_base.h
@@ -31,7 +31,6 @@
 #include <c10/util/BFloat16.h>
 #include <c10/util/BFloat16-math.h>
 #include <c10/util/copysign.h>
-#include <c10/util/math_compat.h>
 #include <ATen/native/cpu/zmath.h>
 #include <c10/util/TypeCast.h>
 #include <c10/macros/Macros.h>
diff --git a/aten/src/ATen/native/Distributions.cpp b/aten/src/ATen/native/Distributions.cpp
index c183beab157f..4d4eb2efaf40 100644
--- a/aten/src/ATen/native/Distributions.cpp
+++ b/aten/src/ATen/native/Distributions.cpp
@@ -4,7 +4,6 @@
 #include <ATen/TensorIterator.h>
 #include <ATen/TensorOperators.h>
 #include <c10/util/Exception.h>
-#include <c10/util/math_compat.h>
 #include <c10/util/Optional.h>
 
 #include <ATen/CPUGeneratorImpl.h>
diff --git a/aten/src/ATen/native/Math.h b/aten/src/ATen/native/Math.h
index f944a518ed63..092ee00992e9 100644
--- a/aten/src/ATen/native/Math.h
+++ b/aten/src/ATen/native/Math.h
@@ -6,7 +6,6 @@
 #include <c10/util/BFloat16.h>
 #include <c10/util/Half.h>
 #include <c10/util/MathConstants.h>
-#include <c10/util/math_compat.h>
 #include <cfloat>
 #include <cmath>
 #include <cstdint>
diff --git a/aten/src/ATen/native/cpu/UnaryOpsKernel.cpp b/aten/src/ATen/native/cpu/UnaryOpsKernel.cpp
index 8771af2bd0d2..2522c57a2d62 100644
--- a/aten/src/ATen/native/cpu/UnaryOpsKernel.cpp
+++ b/aten/src/ATen/native/cpu/UnaryOpsKernel.cpp
@@ -18,7 +18,6 @@
 #include <ATen/native/cpu/zmath.h>
 #include <ATen/OpMathType.h>
 
-#include <c10/util/math_compat.h>
 #include <c10/util/MathConstants.h>
 #include <c10/core/Scalar.h>
 #include <c10/util/TypeSafeSignMath.h>
diff --git a/aten/src/ATen/native/cpu/zmath.h b/aten/src/ATen/native/cpu/zmath.h
index 3f3971e6e76e..9b52039e84f9 100644
--- a/aten/src/ATen/native/cpu/zmath.h
+++ b/aten/src/ATen/native/cpu/zmath.h
@@ -2,7 +2,6 @@
 
 // Complex number math operations that act as no-ops for other dtypes.
 #include <c10/util/complex.h>
-#include <c10/util/math_compat.h>
 #include <c10/util/MathConstants.h>
 #include<ATen/NumericUtils.h>
 
diff --git a/aten/src/ATen/native/quantized/cpu/AdaptiveAveragePooling.cpp b/aten/src/ATen/native/quantized/cpu/AdaptiveAveragePooling.cpp
index 1317817902cf..74476e0a80ae 100644
--- a/aten/src/ATen/native/quantized/cpu/AdaptiveAveragePooling.cpp
+++ b/aten/src/ATen/native/quantized/cpu/AdaptiveAveragePooling.cpp
@@ -16,7 +16,6 @@
 #endif
 
 #include <c10/util/irange.h>
-#include <c10/util/math_compat.h>
 
 #include <algorithm>
 #include <cmath>
diff --git a/aten/src/ATen/native/quantized/cpu/AveragePool2d.cpp b/aten/src/ATen/native/quantized/cpu/AveragePool2d.cpp
index bb72a2010ca3..640c2eeab564 100644
--- a/aten/src/ATen/native/quantized/cpu/AveragePool2d.cpp
+++ b/aten/src/ATen/native/quantized/cpu/AveragePool2d.cpp
@@ -18,7 +18,6 @@
 #endif
 
 #include <c10/util/irange.h>
-#include <c10/util/math_compat.h>
 
 #include <algorithm>
 #include <cmath>
diff --git a/aten/src/ATen/native/quantized/cpu/AveragePool3d.cpp b/aten/src/ATen/native/quantized/cpu/AveragePool3d.cpp
index 93534b70c2c0..91a0b8229bf6 100644
--- a/aten/src/ATen/native/quantized/cpu/AveragePool3d.cpp
+++ b/aten/src/ATen/native/quantized/cpu/AveragePool3d.cpp
@@ -14,8 +14,6 @@
 #include <ATen/ops/avg_pool3d_native.h>
 #endif
 
-#include <c10/util/math_compat.h>
-
 #include <vector>
 
 namespace at {
diff --git a/c10/util/BFloat16-math.h b/c10/util/BFloat16-math.h
index fcb46044a34a..88a6b849d37b 100644
--- a/c10/util/BFloat16-math.h
+++ b/c10/util/BFloat16-math.h
@@ -2,7 +2,6 @@
 
 #include <c10/util/BFloat16.h>
 #include <c10/util/Half.h>
-#include <c10/util/math_compat.h>
 
 C10_CLANG_DIAGNOSTIC_PUSH()
 #if C10_CLANG_HAS_WARNING("-Wimplicit-float-conversion")
diff --git a/c10/util/complex_math.cpp b/c10/util/complex_math.cpp
index 4029133c33be..886aadb14151 100644
--- a/c10/util/complex_math.cpp
+++ b/c10/util/complex_math.cpp
@@ -1,5 +1,4 @@
 #include <c10/util/complex.h>
-#include <c10/util/math_compat.h>
 
 #include <cmath>
 
diff --git a/c10/util/copysign.h b/c10/util/copysign.h
index 1707fc47d3f9..e77b7e61b0ae 100644
--- a/c10/util/copysign.h
+++ b/c10/util/copysign.h
@@ -2,7 +2,6 @@
 
 #include <c10/util/BFloat16.h>
 #include <c10/util/Half.h>
-#include <c10/util/math_compat.h>
 
 namespace c10 {
 
diff --git a/torch/csrc/jit/runtime/register_ops_utils.h b/torch/csrc/jit/runtime/register_ops_utils.h
index 2a269931afbc..e1798cd3abcc 100644
--- a/torch/csrc/jit/runtime/register_ops_utils.h
+++ b/torch/csrc/jit/runtime/register_ops_utils.h
@@ -32,7 +32,6 @@
 #include <c10/core/thread_pool.h>
 #include <c10/util/SmallVector.h>
 #include <c10/util/irange.h>
-#include <c10/util/math_compat.h>
 #include <c10/util/string_utils.h>
 
 namespace torch::jit {
diff --git a/torch/csrc/jit/tensorexpr/eval.h b/torch/csrc/jit/tensorexpr/eval.h
index 4950dfbc1983..64ac1edf8f18 100644
--- a/torch/csrc/jit/tensorexpr/eval.h
+++ b/torch/csrc/jit/tensorexpr/eval.h
@@ -9,7 +9,6 @@
 
 #include <c10/macros/Macros.h>
 #include <c10/util/Logging.h>
-#include <c10/util/math_compat.h>
 #include <c10/util/string_utils.h>
 #include <torch/csrc/jit/tensorexpr/codegen.h>
 #include <torch/csrc/jit/tensorexpr/exceptions.h>