diff --git a/kernels/portable/cpu/op_var_mean.cpp b/kernels/portable/cpu/op_var_mean.cpp
index 6e72b268610..dbb88f2f32d 100644
--- a/kernels/portable/cpu/op_var_mean.cpp
+++ b/kernels/portable/cpu/op_var_mean.cpp
@@ -35,30 +35,65 @@ void compute_var_mean(
       mean_data[out_ix] = NAN;
     }
   } else if (in.numel() > 0) {
-    MapReduceOverDimListPlan plan(in, dim_list);
-    const bool success = parallel_for_each_reduce_over_dim_list_output_index(
-        in, dim_list, var_out, [&](const auto begin, const auto end) {
-          for (const auto out_ix : c10::irange(begin, end)) {
-            // Pass 1: compute sum -> mean
-            CTYPE_OUT sum = plan.execute<CTYPE_IN, CTYPE_OUT>(
-                [](CTYPE_IN v) { return static_cast<CTYPE_OUT>(v); },
-                [](CTYPE_OUT outv, CTYPE_OUT acc) { return acc + outv; },
-                out_ix);
-            CTYPE_OUT mean = sum / static_cast<CTYPE_OUT>(num);
-            mean_data[out_ix] = mean;
-            // Pass 2: compute sum of squared deviations
-            CTYPE_OUT sum2 = plan.execute<CTYPE_IN, CTYPE_OUT>(
-                [mean](CTYPE_IN v) {
-                  return (
-                      (static_cast<CTYPE_OUT>(v) - mean) *
-                      (static_cast<CTYPE_OUT>(v) - mean));
-                },
-                [](CTYPE_OUT outv, CTYPE_OUT acc) { return acc + outv; },
-                out_ix);
-            var_data[out_ix] = sum2 / denominator;
+    // Fast path: contiguous tensor, single innermost dim reduction, same dtype.
+    bool used_fast_path = false;
+    if (dim_list.has_value() && dim_list.value().size() == 1 &&
+        in.scalar_type() == var_out.scalar_type()) {
+      const int64_t d = dim_list.value()[0] < 0 ? dim_list.value()[0] + in.dim()
+                                                : dim_list.value()[0];
+      if (d >= 0 && d < in.dim() && d == in.dim() - 1 &&
+          tensor_is_contiguous(in)) {
+        used_fast_path = true;
+        const int64_t reduce_size = in.size(d);
+        const int64_t outer_size = in.numel() / reduce_size;
+        const CTYPE_OUT cnum = static_cast<CTYPE_OUT>(num);
+        const CTYPE_OUT cdenom = static_cast<CTYPE_OUT>(denominator);
+        const CTYPE_IN* in_data = in.const_data_ptr<CTYPE_IN>();
+        for (int64_t i = 0; i < outer_size; i++) {
+          const CTYPE_IN* row = in_data + i * reduce_size;
+          // Pass 1: compute mean
+          CTYPE_OUT sum = 0;
+          for (int64_t j = 0; j < reduce_size; j++) {
+            sum += static_cast<CTYPE_OUT>(row[j]);
           }
-        });
-    ET_KERNEL_CHECK_MSG(ctx, success, Internal, , "parallel_for failed");
+          CTYPE_OUT mean = sum / cnum;
+          mean_data[i] = mean;
+          // Pass 2: compute variance
+          CTYPE_OUT sum2 = 0;
+          for (int64_t j = 0; j < reduce_size; j++) {
+            CTYPE_OUT diff = static_cast<CTYPE_OUT>(row[j]) - mean;
+            sum2 += diff * diff;
+          }
+          var_data[i] = sum2 / cdenom;
+        }
+      }
+    }
+    if (!used_fast_path) {
+      MapReduceOverDimListPlan plan(in, dim_list);
+      const bool success = parallel_for_each_reduce_over_dim_list_output_index(
+          in, dim_list, var_out, [&](const auto begin, const auto end) {
+            for (const auto out_ix : c10::irange(begin, end)) {
+              // Pass 1: compute sum -> mean
+              CTYPE_OUT sum = plan.execute<CTYPE_IN, CTYPE_OUT>(
+                  [](CTYPE_IN v) { return static_cast<CTYPE_OUT>(v); },
+                  [](CTYPE_OUT outv, CTYPE_OUT acc) { return acc + outv; },
+                  out_ix);
+              CTYPE_OUT mean = sum / static_cast<CTYPE_OUT>(num);
+              mean_data[out_ix] = mean;
+              // Pass 2: compute sum of squared deviations
+              CTYPE_OUT sum2 = plan.execute<CTYPE_IN, CTYPE_OUT>(
+                  [mean](CTYPE_IN v) {
+                    return (
+                        (static_cast<CTYPE_OUT>(v) - mean) *
+                        (static_cast<CTYPE_OUT>(v) - mean));
+                  },
+                  [](CTYPE_OUT outv, CTYPE_OUT acc) { return acc + outv; },
+                  out_ix);
+              var_data[out_ix] = sum2 / denominator;
+            }
+          });
+      ET_KERNEL_CHECK_MSG(ctx, success, Internal, , "parallel_for failed");
+    } // !used_fast_path
   }
 }