diff --git a/dnn/src/x86/conv_bias/opr_impl.cpp b/dnn/src/x86/conv_bias/opr_impl.cpp
index 3244b47c613d628c5a0d4860017061da6ac0f6f0..a71d55819862ef3bdc98f689dd29ac4b66c68358 100644
--- a/dnn/src/x86/conv_bias/opr_impl.cpp
+++ b/dnn/src/x86/conv_bias/opr_impl.cpp
@@ -185,7 +185,7 @@ SmallVector<AlgoCategory> ConvBiasImpl::suggest_algo_category_order(
     }
     //! conv1x1
     im2col_prefer |= (FH == 1 && FW == 1);
-    //! x86 8x8x16 not optmized, so it will use fallback im2col+matmul
+    //! x86 8x8x16 not optimized, so it will use fallback im2col+matmul
     if (param.deduce_algo_data_type() == AlgoDataType::INT8X8X16) {
         im2col_prefer = true;
     }
diff --git a/imperative/python/megengine/functional/debug_param.py b/imperative/python/megengine/functional/debug_param.py
index 08bace859c4e3377ad7c6e17645f9de018984f65..7467eb73187c508a32d7d785a2a2714a8f3d8a5c 100644
--- a/imperative/python/megengine/functional/debug_param.py
+++ b/imperative/python/megengine/functional/debug_param.py
@@ -40,7 +40,7 @@ def set_execution_strategy(option):
         * HEURISTIC uses heuristic to choose the fastest algorithm.
         * PROFILE runs possible algorithms on real device to find the best one.
         * REPRODUCIBLE uses the algorithms that is reproducible.
-        * OPTMIZED uses the algorithms that is optimized.
+        * OPTIMIZED uses the algorithms that is optimized.
 
         The default strategy is HEURISTIC, this options can be combined to
         form a combination option, e.g. PROFILE | REPRODUCIBLE
diff --git a/imperative/python/src/graph_rt.cpp b/imperative/python/src/graph_rt.cpp
index 240bbdbf6c3e9f20ef1d130bf3df34087ae73939..228d69ab5ae1384d53b9dcf01ecc78eb7b7fad59 100644
--- a/imperative/python/src/graph_rt.cpp
+++ b/imperative/python/src/graph_rt.cpp
@@ -266,7 +266,7 @@ void init_graph_rt(py::module m) {
                 {"HEURISTIC", [&]() { stg = _AlgoStrategy::HEURISTIC; }},
                 {"PROFILE", [&]() { stg = _AlgoStrategy::PROFILE; }},
                 {"REPRODUCIBLE", [&]() { stg = _AlgoStrategy::REPRODUCIBLE; }},
-                {"OPTMIZED", [&]() { stg = _AlgoStrategy::OPTMIZED; }},
+                {"OPTIMIZED", [&]() { stg = _AlgoStrategy::OPTIMIZED; }},
         };
         auto it = m.find(strategy);
         mgb_assert(it != m.end(), "Invalid strategy string!");
diff --git a/sdk/load-and-run/src/mgblar.cpp b/sdk/load-and-run/src/mgblar.cpp
index e51ffa6a25b7977c6b52cac54ed60278c425bbc1..87b3a19210ecd8d187d277bb86d7f5495a2a57c7 100644
--- a/sdk/load-and-run/src/mgblar.cpp
+++ b/sdk/load-and-run/src/mgblar.cpp
@@ -709,7 +709,7 @@ void run_test_st(Args &env) {
             strategy = S::PROFILE;
         }
     } else if (env.use_fast_run) {
-        strategy = S::PROFILE | S::OPTMIZED;
+        strategy = S::PROFILE | S::OPTIMIZED;
     } else if (env.reproducible) {
         strategy = S::HEURISTIC | S::REPRODUCIBLE;
     }
diff --git a/src/gopt/test/inference.cpp b/src/gopt/test/inference.cpp
index 08c3b0a1c36dda86e53483f3f97a4ae13a96bdda..70518421cc9dd7266aa771c0d4b1b524deb6764b 100644
--- a/src/gopt/test/inference.cpp
+++ b/src/gopt/test/inference.cpp
@@ -1756,8 +1756,8 @@ TEST(TestGoptInference, FastProfileCache) {
     using S = opr::Convolution::ExecutionPolicy::Strategy;
     ASSERT_EQ(S::HEURISTIC, conv.execution_policy_transient().strategy);
     gopt::modify_opr_algo_strategy_inplace({z + 2.3f},
-                                           S::PROFILE | S::OPTMIZED);
-    ASSERT_EQ(S::PROFILE | S::OPTMIZED, conv.execution_policy().strategy);
+                                           S::PROFILE | S::OPTIMIZED);
+    ASSERT_EQ(S::PROFILE | S::OPTIMIZED, conv.execution_policy().strategy);
 }
 
 TEST(TestGoptInference, AlgoWorkspaceLimit) {
diff --git a/src/opr/impl/search_policy/algo_chooser.cpp b/src/opr/impl/search_policy/algo_chooser.cpp
index 9f712c4040913b84fe0af16e3f4786356dd91972..97b554c1d2a4ba166556ef8206f2cf8ef866c4cf 100644
--- a/src/opr/impl/search_policy/algo_chooser.cpp
+++ b/src/opr/impl/search_policy/algo_chooser.cpp
@@ -283,7 +283,7 @@ std::vector<megdnn::Algorithm::SearchItem> flatten_search_space(
 static bool algo_attribute_match_strategy(AlgoAttribute attribute,
                                           ExecutionStrategy selected_strategy) {
     bool ret = true;
-    if (selected_strategy & ExecutionStrategy::OPTMIZED) {
+    if (selected_strategy & ExecutionStrategy::OPTIMIZED) {
         ret &= (!static_cast<bool>(AlgoAttribute::NAIVE & attribute));
     } else if (selected_strategy & ExecutionStrategy::REPRODUCIBLE) {
         ret &= static_cast<bool>(AlgoAttribute::REPRODUCIBLE & attribute);
diff --git a/src/opr/test/dnn/convolution.cpp b/src/opr/test/dnn/convolution.cpp
index 7736975d2191bc77323cc7362aed004c347020d6..ccd3aebcd535b7722f1749eac99339f6d9ca49bb 100644
--- a/src/opr/test/dnn/convolution.cpp
+++ b/src/opr/test/dnn/convolution.cpp
@@ -357,7 +357,7 @@ TEST(TestOprDNN, ConvBiasExePolicy) {
 #if MGB_ENABLE_FASTRUN
     for (auto strategy :
          SmallVector<S>{S::PROFILE, S::HEURISTIC, S::PROFILE | S::REPRODUCIBLE,
-          S::PROFILE | S::HEURISTIC, S::PROFILE | S::OPTMIZED}) {
+          S::PROFILE | S::HEURISTIC, S::PROFILE | S::OPTIMIZED}) {
 #else
     for (auto strategy :
          SmallVector<S>{S : HEURISTIC, S::PROFILE | S::HEURISTIC}) {
@@ -444,7 +444,7 @@ TEST(TestOprDNN, ConvolutionExePolicy) {
 #if MGB_ENABLE_FASTRUN
     for (auto strategy :
          SmallVector<S>{S::PROFILE, S::HEURISTIC, S::PROFILE | S::REPRODUCIBLE,
-          S::PROFILE | S::HEURISTIC, S::PROFILE | S::OPTMIZED}) {
+          S::PROFILE | S::HEURISTIC, S::PROFILE | S::OPTIMIZED}) {
 #else
     for (auto strategy :
          SmallVector<S>{S : HEURISTIC, S::PROFILE | S::HEURISTIC}) {
@@ -1717,7 +1717,7 @@ TEST(TestOprDNN, LocalShareForwardExecPolicy) {
 #if MGB_ENABLE_FASTRUN
     for (auto strategy :
          SmallVector<S>{S::PROFILE, S::HEURISTIC, S::PROFILE | S::REPRODUCIBLE,
-          S::PROFILE | S::HEURISTIC, S::PROFILE | S::OPTMIZED}) {
+          S::PROFILE | S::HEURISTIC, S::PROFILE | S::OPTIMIZED}) {
 #else
     for (auto strategy :
          SmallVector<S>{S : HEURISTIC, S::PROFILE | S::HEURISTIC}) {
@@ -1828,7 +1828,7 @@ TEST(TestOprDNN, DeformableConvForward) {
 #if MGB_ENABLE_FASTRUN
     for (auto strategy :
          SmallVector<S>{S::PROFILE, S::HEURISTIC, S::PROFILE | S::REPRODUCIBLE,
-          S::PROFILE | S::HEURISTIC, S::PROFILE | S::OPTMIZED}) {
+          S::PROFILE | S::HEURISTIC, S::PROFILE | S::OPTIMIZED}) {
 #else
     for (auto strategy :
          SmallVector<S>{S : HEURISTIC, S::PROFILE | S::HEURISTIC}) {
@@ -1997,7 +1997,7 @@ TEST(TestOprDNN, BatchConvBiasForward) {
 #if MGB_ENABLE_FASTRUN
     for (auto strategy :
          SmallVector<S>{S::PROFILE, S::HEURISTIC, S::PROFILE | S::REPRODUCIBLE,
-          S::PROFILE | S::HEURISTIC, S::PROFILE | S::OPTMIZED}) {
+          S::PROFILE | S::HEURISTIC, S::PROFILE | S::OPTIMIZED}) {
 #else
     for (auto strategy :
          SmallVector<S>{S : HEURISTIC, S::PROFILE | S::HEURISTIC}) {
diff --git a/tools/param_defs/mgb_opr_param_defs.py b/tools/param_defs/mgb_opr_param_defs.py
index cd46edd2a82fba2b1e9b83335b14b040fee7c513..b32d2087d157b3af9a4f0e2108c08ccdbdf6882e 100644
--- a/tools/param_defs/mgb_opr_param_defs.py
+++ b/tools/param_defs/mgb_opr_param_defs.py
@@ -41,12 +41,13 @@ pdef('PersistentOutputStorage').add_fields(
           Doc('REPRODUCIBLE',
               'when profile or heuristic algo selection it require the algos'
               'must be reproducible'),
-          Doc('OPTMIZED',
+          Doc('OPTIMIZED',
               'profile require algos are optmized to achieve fast-profile'),
           default=('HEURISTIC',),
           member_alias=[(('HEURISTIC', 'REPRODUCIBLE'), 'HEURISTIC_REPRODUCIBLE'),
                         (('PROFILE', 'REPRODUCIBLE'), 'PROFILE_REPRODUCIBLE'),
                         (('PROFILE', 'HEURISTIC'), 'PROFILE_HEURISTIC'),
+                        (('OPTIMIZED',), 'OPTMIZED'),
                         ]).
  add_fields('uint64',
             Doc('workspace_limit', 'workspace limit in bytes'),