diff --git a/yateto/arch.py b/yateto/arch.py
index 334af1b..12a6e25 100644
--- a/yateto/arch.py
+++ b/yateto/arch.py
@@ -132,6 +132,7 @@ def getArchitectureIdentifiedBy(ident):
     'a64fx': Architecture(name, precision, 64, True),
     'neon': Architecture(name, precision, 16, False),
     'apple-m1': Architecture(name, precision, 16, False),
+    'apple-m2': Architecture(name, precision, 16, False),
     'sve128': Architecture(name, precision, 16, False),
     'sve256': Architecture(name, precision, 32, False),
     'sve512': Architecture(name, precision, 64, False),
diff --git a/yateto/codegen/gemm/gemmgen.py b/yateto/codegen/gemm/gemmgen.py
index 04d6264..1537802 100644
--- a/yateto/codegen/gemm/gemmgen.py
+++ b/yateto/codegen/gemm/gemmgen.py
@@ -399,7 +399,7 @@ def _kernel(self, routine_name):
     #flags += ["LIBXSMM_GEMM_FLAG_ALIGN_C"]
     libxsmm_flag_str = " | ".join(flags)
 
-    prefetch_flag =  "LIBXSMM_GEMM_PREFETCH_SIGONLY" if not self._arch.enablePrefetch else "LIBXSMM_GEMM_PREFETCH_BL2_VIA_C"
+    prefetch_flag =  "LIBXSMM_GEMM_PREFETCH_NONE" if not self._arch.enablePrefetch else "LIBXSMM_GEMM_PREFETCH_BL2_VIA_C"
 
     kernel_var_name = f'{routine_name}_var'
     return """
diff --git a/yateto/gemm_configuration.py b/yateto/gemm_configuration.py
index d3fbd64..54dc1ed 100644
--- a/yateto/gemm_configuration.py
+++ b/yateto/gemm_configuration.py
@@ -162,7 +162,7 @@ def preference(self, m, n, k, sparseA, sparseB, transA, transB, alpha, beta, ali
     return Preference.LOW
 
   def _archSupported(self):
-    supported_set = {'noarch', 'wsm', 'snb', 'hsw', 'skx', 'knc', 'knl', 'naples', 'rome', 'milan', 'bergamo', "a64fx", "thunderx2t99", 'neon', 'sve128', 'sve256', 'sve512', 'apple-m1'}
+    supported_set = {'noarch', 'wsm', 'snb', 'hsw', 'skx', 'knc', 'knl', 'naples', 'rome', 'milan', 'bergamo', "a64fx", "thunderx2t99", 'neon', 'sve128', 'sve256', 'sve512', 'apple-m1', "apple-m2"}
 
     if self._arch.name.lower() in supported_set:
       return True
@@ -210,7 +210,7 @@ def __init__(self, arch, cmd: str = 'pspamm.py', threshold: int = 128):
     self._threshold = threshold
 
   def _archSupported(self):
-    supported_set = {'thunderx2t99', 'knl', 'skx', 'a64fx', 'hsw', 'naples', 'rome', 'milan', 'bergamo', 'neon', 'sve128', 'sve256', 'sve512', 'sve1024', 'sve2048', 'apple-m1'}
+    supported_set = {'thunderx2t99', 'knl', 'skx', 'a64fx', 'hsw', 'naples', 'rome', 'milan', 'bergamo', 'neon', 'sve128', 'sve256', 'sve512', 'sve1024', 'sve2048', 'apple-m1', 'apple-m2'}
     if self._arch.name.lower() in supported_set:
       return True
     else:
@@ -303,6 +303,8 @@ def __init__(self, arch):
       'knl' : [libxsmm_jit, libxsmm, pspamm, mkl, blis, eigen],
       'skx' : [libxsmm_jit, libxsmm, pspamm, mkl, blis, eigen],
       'thunderx2t99' : [libxsmm_jit, pspamm, openblas, blis, eigen],
+      'apple-m1' : [libxsmm_jit, pspamm, openblas, blis, eigen],
+      'apple-m2' : [libxsmm_jit, pspamm, openblas, blis, eigen],
       'a64fx' : [libxsmm_jit, pspamm, openblas, blis, eigen],
       'neon' : [libxsmm_jit, pspamm, openblas, blis, eigen],
       'sve128' : [libxsmm_jit, pspamm, openblas, blis, eigen],
diff --git a/yateto/util.py b/yateto/util.py
index 9a28d20..108ecf1 100644
--- a/yateto/util.py
+++ b/yateto/util.py
@@ -27,7 +27,7 @@ def create_collection(matrices):
 def tensor_from_constant_expression(name: str,
                                     expression,
                                     target_indices: Indices = None,
-                                    dtype: dtype = np.float128,
+                                    dtype: dtype = np.longdouble,
                                     tensor_args: dict = dict()):
   """
   Computes the result of an expression and returns
@@ -55,7 +55,7 @@ def tensor_collection_from_constant_expression(base_name: str,
                                                expressions,
                                                group_indices,
                                                target_indices: Indices = None,
-                                               dtype=np.float128,
+                                               dtype=np.longdouble,
                                                tensor_args: dict = {}):
   """
   Computes the result of an expression group and returns