New try

pystencils/cpu/cpujit.py

@@ -494,19 +494,22 @@ def run_compile_step(command):
 
 
 class ExtensionModuleCode:
-    def __init__(self, module_name='generated', custom_backend=None, generated_code=None):
+    def __init__(self, module_name='generated', custom_backend=None):
         self.module_name = module_name
 
         self._ast_nodes = []
         self._function_names = []
         self._custom_backend = custom_backend
-        self._generated_code = generated_code
+        self.code_string = str()
+        self._code_hash = None
 
     def add_function(self, ast, name=None):
         self._ast_nodes.append(ast)
         self._function_names.append(name if name is not None else ast.function_name)
 
-    def write_to_file(self, restrict_qualifier, function_prefix, file):
+    def create_code_string(self, restrict_qualifier, function_prefix):
+        self.code_string = str()
+
         headers = {'<math.h>', '<stdint.h>'}
         for ast in self._ast_nodes:
             headers.update(get_headers(ast))
@@ -515,22 +518,28 @@ class ExtensionModuleCode:
         header_list.insert(0, '"Python.h"')
 
         includes = "\n".join(["#include %s" % (include_file,) for include_file in header_list])
-        print(includes, file=file)
-        print("\n", file=file)
-        print(f"#define RESTRICT {restrict_qualifier}", file=file)
-        print(f"#define FUNC_PREFIX {function_prefix}", file=file)
-        print("\n", file=file)
+        self.code_string += includes
+        self.code_string += "\n"
+        self.code_string += f"#define RESTRICT {restrict_qualifier} \n"
+        self.code_string += f"#define FUNC_PREFIX {function_prefix}"
+        self.code_string += "\n"
 
         for ast, name in zip(self._ast_nodes, self._function_names):
             old_name = ast.function_name
             ast.function_name = "kernel_" + name
-            if self._generated_code:
-                print(self._generated_code, file=file)
-            else:
-                print(generate_c(ast, custom_backend=self._custom_backend), file=file)
-            print(create_function_boilerplate_code(ast.get_parameters(), name, ast), file=file)
+            self.code_string += generate_c(ast, custom_backend=self._custom_backend)
+            self.code_string += create_function_boilerplate_code(ast.get_parameters(), name, ast)
             ast.function_name = old_name
-        print(create_module_boilerplate_code(self.module_name, self._function_names), file=file)
+
+        self._code_hash = "mod_" + hashlib.sha256(self.code_string.encode()).hexdigest()
+        self.code_string += create_module_boilerplate_code(self._code_hash, self._function_names)
+
+    def get_hash_of_code(self):
+        return self._code_hash
+
+    def write_to_file(self, file):
+        assert self.code_string, "The code must be generated first"
+        print(self.code_string, file=file)
 
 
 def compile_module(code, code_hash, base_dir):
@@ -554,7 +563,7 @@ def compile_module(code, code_hash, base_dir):
 
     if not os.path.exists(object_file):
         with file_handle_for_atomic_write(src_file) as f:
-            code.write_to_file(compiler_config['restrict_qualifier'], function_prefix, f)
+            code.write_to_file(f)
 
         if windows:
             compile_cmd = ['cl.exe', '/c', '/EHsc'] + compiler_config['flags'].split()
@@ -588,15 +597,15 @@ def compile_module(code, code_hash, base_dir):
 def compile_and_load(ast, custom_backend=None):
     cache_config = get_cache_config()
 
-    generated_code = generate_c(ast, dialect='c', custom_backend=custom_backend)
-    fields_accessed = str(ast.fields_accessed)
+    compiler_config = get_compiler_config()
+    function_prefix = '__declspec(dllexport)' if compiler_config['os'].lower() == 'windows' else ''
 
-    # Also the information of the field size should be contained in the hash string. Due to padding the generated code
-    # can look similar for different field sizes.
-    code_hash_str = "mod_" + hashlib.sha256((generated_code + fields_accessed).encode()).hexdigest()
-    code = ExtensionModuleCode(module_name=code_hash_str, custom_backend=custom_backend, generated_code=generated_code)
+    code = ExtensionModuleCode(custom_backend=custom_backend)
     code.add_function(ast, ast.function_name)
 
+    code.create_code_string(compiler_config['restrict_qualifier'], function_prefix)
+    code_hash_str = code.get_hash_of_code()
+
     if cache_config['object_cache'] is False:
         with TemporaryDirectory() as base_dir:
             lib_file = compile_module(code, code_hash_str, base_dir)

pystencils_tests/test_conditional_vec.py

@@ -5,7 +5,6 @@ import pytest
 import pystencils as ps
 from pystencils.astnodes import Block, Conditional
 from pystencils.backends.simd_instruction_sets import get_supported_instruction_sets
-from pystencils.backends.x86_instruction_sets import get_vector_instruction_set_x86
 from pystencils.cpu.vectorization import vec_all, vec_any
 
 
@@ -29,7 +28,7 @@ def test_vec_any():
     kernel = ast.compile()
     kernel(data=data_arr)
 
-    width = get_vector_instruction_set_x86(instruction_set=instruction_set)['width']
+    width = ast.instruction_set['width']
 
     np.testing.assert_equal(data_arr[3:9, 0:width], 2.0)
 

pystencils_tests/test_vectorization_specific.py

@@ -4,8 +4,7 @@ import numpy as np
 import sympy as sp
 
 import pystencils as ps
-from pystencils.backends.simd_instruction_sets import get_supported_instruction_sets
-from pystencils.backends.x86_instruction_sets import get_vector_instruction_set_x86
+from pystencils.backends.simd_instruction_sets import get_supported_instruction_sets, get_vector_instruction_set
 from pystencils.data_types import cast_func, VectorType
 
 supported_instruction_sets = get_supported_instruction_sets() if get_supported_instruction_sets() else []
@@ -57,7 +56,7 @@ def test_vectorized_abs(instruction_set, dtype):
 @pytest.mark.parametrize('instruction_set', supported_instruction_sets)
 def test_alignment_and_correct_ghost_layers(instruction_set, dtype):
     itemsize = 8 if dtype == 'double' else 4
-    alignment = get_vector_instruction_set_x86(dtype, instruction_set)['width'] * itemsize
+    alignment = get_vector_instruction_set(dtype, instruction_set)['width'] * itemsize
     dtype = np.float64 if dtype == 'double' else np.float32
 
     domain_size = (128, 128)