doc/notebooks/demo_kerncraft.ipynb

+%% Cell type:code id: tags:
+
+``` python
+%matplotlib notebook
+import matplotlib.pyplot as plt
+from pystencils.session import *
+import pystencils
+```
+
+%% Cell type:code id: tags:
+
+``` python
+src_arr = np.zeros([1024,1024], dtype=np.float64)
+dst_arr = np.zeros_like(src_arr)
+
+dst, src = ps.fields(dst=dst_arr, src=src_arr)
+```
+
+%% Cell type:code id: tags:
+
+``` python
+grad_x, grad_y = sp.symbols("grad_x, grad_y")
+
+symbolic_description = [
+    ps.Assignment(dst[0, 0], (src[1, 0] + src[-1, 0] + src[0, 1] + src[0, -1]) / 4)
+]
+kernel = ps.create_kernel(symbolic_description)
+symbolic_description
+```
+
+%% Output
+
+
+    $\displaystyle \left[ {{dst}_{(0,0)}} \leftarrow \frac{{{src}_{(-1,0)}}}{4} + \frac{{{src}_{(0,-1)}}}{4} + \frac{{{src}_{(0,1)}}}{4} + \frac{{{src}_{(1,0)}}}{4}\right]$
+    ⎡         src_W   src_S   src_N   src_E⎤
+    ⎢dst_C := ───── + ───── + ───── + ─────⎥
+    ⎣           4       4       4       4  ⎦
+
+%% Cell type:code id: tags:
+
+``` python
+@ps.kernel
+def symbolic_description_using_function():
+    dst[0, 0] @= (src[1, 0] + src[-1, 0] + src[0, 1] + src[0, -1]) / 4
+symbolic_description_using_function
+```
+
+%% Output
+
+
+    $\displaystyle \left[ {{dst}_{(0,0)}} \leftarrow \frac{{{src}_{(-1,0)}}}{4} + \frac{{{src}_{(0,-1)}}}{4} + \frac{{{src}_{(0,1)}}}{4} + \frac{{{src}_{(1,0)}}}{4}\right]$
+    ⎡         src_W   src_S   src_N   src_E⎤
+    ⎢dst_C := ───── + ───── + ───── + ─────⎥
+    ⎣           4       4       4       4  ⎦
+
+%% Cell type:code id: tags:
+
+``` python
+kernel = ps.create_kernel(symbolic_description_using_function)
+ps.show_code(kernel)
+```
+
+%% Output
+
+
+
+%% Cell type:code id: tags:
+
+``` python
+src_arr[:,0] = 1
+src_arr[0,:] = 1
+compiled_kernel = kernel.compile()
+
+plt.imshow(src_arr)
+plt.colorbar();
+```
+
+%% Output
+
+
+
+%% Cell type:code id: tags:
+
+``` python
+for _ in range(1000):
+    compiled_kernel(src=src_arr, dst=dst_arr)
+    src_arr, dst_arr = dst_arr, src_arr
+
+plt.imshow(src_arr)
+plt.colorbar();
+```
+
+%% Output
+
+
+
+%% Cell type:code id: tags:
+
+``` python
+from pystencils import kerncraft_coupling as kc
+import kerncraft
+```
+
+%% Cell type:code id: tags:
+
+``` python
+analyzed = kc.PyStencilsKerncraftKernel(kernel)
+```
+
+%% Cell type:code id: tags:
+
+``` python
+print(analyzed.as_code())
+```
+
+%% Output
+
+    
+    #include "kerncraft.h"
+    #include <stdlib.h>
+    #include <stdint.h>
+    #include <stdbool.h>
+    #include <math.h>
+    #include <stdio.h>
+    
+    
+    
+    #define RESTRICT __restrict__
+    #define FUNC_PREFIX
+    void dummy(void *);
+    void timing(double* wcTime, double* cpuTime);
+    extern int var_false;
+    
+    
+    FUNC_PREFIX void kernel(double * RESTRICT _data_dst, double * RESTRICT const _data_src)
+    {
+       for (int ctr_0 = 1; ctr_0 < 1023; ctr_0 += 1)
+       {
+          double * RESTRICT _data_dst_00 = _data_dst + 1024*ctr_0;
+          double * RESTRICT _data_src_01 = _data_src + 1024*ctr_0 + 1024;
+          double * RESTRICT _data_src_00 = _data_src + 1024*ctr_0;
+          double * RESTRICT _data_src_0m1 = _data_src + 1024*ctr_0 - 1024;
+          for (int ctr_1 = 1; ctr_1 < 1023; ctr_1 += 1)
+          {
+             _data_dst_00[ctr_1] = 0.25*_data_src_00[ctr_1 + 1] + 0.25*_data_src_00[ctr_1 - 1] + 0.25*_data_src_01[ctr_1] + 0.25*_data_src_0m1[ctr_1];
+          }
+       }
+    }
+    
+    
+    int main(int argc, char **argv)
+    {
+    
+      // Initialization dst
+      double * dst = (double *) aligned_malloc(sizeof(double) * 1048576, 64);
+      for (unsigned long long i = 0; i < 1048576; ++i)
+        dst[i] = 0.23;
+    
+      if(var_false)
+        dummy(dst);
+    
+      // Initialization src
+      double * src = (double *) aligned_malloc(sizeof(double) * 1048576, 64);
+      for (unsigned long long i = 0; i < 1048576; ++i)
+        src[i] = 0.23;
+    
+      if(var_false)
+        dummy(src);
+    
+      for(int warmup = 1; warmup >= 0; --warmup) {
+        int repeat = 2;
+        if(warmup == 0) {
+          repeat = atoi(argv[1]);
+        }
+    
+        for (; repeat > 0; --repeat)
+        {
+          kernel(dst,src);
+    
+          // Dummy calls
+          if(var_false) dummy((void*)dst);
+          if(var_false) dummy((void*)src);
+        }
+    
+      }
+    }
+
+%% Cell type:code id: tags:
+
+``` python
+analyzed.array_sizes()
+```
+
+%% Output
+
+    {'src': 1048576, 'dst': 1048576}
+
+%% Cell type:code id: tags:
+
+``` python
+list(analyzed.get_loop_stack())
+```
+
+%% Output
+
+    [{'index': 'ctr_0', 'start': 1, 'stop': 1023, 'increment': 1},
+     {'index': 'ctr_1', 'start': 1, 'stop': 1023, 'increment': 1}]
+
+%% Cell type:code id: tags:
+
+``` python
+analyzed.print_kernel_info()
+```
+
+%% Output
+
+    loop stack:             idx |        min        max       step
+                       ---------+---------------------------------
+                          ctr_0 |          1       1023          1
+                          ctr_1 |          1       1023          1
+    
+    
+    data sources:          name |  offsets   ...
+                       ---------+------------...
+                            src | [ctr_0, ctr_1 - 1]
+                                | [ctr_0 + 1, ctr_1]
+                                | [ctr_0 - 1, ctr_1]
+                                | [ctr_0, ctr_1 + 1]
+    
+    
+    data destinations:     name |  offsets   ...
+                       ---------+------------...
+                            dst | [ctr_0, ctr_1]
+    
+    
+    FLOPs:      op | count
+               ----+-------
+                 + |    3
+                 * |    4
+                    =======
+                        7
+    
+
+%% Cell type:code id: tags:
+
+``` python
+analyzed.print_constants_info()
+```
+
+%% Output
+
+    constants:     name | value
+               ---------+-----------
+    
+    
+
+%% Cell type:code id: tags:
+
+``` python
+analyzed.print_variables_info()
+```
+
+%% Output
+
+    variables:     name |   type          size
+               ---------+----------------------------------
+                    src |           float (1024, 1024)
+                    dst |           float (1024, 1024)
+    
+    
+
+%% Cell type:code id: tags:
+
+``` python
+machine = kerncraft.machinemodel.MachineModel(path_to_yaml="some_machine.yaml")
+model = kerncraft.models.ECMData(analyzed, machine, kc.KerncraftParameters())
+```
+
+%% Cell type:code id: tags:
+
+``` python
+model.analyze()
+```
+
+%% Output
+
+    {'cycles': [('L2', 3.0255005100102004),
+      ('L3', 6.0516410328206565),
+      ('MEM', 0.0)],
+     'misses': [2.023880477609552, 2.023880477609552, 0.0, 0.0],
+     'hits': [62.97805956119122, 0.0, 2.023880477609552, 0.0],
+     'evicts': [1.0016200324006481, 1.001940038800776, 0.0, 0.0],
+     'verbose infos': {'memory hierarchy': [{'index': 0,
+        'level': 'L1',
+        'total loads': 320.1241624832497,
+        'total misses': 68.21192423848477,
+        'total hits': 251.9122382447649,
+        'total stores': 64.0,
+        'total evicts': 64.10368207364148,
+        'total lines load': 65.00194003880078,
+        'total lines misses': 2.023880477609552,
+        'total lines hits': 62.97805956119122,
+        'total lines stores': 16.0,
+        'total lines evicts': 1.0016200324006481,
+        'cycles': None},
+       {'index': 1,
+        'level': 'L2',
+        'total loads': 129.52835056701133,
+        'total misses': 129.52835056701133,
+        'total hits': 0.0,
+        'total stores': 64.10368207364148,
+        'total evicts': 64.12416248324966,
+        'total lines load': 2.023880477609552,
+        'total lines misses': 2.023880477609552,
+        'total lines hits': 0.0,
+        'total lines stores': 1.0016200324006481,
+        'total lines evicts': 1.001940038800776,
+        'cycles': None},
+       {'index': 2,
+        'level': 'L3',
+        'total loads': 129.52835056701133,
+        'total misses': 0.0,
+        'total hits': 129.52835056701133,
+        'total stores': 64.12416248324966,
+        'total evicts': 0.0,
+        'total lines load': 2.023880477609552,
+        'total lines misses': 0.0,
+        'total lines hits': 2.023880477609552,
+        'total lines stores': 1.001940038800776,
+        'total lines evicts': 0.0,
+        'cycles': None},
+       {'index': 3,
+        'level': 'MEM',
+        'total loads': 0.0,
+        'total misses': 0.0,
+        'total hits': 0.0,
+        'total stores': 0.0,
+        'total evicts': 0.0,
+        'total lines load': 0.0,
+        'total lines misses': 0.0,
+        'total lines hits': 0.0,
+        'total lines stores': 0.0,
+        'total lines evicts': 0.0,
+        'cycles': None}],
+      'cache stats': [{'name': 'L1',
+        'LOAD_count': 406254,
+        'LOAD_byte': 2000736,
+        'STORE_count': 99998,
+        'STORE_byte': 399992,
+        'HIT_count': 393605,
+        'HIT_byte': 1574420,
+        'MISS_count': 12649,
+        'MISS_byte': 426316,
+        'EVICT_count': 6260,
+        'EVICT_byte': 400640},
+       {'name': 'L2',
+        'LOAD_count': 12649,
+        'LOAD_byte': 809536,
+        'STORE_count': 6260,
+        'STORE_byte': 400640,
+        'HIT_count': 0,
+        'HIT_byte': 0,
+        'MISS_count': 12649,
+        'MISS_byte': 809536,
+        'EVICT_count': 6262,
+        'EVICT_byte': 400768},
+       {'name': 'L3',
+        'LOAD_count': 12649,
+        'LOAD_byte': 809536,
+        'STORE_count': 6262,
+        'STORE_byte': 400768,
+        'HIT_count': 12649,
+        'HIT_byte': 809536,
+        'MISS_count': 0,
+        'MISS_byte': 0,
+        'EVICT_count': 0,
+        'EVICT_byte': 0},
+       {'name': 'MEM',
+        'LOAD_count': 0,
+        'LOAD_byte': 0,
+        'HIT_count': 0,
+        'HIT_byte': 0,
+        'STORE_count': 0,
+        'STORE_byte': 0,
+        'EVICT_count': 0,
+        'EVICT_byte': 0,
+        'MISS_count': 0,
+        'MISS_byte': 0}],
+      'cachelines in stats': 6249.875},
+     'iterations per cacheline': 16,
+     'L2': 3.0255005100102004,
+     'L3': 6.0516410328206565,
+     'memory bandwidth kernel': 'load',
+     'memory bandwidth': PrefixedUnit(61.82, 'G', 'B/s'),
+     'MEM': 0.0,
+     'flops per iteration': 7}
+
+%% Cell type:code id: tags:
+
+``` python
+model.report()
+```
+
+%% Output
+
+    L2 = 3.03 cy/CL
+    L3 = 6.05 cy/CL
+    MEM = 0.00 cy/CL
+
+%% Cell type:code id: tags:
+
+``` python
+```

pystencils/kerncraft_coupling/kerncraft_interface.py

@@ -96,7 +96,8 @@ class PyStencilsKerncraftKernel(KernelCode):
         for field in fields_accessed:
             layout = get_layout_tuple(field)
             permuted_shape = list(field.shape[i] for i in layout)
-            self.set_variable(field.name, str(field.dtype), tuple(permuted_shape))
+            self.set_variable(field.name, ('double',) if str(field.dtype) == 'float64' else
+                    ('float',), tuple(permuted_shape))
 
         # Scalars may be safely ignored
         # for param in ast.get_parameters():