cumo 0.1.0 → 0.1.1

data/ext/cumo/narray/gen/tmpl/store_array_kernel.cu

@@ -29,30 +29,30 @@ __global__ void <%="cumo_#{c_iter}_stride_scalar_kernel"%>(char *p1, ssize_t s1,
 
 void <%="cumo_#{c_iter}_index_kernel_launch"%>(char *p1, size_t *idx1, dtype* z, uint64_t n)
 {
-    size_t gridDim = get_gridDim(n);
-    size_t blockDim = get_blockDim(n);
-    <%="cumo_#{c_iter}_index_kernel"%><<<gridDim, blockDim>>>(p1,idx1,z,n);
+    size_t grid_dim = cumo_get_grid_dim(n);
+    size_t block_dim = cumo_get_block_dim(n);
+    <%="cumo_#{c_iter}_index_kernel"%><<<grid_dim, block_dim>>>(p1,idx1,z,n);
 }
 
 void <%="cumo_#{c_iter}_stride_kernel_launch"%>(char *p1, ssize_t s1, dtype* z, uint64_t n)
 {
-    size_t gridDim = get_gridDim(n);
-    size_t blockDim = get_blockDim(n);
-    <%="cumo_#{c_iter}_stride_kernel"%><<<gridDim, blockDim>>>(p1,s1,z,n);
+    size_t grid_dim = cumo_get_grid_dim(n);
+    size_t block_dim = cumo_get_block_dim(n);
+    <%="cumo_#{c_iter}_stride_kernel"%><<<grid_dim, block_dim>>>(p1,s1,z,n);
 }
 
 void <%="cumo_#{c_iter}_index_scalar_kernel_launch"%>(char *p1, size_t *idx1, dtype z, uint64_t n)
 {
-    size_t gridDim = get_gridDim(n);
-    size_t blockDim = get_blockDim(n);
-    <%="cumo_#{c_iter}_index_scalar_kernel"%><<<gridDim, blockDim>>>(p1,idx1,z,n);
+    size_t grid_dim = cumo_get_grid_dim(n);
+    size_t block_dim = cumo_get_block_dim(n);
+    <%="cumo_#{c_iter}_index_scalar_kernel"%><<<grid_dim, block_dim>>>(p1,idx1,z,n);
 }
 
 void <%="cumo_#{c_iter}_stride_scalar_kernel_launch"%>(char *p1, ssize_t s1, dtype z, uint64_t n)
 {
-    size_t gridDim = get_gridDim(n);
-    size_t blockDim = get_blockDim(n);
-    <%="cumo_#{c_iter}_stride_scalar_kernel"%><<<gridDim, blockDim>>>(p1,s1,z,n);
+    size_t grid_dim = cumo_get_grid_dim(n);
+    size_t block_dim = cumo_get_block_dim(n);
+    <%="cumo_#{c_iter}_stride_scalar_kernel"%><<<grid_dim, block_dim>>>(p1,s1,z,n);
 }
 
 <% end %>

data/ext/cumo/narray/gen/tmpl/store_bit.c

@@ -1,56 +1,56 @@
 //<% unless c_iter.include? 'robject' %>
-void <%="cumo_#{c_iter}_index_index_kernel_launch"%>(char *p1, size_t p2, BIT_DIGIT *a2, size_t *idx1, size_t *idx2, uint64_t n);
-void <%="cumo_#{c_iter}_stride_index_kernel_launch"%>(char *p1, size_t p2, BIT_DIGIT *a2, ssize_t s1, size_t *idx2, uint64_t n);
-void <%="cumo_#{c_iter}_index_stride_kernel_launch"%>(char *p1, size_t p2, BIT_DIGIT *a2, size_t *idx1, ssize_t s2, uint64_t n);
-void <%="cumo_#{c_iter}_stride_stride_kernel_launch"%>(char *p1, size_t p2, BIT_DIGIT *a2, ssize_t s1, ssize_t s2, uint64_t n);
+void <%="cumo_#{c_iter}_index_index_kernel_launch"%>(char *p1, size_t p2, CUMO_BIT_DIGIT *a2, size_t *idx1, size_t *idx2, uint64_t n);
+void <%="cumo_#{c_iter}_stride_index_kernel_launch"%>(char *p1, size_t p2, CUMO_BIT_DIGIT *a2, ssize_t s1, size_t *idx2, uint64_t n);
+void <%="cumo_#{c_iter}_index_stride_kernel_launch"%>(char *p1, size_t p2, CUMO_BIT_DIGIT *a2, size_t *idx1, ssize_t s2, uint64_t n);
+void <%="cumo_#{c_iter}_stride_stride_kernel_launch"%>(char *p1, size_t p2, CUMO_BIT_DIGIT *a2, ssize_t s1, ssize_t s2, uint64_t n);
 //<% end %>
 
 static void
-<%=c_iter%>(na_loop_t *const lp)
+<%=c_iter%>(cumo_na_loop_t *const lp)
 {
     size_t     i;
     char      *p1;
     size_t     p2;
     ssize_t    s1, s2;
     size_t    *idx1, *idx2;
-    BIT_DIGIT *a2;
+    CUMO_BIT_DIGIT *a2;
 
-    INIT_COUNTER(lp, i);
-    INIT_PTR_IDX(lp, 0, p1, s1, idx1);
-    INIT_PTR_BIT_IDX(lp, 1, a2, p2, s2, idx2);
+    CUMO_INIT_COUNTER(lp, i);
+    CUMO_INIT_PTR_IDX(lp, 0, p1, s1, idx1);
+    CUMO_INIT_PTR_BIT_IDX(lp, 1, a2, p2, s2, idx2);
 
     //<% if c_iter.include? 'robject' %>
     {
-        BIT_DIGIT x;
+        CUMO_BIT_DIGIT x;
         dtype y;
-        SHOW_SYNCHRONIZE_WARNING_ONCE("<%=name%>", "<%=type_name%>");
-        SHOW_SYNCHRONIZE_FIXME_WARNING_ONCE("<%=name%>", "<%=type_name%>");
+        CUMO_SHOW_SYNCHRONIZE_WARNING_ONCE("<%=name%>", "<%=type_name%>");
+        CUMO_SHOW_SYNCHRONIZE_FIXME_WARNING_ONCE("<%=name%>", "<%=type_name%>");
         if (idx2) {
             if (idx1) {
                 for (; i--;) {
-                    LOAD_BIT(a2, p2+*idx2, x); idx2++;
+                    CUMO_LOAD_BIT(a2, p2+*idx2, x); idx2++;
                     y = m_from_sint(x);
-                    SET_DATA_INDEX(p1,idx1,dtype,y);
+                    CUMO_SET_DATA_INDEX(p1,idx1,dtype,y);
                 }
             } else {
                 for (; i--;) {
-                    LOAD_BIT(a2, p2+*idx2, x); idx2++;
+                    CUMO_LOAD_BIT(a2, p2+*idx2, x); idx2++;
                     y = m_from_sint(x);
-                    SET_DATA_STRIDE(p1,s1,dtype,y);
+                    CUMO_SET_DATA_STRIDE(p1,s1,dtype,y);
                 }
             }
         } else {
             if (idx1) {
                 for (; i--;) {
-                    LOAD_BIT(a2, p2, x); p2 += s2;
+                    CUMO_LOAD_BIT(a2, p2, x); p2 += s2;
                     y = m_from_sint(x);
-                    SET_DATA_INDEX(p1,idx1,dtype,y);
+                    CUMO_SET_DATA_INDEX(p1,idx1,dtype,y);
                 }
             } else {
                 for (; i--;) {
-                    LOAD_BIT(a2, p2, x); p2 += s2;
+                    CUMO_LOAD_BIT(a2, p2, x); p2 += s2;
                     y = m_from_sint(x);
-                    SET_DATA_STRIDE(p1,s1,dtype,y);
+                    CUMO_SET_DATA_STRIDE(p1,s1,dtype,y);
                 }
             }
         }
@@ -78,9 +78,9 @@ static void
 static VALUE
 <%=c_func(:nodef)%>(VALUE self, VALUE obj)
 {
-    ndfunc_arg_in_t ain[2] = {{OVERWRITE,0},{Qnil,0}};
-    ndfunc_t ndf = {<%=c_iter%>, FULL_LOOP, 2,0, ain,0};
+    cumo_ndfunc_arg_in_t ain[2] = {{CUMO_OVERWRITE,0},{Qnil,0}};
+    cumo_ndfunc_t ndf = {<%=c_iter%>, CUMO_FULL_LOOP, 2,0, ain,0};
 
-    na_ndloop(&ndf, 2, self, obj);
+    cumo_na_ndloop(&ndf, 2, self, obj);
     return self;
 }

data/ext/cumo/narray/gen/tmpl/store_bit_kernel.cu

@@ -1,66 +1,66 @@
 <% unless c_iter.include? 'robject' %>
-__global__ void <%="cumo_#{c_iter}_index_index_kernel"%>(char *p1, size_t p2, BIT_DIGIT *a2, size_t *idx1, size_t *idx2, uint64_t n)
+__global__ void <%="cumo_#{c_iter}_index_index_kernel"%>(char *p1, size_t p2, CUMO_BIT_DIGIT *a2, size_t *idx1, size_t *idx2, uint64_t n)
 {
     for (uint64_t i = blockIdx.x * blockDim.x + threadIdx.x; i < n; i += blockDim.x * gridDim.x) {
-        BIT_DIGIT x;
-        LOAD_BIT(a2, p2 + idx2[i], x);
+        CUMO_BIT_DIGIT x;
+        CUMO_LOAD_BIT(a2, p2 + idx2[i], x);
         *(dtype*)(p1 + idx1[i]) = m_from_real(x);
     }
 }
 
-__global__ void <%="cumo_#{c_iter}_stride_index_kernel"%>(char *p1, size_t p2, BIT_DIGIT *a2, ssize_t s1, size_t *idx2, uint64_t n)
+__global__ void <%="cumo_#{c_iter}_stride_index_kernel"%>(char *p1, size_t p2, CUMO_BIT_DIGIT *a2, ssize_t s1, size_t *idx2, uint64_t n)
 {
     for (uint64_t i = blockIdx.x * blockDim.x + threadIdx.x; i < n; i += blockDim.x * gridDim.x) {
-        BIT_DIGIT x;
-        LOAD_BIT(a2, p2 + idx2[i], x);
+        CUMO_BIT_DIGIT x;
+        CUMO_LOAD_BIT(a2, p2 + idx2[i], x);
         *(dtype*)(p1 + (i * s1)) = m_from_real(x);
     }
 }
 
-__global__ void <%="cumo_#{c_iter}_index_stride_kernel"%>(char *p1, size_t p2, BIT_DIGIT *a2, size_t *idx1, ssize_t s2, uint64_t n)
+__global__ void <%="cumo_#{c_iter}_index_stride_kernel"%>(char *p1, size_t p2, CUMO_BIT_DIGIT *a2, size_t *idx1, ssize_t s2, uint64_t n)
 {
     for (uint64_t i = blockIdx.x * blockDim.x + threadIdx.x; i < n; i += blockDim.x * gridDim.x) {
-        BIT_DIGIT x;
-        LOAD_BIT(a2, p2 + (i * s2), x);
+        CUMO_BIT_DIGIT x;
+        CUMO_LOAD_BIT(a2, p2 + (i * s2), x);
         *(dtype*)(p1 + idx1[i]) = m_from_real(x);
     }
 }
 
-__global__ void <%="cumo_#{c_iter}_stride_stride_kernel"%>(char *p1, size_t p2, BIT_DIGIT *a2, ssize_t s1, ssize_t s2, uint64_t n)
+__global__ void <%="cumo_#{c_iter}_stride_stride_kernel"%>(char *p1, size_t p2, CUMO_BIT_DIGIT *a2, ssize_t s1, ssize_t s2, uint64_t n)
 {
     for (uint64_t i = blockIdx.x * blockDim.x + threadIdx.x; i < n; i += blockDim.x * gridDim.x) {
-        BIT_DIGIT x;
-        LOAD_BIT(a2, p2 + (i * s2), x);
+        CUMO_BIT_DIGIT x;
+        CUMO_LOAD_BIT(a2, p2 + (i * s2), x);
         *(dtype*)(p1 + (i * s1)) = m_from_real(x);
     }
 }
 
-void <%="cumo_#{c_iter}_index_index_kernel_launch"%>(char *p1, size_t p2, BIT_DIGIT *a2, size_t *idx1, size_t *idx2, uint64_t n)
+void <%="cumo_#{c_iter}_index_index_kernel_launch"%>(char *p1, size_t p2, CUMO_BIT_DIGIT *a2, size_t *idx1, size_t *idx2, uint64_t n)
 {
-    size_t gridDim = get_gridDim(n);
-    size_t blockDim = get_blockDim(n);
-    <%="cumo_#{c_iter}_index_index_kernel"%><<<gridDim, blockDim>>>(p1,p2,a2,idx1,idx2,n);
+    size_t grid_dim = cumo_get_grid_dim(n);
+    size_t block_dim = cumo_get_block_dim(n);
+    <%="cumo_#{c_iter}_index_index_kernel"%><<<grid_dim, block_dim>>>(p1,p2,a2,idx1,idx2,n);
 }
 
-void <%="cumo_#{c_iter}_stride_index_kernel_launch"%>(char *p1, size_t p2, BIT_DIGIT *a2, ssize_t s1, size_t *idx2, uint64_t n)
+void <%="cumo_#{c_iter}_stride_index_kernel_launch"%>(char *p1, size_t p2, CUMO_BIT_DIGIT *a2, ssize_t s1, size_t *idx2, uint64_t n)
 {
-    size_t gridDim = get_gridDim(n);
-    size_t blockDim = get_blockDim(n);
-    <%="cumo_#{c_iter}_stride_index_kernel"%><<<gridDim, blockDim>>>(p1,p2,a2,s1,idx2,n);
+    size_t grid_dim = cumo_get_grid_dim(n);
+    size_t block_dim = cumo_get_block_dim(n);
+    <%="cumo_#{c_iter}_stride_index_kernel"%><<<grid_dim, block_dim>>>(p1,p2,a2,s1,idx2,n);
 }
 
-void <%="cumo_#{c_iter}_index_stride_kernel_launch"%>(char *p1, size_t p2, BIT_DIGIT *a2, size_t *idx1, ssize_t s2, uint64_t n)
+void <%="cumo_#{c_iter}_index_stride_kernel_launch"%>(char *p1, size_t p2, CUMO_BIT_DIGIT *a2, size_t *idx1, ssize_t s2, uint64_t n)
 {
-    size_t gridDim = get_gridDim(n);
-    size_t blockDim = get_blockDim(n);
-    <%="cumo_#{c_iter}_index_stride_kernel"%><<<gridDim, blockDim>>>(p1,p2,a2,idx1,s2,n);
+    size_t grid_dim = cumo_get_grid_dim(n);
+    size_t block_dim = cumo_get_block_dim(n);
+    <%="cumo_#{c_iter}_index_stride_kernel"%><<<grid_dim, block_dim>>>(p1,p2,a2,idx1,s2,n);
 }
 
-void <%="cumo_#{c_iter}_stride_stride_kernel_launch"%>(char *p1, size_t p2, BIT_DIGIT *a2, ssize_t s1, ssize_t s2, uint64_t n)
+void <%="cumo_#{c_iter}_stride_stride_kernel_launch"%>(char *p1, size_t p2, CUMO_BIT_DIGIT *a2, ssize_t s1, ssize_t s2, uint64_t n)
 {
-    size_t gridDim = get_gridDim(n);
-    size_t blockDim = get_blockDim(n);
-    <%="cumo_#{c_iter}_stride_stride_kernel"%><<<gridDim, blockDim>>>(p1,p2,a2,s1,s2,n);
+    size_t grid_dim = cumo_get_grid_dim(n);
+    size_t block_dim = cumo_get_block_dim(n);
+    <%="cumo_#{c_iter}_stride_stride_kernel"%><<<grid_dim, block_dim>>>(p1,p2,a2,s1,s2,n);
 }
 
 <% end %>

data/ext/cumo/narray/gen/tmpl/store_from.c

@@ -6,46 +6,46 @@ void <%="cumo_#{c_iter}_stride_stride_kernel_launch"%>(char *p1, char *p2, ssize
 //<% end %>
 
 static void
-<%=c_iter%>(na_loop_t *const lp)
+<%=c_iter%>(cumo_na_loop_t *const lp)
 {
     size_t  i, s1, s2;
     char   *p1, *p2;
     size_t *idx1, *idx2;
 
-    INIT_COUNTER(lp, i);
-    INIT_PTR_IDX(lp, 0, p1, s1, idx1);
-    INIT_PTR_IDX(lp, 1, p2, s2, idx2);
+    CUMO_INIT_COUNTER(lp, i);
+    CUMO_INIT_PTR_IDX(lp, 0, p1, s1, idx1);
+    CUMO_INIT_PTR_IDX(lp, 1, p2, s2, idx2);
     //<% if c_iter.include? 'robject' %>
-    SHOW_SYNCHRONIZE_FIXME_WARNING_ONCE("<%=name%>", "<%=type_name%>");
+    CUMO_SHOW_SYNCHRONIZE_FIXME_WARNING_ONCE("<%=name%>", "<%=type_name%>");
     {
         <%=dtype%> x;
         dtype y;
         if (idx2) {
             if (idx1) {
                 for (; i--;) {
-                    GET_DATA_INDEX(p2,idx2,<%=dtype%>,x);
+                    CUMO_GET_DATA_INDEX(p2,idx2,<%=dtype%>,x);
                     y = <%=macro%>(x);
-                    SET_DATA_INDEX(p1,idx1,dtype,y);
+                    CUMO_SET_DATA_INDEX(p1,idx1,dtype,y);
                 }
             } else {
                 for (; i--;) {
-                    GET_DATA_INDEX(p2,idx2,<%=dtype%>,x);
+                    CUMO_GET_DATA_INDEX(p2,idx2,<%=dtype%>,x);
                     y = <%=macro%>(x);
-                    SET_DATA_STRIDE(p1,s1,dtype,y);
+                    CUMO_SET_DATA_STRIDE(p1,s1,dtype,y);
                 }
             }
         } else {
             if (idx1) {
                 for (; i--;) {
-                    GET_DATA_STRIDE(p2,s2,<%=dtype%>,x);
+                    CUMO_GET_DATA_STRIDE(p2,s2,<%=dtype%>,x);
                     y = <%=macro%>(x);
-                    SET_DATA_INDEX(p1,idx1,dtype,y);
+                    CUMO_SET_DATA_INDEX(p1,idx1,dtype,y);
                 }
             } else {
                 for (; i--;) {
-                    GET_DATA_STRIDE(p2,s2,<%=dtype%>,x);
+                    CUMO_GET_DATA_STRIDE(p2,s2,<%=dtype%>,x);
                     y = <%=macro%>(x);
-                    SET_DATA_STRIDE(p1,s1,dtype,y);
+                    CUMO_SET_DATA_STRIDE(p1,s1,dtype,y);
                 }
             }
         }
@@ -73,9 +73,9 @@ static void
 static VALUE
 <%=c_func(:nodef)%>(VALUE self, VALUE obj)
 {
-    ndfunc_arg_in_t ain[2] = {{OVERWRITE,0},{Qnil,0}};
-    ndfunc_t ndf = { <%=c_iter%>, FULL_LOOP, 2, 0, ain, 0 };
+    cumo_ndfunc_arg_in_t ain[2] = {{CUMO_OVERWRITE,0},{Qnil,0}};
+    cumo_ndfunc_t ndf = { <%=c_iter%>, CUMO_FULL_LOOP, 2, 0, ain, 0 };
 
-    na_ndloop(&ndf, 2, self, obj);
+    cumo_na_ndloop(&ndf, 2, self, obj);
     return self;
 }

data/ext/cumo/narray/gen/tmpl/store_from_kernel.cu

@@ -29,30 +29,30 @@ __global__ void <%="cumo_#{c_iter}_stride_stride_kernel"%>(char *p1, char *p2, s
 
 void <%="cumo_#{c_iter}_index_index_kernel_launch"%>(char *p1, char *p2, size_t *idx1, size_t *idx2, uint64_t n)
 {
-    size_t gridDim = get_gridDim(n);
-    size_t blockDim = get_blockDim(n);
-    <%="cumo_#{c_iter}_index_index_kernel"%><<<gridDim, blockDim>>>(p1,p2,idx1,idx2,n);
+    size_t grid_dim = cumo_get_grid_dim(n);
+    size_t block_dim = cumo_get_block_dim(n);
+    <%="cumo_#{c_iter}_index_index_kernel"%><<<grid_dim, block_dim>>>(p1,p2,idx1,idx2,n);
 }
 
 void <%="cumo_#{c_iter}_stride_index_kernel_launch"%>(char *p1, char *p2, ssize_t s1, size_t *idx2, uint64_t n)
 {
-    size_t gridDim = get_gridDim(n);
-    size_t blockDim = get_blockDim(n);
-    <%="cumo_#{c_iter}_stride_index_kernel"%><<<gridDim, blockDim>>>(p1,p2,s1,idx2,n);
+    size_t grid_dim = cumo_get_grid_dim(n);
+    size_t block_dim = cumo_get_block_dim(n);
+    <%="cumo_#{c_iter}_stride_index_kernel"%><<<grid_dim, block_dim>>>(p1,p2,s1,idx2,n);
 }
 
 void <%="cumo_#{c_iter}_index_stride_kernel_launch"%>(char *p1, char *p2, size_t *idx1, ssize_t s2, uint64_t n)
 {
-    size_t gridDim = get_gridDim(n);
-    size_t blockDim = get_blockDim(n);
-    <%="cumo_#{c_iter}_index_stride_kernel"%><<<gridDim, blockDim>>>(p1,p2,idx1,s2,n);
+    size_t grid_dim = cumo_get_grid_dim(n);
+    size_t block_dim = cumo_get_block_dim(n);
+    <%="cumo_#{c_iter}_index_stride_kernel"%><<<grid_dim, block_dim>>>(p1,p2,idx1,s2,n);
 }
 
 void <%="cumo_#{c_iter}_stride_stride_kernel_launch"%>(char *p1, char *p2, ssize_t s1, ssize_t s2, uint64_t n)
 {
-    size_t gridDim = get_gridDim(n);
-    size_t blockDim = get_blockDim(n);
-    <%="cumo_#{c_iter}_stride_stride_kernel"%><<<gridDim, blockDim>>>(p1,p2,s1,s2,n);
+    size_t grid_dim = cumo_get_grid_dim(n);
+    size_t block_dim = cumo_get_block_dim(n);
+    <%="cumo_#{c_iter}_stride_stride_kernel"%><<<grid_dim, block_dim>>>(p1,p2,s1,s2,n);
 }
 
 <% end %>

data/ext/cumo/narray/gen/tmpl/to_a.c

@@ -1,5 +1,5 @@
 static void
-<%=c_iter%>(na_loop_t *const lp)
+<%=c_iter%>(cumo_na_loop_t *const lp)
 {
     size_t i, s1;
     char *p1;
@@ -7,20 +7,20 @@ static void
     dtype x;
     volatile VALUE a, y;
 
-    INIT_COUNTER(lp, i);
-    INIT_PTR_IDX(lp, 0, p1, s1, idx1);
+    CUMO_INIT_COUNTER(lp, i);
+    CUMO_INIT_PTR_IDX(lp, 0, p1, s1, idx1);
     a = rb_ary_new2(i);
     rb_ary_push(lp->args[1].value, a);
-    //SHOW_SYNCHRONIZE_FIXME_WARNING_ONCE("<%=name%>", "<%=type_name%>");
+    //CUMO_SHOW_SYNCHRONIZE_FIXME_WARNING_ONCE("<%=name%>", "<%=type_name%>");
     if (idx1) {
         for (; i--;) {
-            GET_DATA_INDEX(p1,idx1,dtype,x);
+            CUMO_GET_DATA_INDEX(p1,idx1,dtype,x);
             y = m_data_to_num(x);
             rb_ary_push(a,y);
         }
     } else {
         for (; i--;) {
-            GET_DATA_STRIDE(p1,s1,dtype,x);
+            CUMO_GET_DATA_STRIDE(p1,s1,dtype,x);
             y = m_data_to_num(x);
             rb_ary_push(a,y);
         }
@@ -35,9 +35,9 @@ static void
 static VALUE
 <%=c_func(0)%>(VALUE self)
 {
-    ndfunc_arg_in_t ain[3] = {{Qnil,0},{sym_loop_opt},{sym_option}};
-    ndfunc_arg_out_t aout[1] = {{rb_cArray,0}}; // dummy?
-    ndfunc_t ndf = { <%=c_iter%>, FULL_LOOP_NIP, 3, 1, ain, aout };
+    cumo_ndfunc_arg_in_t ain[3] = {{Qnil,0},{cumo_sym_loop_opt},{cumo_sym_option}};
+    cumo_ndfunc_arg_out_t aout[1] = {{rb_cArray,0}}; // dummy?
+    cumo_ndfunc_t ndf = { <%=c_iter%>, CUMO_FULL_LOOP_NIP, 3, 1, ain, aout };
     cumo_cuda_runtime_check_status(cudaDeviceSynchronize());
-    return na_ndloop_cast_narray_to_rarray(&ndf, self, Qnil);
+    return cumo_na_ndloop_cast_narray_to_rarray(&ndf, self, Qnil);
 }

data/ext/cumo/narray/gen/tmpl/unary.c

@@ -8,47 +8,47 @@ void <%="cumo_#{c_iter}_contiguous_kernel_launch"%>(char *p1, char *p2, uint64_t
 <% end %>
 
 static void
-<%=c_iter%>(na_loop_t *const lp)
+<%=c_iter%>(cumo_na_loop_t *const lp)
 {
     size_t  n;
     char   *p1, *p2;
     ssize_t s1, s2;
     size_t *idx1, *idx2;
 
-    INIT_COUNTER(lp, n);
-    INIT_PTR_IDX(lp, 0, p1, s1, idx1);
-    INIT_PTR_IDX(lp, 1, p2, s2, idx2);
+    CUMO_INIT_COUNTER(lp, n);
+    CUMO_INIT_PTR_IDX(lp, 0, p1, s1, idx1);
+    CUMO_INIT_PTR_IDX(lp, 1, p2, s2, idx2);
 
     <% if type_name == 'robject' || name == 'map' %>
     {
         size_t i;
         dtype x;
-        SHOW_SYNCHRONIZE_FIXME_WARNING_ONCE("<%=name%>", "<%=type_name%>");
+        CUMO_SHOW_SYNCHRONIZE_FIXME_WARNING_ONCE("<%=name%>", "<%=type_name%>");
         if (idx1) {
             if (idx2) {
                 for (i=0; i<n; i++) {
-                    GET_DATA_INDEX(p1,idx1,dtype,x);
+                    CUMO_GET_DATA_INDEX(p1,idx1,dtype,x);
                     x = m_<%=name%>(x);
-                    SET_DATA_INDEX(p2,idx2,dtype,x);
+                    CUMO_SET_DATA_INDEX(p2,idx2,dtype,x);
                 }
             } else {
                 for (i=0; i<n; i++) {
-                    GET_DATA_INDEX(p1,idx1,dtype,x);
+                    CUMO_GET_DATA_INDEX(p1,idx1,dtype,x);
                     x = m_<%=name%>(x);
-                    SET_DATA_STRIDE(p2,s2,dtype,x);
+                    CUMO_SET_DATA_STRIDE(p2,s2,dtype,x);
                 }
             }
         } else {
             if (idx2) {
                 for (i=0; i<n; i++) {
-                    GET_DATA_STRIDE(p1,s1,dtype,x);
+                    CUMO_GET_DATA_STRIDE(p1,s1,dtype,x);
                     x = m_<%=name%>(x);
-                    SET_DATA_INDEX(p2,idx2,dtype,x);
+                    CUMO_SET_DATA_INDEX(p2,idx2,dtype,x);
                 }
             } else {
                 //<% if need_align %>
-                if (is_aligned(p1,sizeof(dtype)) &&
-                    is_aligned(p2,sizeof(dtype)) ) {
+                if (cumo_is_aligned(p1,sizeof(dtype)) &&
+                    cumo_is_aligned(p2,sizeof(dtype)) ) {
                     if (s1 == sizeof(dtype) &&
                         s2 == sizeof(dtype) ) {
                         for (i=0; i<n; i++) {
@@ -56,8 +56,8 @@ static void
                         }
                         return;
                     }
-                    if (is_aligned_step(s1,sizeof(dtype)) &&
-                        is_aligned_step(s2,sizeof(dtype)) ) {
+                    if (cumo_is_aligned_step(s1,sizeof(dtype)) &&
+                        cumo_is_aligned_step(s2,sizeof(dtype)) ) {
                         //<% end %>
                         for (i=0; i<n; i++) {
                             *(dtype*)p2 = m_<%=name%>(*(dtype*)p1);
@@ -69,9 +69,9 @@ static void
                     }
                 }
                 for (i=0; i<n; i++) {
-                    GET_DATA_STRIDE(p1,s1,dtype,x);
+                    CUMO_GET_DATA_STRIDE(p1,s1,dtype,x);
                     x = m_<%=name%>(x);
-                    SET_DATA_STRIDE(p2,s2,dtype,x);
+                    CUMO_SET_DATA_STRIDE(p2,s2,dtype,x);
                 }
                 //<% end %>
             }
@@ -90,15 +90,15 @@ static void
                 <%="cumo_#{c_iter}_stride_index_kernel_launch"%>(p1,p2,s1,idx2,n);
             } else {
                 //<% if need_align %>
-                if (is_aligned(p1,sizeof(dtype)) &&
-                    is_aligned(p2,sizeof(dtype)) ) {
+                if (cumo_is_aligned(p1,sizeof(dtype)) &&
+                    cumo_is_aligned(p2,sizeof(dtype)) ) {
                     if (s1 == sizeof(dtype) &&
                         s2 == sizeof(dtype) ) {
                         <%="cumo_#{c_iter}_contiguous_kernel_launch"%>(p1,p2,n);
                         return;
                     }
-                    if (is_aligned_step(s1,sizeof(dtype)) &&
-                        is_aligned_step(s2,sizeof(dtype)) ) {
+                    if (cumo_is_aligned_step(s1,sizeof(dtype)) &&
+                        cumo_is_aligned_step(s2,sizeof(dtype)) ) {
                         //<% end %>
                         <%="cumo_#{c_iter}_stride_stride_kernel_launch"%>(p1,p2,s1,s2,n);
                         return;
@@ -121,12 +121,12 @@ static void
 static VALUE
 <%=c_func(0)%>(VALUE self)
 {
-    ndfunc_arg_in_t ain[1] = {{cT,0}};
-    ndfunc_arg_out_t aout[1] = {{cT,0}};
-    ndfunc_t ndf = {<%=c_iter%>, FULL_LOOP, 1,1, ain,aout};
+    cumo_ndfunc_arg_in_t ain[1] = {{cT,0}};
+    cumo_ndfunc_arg_out_t aout[1] = {{cT,0}};
+    cumo_ndfunc_t ndf = {<%=c_iter%>, CUMO_FULL_LOOP, 1,1, ain,aout};
 
     <% if name == 'map' %>
     cumo_cuda_runtime_check_status(cudaDeviceSynchronize());
     <% end %>
-    return na_ndloop(&ndf, 1, self);
+    return cumo_na_ndloop(&ndf, 1, self);
 }