Review: use int64_t for blockDim.x, rename nb->s for clarity

Author: am17an

ggml/src/ggml-cuda/set-rows.cu

@@ -20,11 +20,11 @@ static __global__ void k_set_rows(
         const src_t * __restrict__ src0, const int64_t * __restrict__ src1, dst_t * __restrict__ dst,
         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
         const int64_t ne10, const int64_t ne11, const int64_t ne12, const int64_t ne13,
-        const size_t nb01, const size_t nb02, const size_t nb03,
-        const size_t nb10, const size_t nb11, const size_t nb12,
-        const size_t nb1, const size_t nb2, const size_t nb3) {
+        const int64_t s01, const int64_t s02, const int64_t s03,
+        const int64_t s10, const int64_t s11, const int64_t s12,
+        const int64_t s1, const int64_t s2, const int64_t s3) {
 
-    const int64_t i = blockDim.x * blockIdx.x + threadIdx.x;
+    const int64_t i = int64_t(blockDim.x) * blockIdx.x + threadIdx.x;
     const int64_t ne_total = ne00 * ne01 * ne02 * ne03;
 
     if (i >= ne_total) {
@@ -40,10 +40,10 @@ static __global__ void k_set_rows(
     const int64_t i11 = i02 % ne11;
     const int64_t i10 = i01;
 
-    const int64_t dst_row = *(src1 + i10*nb10 + i11*nb11 + i12*nb12);
+    const int64_t dst_row = *(src1 + i10*s10 + i11*s11 + i12*s12);
 
-    const src_t * src0_row = src0 + i01*nb01 + i02*nb02 + i03*nb03;
-    dst_t * dst_row_ptr    = dst + dst_row*nb1 + i02*nb2 + i03*nb3;
+    const src_t * src0_row = src0 + i01*s01 + i02*s02 + i03*s03;
+    dst_t * dst_row_ptr    = dst + dst_row*s1 + i02*s2 + i03*s3;
 
     const src_t* src_elem = src0_row + i00;
     dst_t* dst_elem = dst_row_ptr + i00;