if(h==1) return;
*(uint16_t*)(p + 1*stride)= v;
if(h==2) return;
- *(uint16_t*)(p + 2*stride)=
+ *(uint16_t*)(p + 2*stride)= v;
*(uint16_t*)(p + 3*stride)= v;
}else if(w==4){
const uint32_t v= size==4 ? val : val*0x01010101;
if(h==1) return;
*(uint32_t*)(p + 1*stride)= v;
if(h==2) return;
- *(uint32_t*)(p + 2*stride)=
+ *(uint32_t*)(p + 2*stride)= v;
*(uint32_t*)(p + 3*stride)= v;
}else if(w==8){
//gcc can't optimize 64bit math on x86_32
if(h==1) return;
*(uint64_t*)(p + 1*stride)= v;
if(h==2) return;
- *(uint64_t*)(p + 2*stride)=
+ *(uint64_t*)(p + 2*stride)= v;
*(uint64_t*)(p + 3*stride)= v;
}else if(w==16){
const uint64_t v= val*0x0100000001ULL;
- *(uint64_t*)(p + 0+0*stride)=
- *(uint64_t*)(p + 8+0*stride)=
- *(uint64_t*)(p + 0+1*stride)=
+ *(uint64_t*)(p + 0+0*stride)= v;
+ *(uint64_t*)(p + 8+0*stride)= v;
+ *(uint64_t*)(p + 0+1*stride)= v;
*(uint64_t*)(p + 8+1*stride)= v;
if(h==2) return;
- *(uint64_t*)(p + 0+2*stride)=
- *(uint64_t*)(p + 8+2*stride)=
- *(uint64_t*)(p + 0+3*stride)=
+ *(uint64_t*)(p + 0+2*stride)= v;
+ *(uint64_t*)(p + 8+2*stride)= v;
+ *(uint64_t*)(p + 0+3*stride)= v;
*(uint64_t*)(p + 8+3*stride)= v;
#else
- *(uint32_t*)(p + 0+0*stride)=
+ *(uint32_t*)(p + 0+0*stride)= val;
*(uint32_t*)(p + 4+0*stride)= val;
if(h==1) return;
- *(uint32_t*)(p + 0+1*stride)=
+ *(uint32_t*)(p + 0+1*stride)= val;
*(uint32_t*)(p + 4+1*stride)= val;
if(h==2) return;
- *(uint32_t*)(p + 0+2*stride)=
- *(uint32_t*)(p + 4+2*stride)=
- *(uint32_t*)(p + 0+3*stride)=
+ *(uint32_t*)(p + 0+2*stride)= val;
+ *(uint32_t*)(p + 4+2*stride)= val;
+ *(uint32_t*)(p + 0+3*stride)= val;
*(uint32_t*)(p + 4+3*stride)= val;
}else if(w==16){
- *(uint32_t*)(p + 0+0*stride)=
- *(uint32_t*)(p + 4+0*stride)=
- *(uint32_t*)(p + 8+0*stride)=
- *(uint32_t*)(p +12+0*stride)=
- *(uint32_t*)(p + 0+1*stride)=
- *(uint32_t*)(p + 4+1*stride)=
- *(uint32_t*)(p + 8+1*stride)=
+ *(uint32_t*)(p + 0+0*stride)= val;
+ *(uint32_t*)(p + 4+0*stride)= val;
+ *(uint32_t*)(p + 8+0*stride)= val;
+ *(uint32_t*)(p +12+0*stride)= val;
+ *(uint32_t*)(p + 0+1*stride)= val;
+ *(uint32_t*)(p + 4+1*stride)= val;
+ *(uint32_t*)(p + 8+1*stride)= val;
*(uint32_t*)(p +12+1*stride)= val;
if(h==2) return;
- *(uint32_t*)(p + 0+2*stride)=
- *(uint32_t*)(p + 4+2*stride)=
- *(uint32_t*)(p + 8+2*stride)=
- *(uint32_t*)(p +12+2*stride)=
- *(uint32_t*)(p + 0+3*stride)=
- *(uint32_t*)(p + 4+3*stride)=
- *(uint32_t*)(p + 8+3*stride)=
+ *(uint32_t*)(p + 0+2*stride)= val;
+ *(uint32_t*)(p + 4+2*stride)= val;
+ *(uint32_t*)(p + 8+2*stride)= val;
+ *(uint32_t*)(p +12+2*stride)= val;
+ *(uint32_t*)(p + 0+3*stride)= val;
+ *(uint32_t*)(p + 4+3*stride)= val;
+ *(uint32_t*)(p + 8+3*stride)= val;
*(uint32_t*)(p +12+3*stride)= val;
#endif
}else