|
Line
Link Here
|
| 0 |
-- |
0 |
++ b/backend/src/libocl/tmpl/ocl_integer.tmpl.cl |
| 1 |
-- a/backend/src/libocl/tmpl/ocl_integer.tmpl.cl |
|
|
|
Lines 216-228
OVERLOADABLE ulong mad_sat(ulong a, ulong b, ulong c) {
Link Here
|
| 216 |
return __gen_ocl_mad_sat(a, b, c); |
216 |
return __gen_ocl_mad_sat(a, b, c); |
| 217 |
} |
217 |
} |
| 218 |
OVERLOADABLE uchar __rotate_left(uchar x, uchar y) { return (x << y) | (x >> (8 - y)); } |
218 |
// the 'volatile' is to make the LLVM optimizer leave these alone, as it would convert them to intrinsics (fshl/fshr) that we don't implement |
|
|
219 |
OVERLOADABLE uchar __rotate_left(uchar x, uchar y) { volatile uchar z; z = (x << y); return z | (x >> (8 - y)); } |
| 219 |
OVERLOADABLE char __rotate_left(char x, char y) { return __rotate_left((uchar)x, (uchar)y); } |
220 |
OVERLOADABLE char __rotate_left(char x, char y) { return __rotate_left((uchar)x, (uchar)y); } |
| 220 |
OVERLOADABLE ushort __rotate_left(ushort x, ushort y) { return (x << y) | (x >> (16 - y)); } |
221 |
OVERLOADABLE ushort __rotate_left(ushort x, ushort y) { volatile ushort z; z = (x << y); return z | (x >> (16 - y)); } |
| 221 |
OVERLOADABLE short __rotate_left(short x, short y) { return __rotate_left((ushort)x, (ushort)y); } |
222 |
OVERLOADABLE short __rotate_left(short x, short y) { return __rotate_left((ushort)x, (ushort)y); } |
| 222 |
OVERLOADABLE uint __rotate_left(uint x, uint y) { return (x << y) | (x >> (32 - y)); } |
223 |
OVERLOADABLE uint __rotate_left(uint x, uint y) { volatile uint z; z = (x << y); return z | (x >> (32 - y)); } |
| 223 |
OVERLOADABLE int __rotate_left(int x, int y) { return __rotate_left((uint)x, (uint)y); } |
224 |
OVERLOADABLE int __rotate_left(int x, int y) { return __rotate_left((uint)x, (uint)y); } |
| 224 |
OVERLOADABLE ulong __rotate_left(ulong x, ulong y) { return (x << y) | (x >> (64 - y)); } |
225 |
OVERLOADABLE ulong __rotate_left(ulong x, ulong y) { volatile ulong z; z = (x << y); return z | (x >> (64 - y)); } |
| 225 |
OVERLOADABLE long __rotate_left(long x, long y) { return __rotate_left((ulong)x, (ulong)y); } |
226 |
OVERLOADABLE long __rotate_left(long x, long y) { return __rotate_left((ulong)x, (ulong)y); } |
| 226 |
#define DEF(type, m) OVERLOADABLE type rotate(type x, type y) { return __rotate_left(x, (type)(y & m)); } |
227 |
#define DEF(type, m) OVERLOADABLE type rotate(type x, type y) { return __rotate_left(x, (type)(y & m)); } |
| 227 |
DEF(char, 7) |
228 |
DEF(char, 7) |
| 228 |
-- a/backend/src/backend/gen_register.hpp |
229 |
++ b/backend/src/backend/gen_register.hpp |
|
Lines 225-230
namespace gbe
Link Here
|
| 225 |
uint32_t width, |
225 |
uint32_t width, |
| 226 |
uint32_t hstride) |
226 |
uint32_t hstride) |
| 227 |
{ |
227 |
{ |
|
|
228 |
this->value.reg = 0;//avoid subgroup crash |
| 228 |
this->type = type; |
229 |
this->type = type; |
| 229 |
this->file = file; |
230 |
this->file = file; |
| 230 |
this->nr = nr; |
231 |
this->nr = nr; |
