Cannot Compile Kernel #20
Description
Tests pass for the CL context, but I am not sure where to go with this:
using GPUArrays
u0 = GPUArray(rand(Float32, 32, 32))
abs.(u0)Couldn't compile kernel:
1 : // dependant type declarations
2 : typedef struct {
3 : float empty; // structs can't be empty
4 : }Base21abs;
5 :
6 : // dependant function declarations
7 : float broadcast_index_2(__global float * restrict arg, int2 shape, int i)
8 : {
9 : ;
10 : return arg[i - 1];
11 : ;
12 : }
13 : // Main inner function
14 : __kernel void broadcast_kernel_7(__global float * restrict A, Base21abs f, int2 sz, __global float * restrict arg_1)
15 : {
16 : int i;
17 : i = get_global_id(0) + 1;
18 : float _ssavalue_0;
19 : _ssavalue_0 = abs(broadcast_index_2(arg_1, sz, i));
20 : A[i - 1] = _ssavalue_0;
21 : ;
22 : }
23 :
With following build error:
<kernel>:19:19: error: call to 'abs' is ambiguous
_ssavalue_0 = abs(broadcast_index_2(arg_1, sz, i));
^~~
cl_kernel.h:2310:24: note: candidate function
uchar __OVERLOADABLE__ abs(char);
^
cl_kernel.h:2311:24: note: candidate function
uchar __OVERLOADABLE__ abs(uchar);
^
cl_kernel.h:2312:25: note: candidate function
ushort __OVERLOADABLE__ abs(short);
^
cl_kernel.h:2313:25: note: candidate function
ushort __OVERLOADABLE__ abs(ushort);
^
cl_kernel.h:2314:23: note: candidate function
uint __OVERLOADABLE__ abs(int);
^
cl_kernel.h:2315:23: note: candidate function
uint __OVERLOADABLE__ abs(uint);
^
cl_kernel.h:2316:24: note: candidate function
ulong __OVERLOADABLE__ abs(long);
^
cl_kernel.h:2317:24: note: candidate function
ulong __OVERLOADABLE__ abs(ulong);
^
cl_kernel.h:2319:25: note: candidate function
uchar2 __OVERLOADABLE__ abs(char2);
^
cl_kernel.h:2321:25: note: candidate function
uchar3 __OVERLOADABLE__ abs(char3);
^
cl_kernel.h:2323:25: note: candidate function
uchar4 __OVERLOADABLE__ abs(char4);
^
cl_kernel.h:2324:25: note: candidate function
uchar8 __OVERLOADABLE__ abs(char8);
^
cl_kernel.h:2325:26: note: candidate function
uchar16 __OVERLOADABLE__ abs(char16);
^
cl_kernel.h:2326:25: note: candidate function
uchar2 __OVERLOADABLE__ abs(uchar2);
^
cl_kernel.h:2328:25: note: candidate function
uchar3 __OVERLOADABLE__ abs(uchar3);
^
cl_kernel.h:2330:25: note: candidate function
uchar4 __OVERLOADABLE__ abs(uchar4);
^
cl_kernel.h:2331:25: note: candidate function
uchar8 __OVERLOADABLE__ abs(uchar8);
^
cl_kernel.h:2332:26: note: candidate function
uchar16 __OVERLOADABLE__ abs(uchar16);
^
cl_kernel.h:2333:26: note: candidate function
ushort2 __OVERLOADABLE__ abs(short2);
^
cl_kernel.h:2335:26: note: candidate function
ushort3 __OVERLOADABLE__ abs(short3);
^
cl_kernel.h:2337:26: note: candidate function
ushort4 __OVERLOADABLE__ abs(short4);
^
cl_kernel.h:2338:26: note: candidate function
ushort8 __OVERLOADABLE__ abs(short8);
^
cl_kernel.h:2339:27: note: candidate function
ushort16 __OVERLOADABLE__ abs(short16);
^
cl_kernel.h:2340:26: note: candidate function
ushort2 __OVERLOADABLE__ abs(ushort2);
^
cl_kernel.h:2342:26: note: candidate function
ushort3 __OVERLOADABLE__ abs(ushort3);
^
cl_kernel.h:2344:26: note: candidate function
ushort4 __OVERLOADABLE__ abs(ushort4);
^
cl_kernel.h:2345:26: note: candidate function
ushort8 __OVERLOADABLE__ abs(ushort8);
^
cl_kernel.h:2346:27: note: candidate function
ushort16 __OVERLOADABLE__ abs(ushort16);
^
cl_kernel.h:2347:24: note: candidate function
uint2 __OVERLOADABLE__ abs(int2);
^
cl_kernel.h:2349:24: note: candidate function
uint3 __OVERLOADABLE__ abs(int3);
^
cl_kernel.h:2351:24: note: candidate function
uint4 __OVERLOADABLE__ abs(int4);
^
cl_kernel.h:2352:24: note: candidate function
uint8 __OVERLOADABLE__ abs(int8);
^
cl_kernel.h:2353:25: note: candidate function
uint16 __OVERLOADABLE__ abs(int16);
^
cl_kernel.h:2354:24: note: candidate function
uint2 __OVERLOADABLE__ abs(uint2);
^
cl_kernel.h:2356:24: note: candidate function
uint3 __OVERLOADABLE__ abs(uint3);
^
cl_kernel.h:2358:24: note: candidate function
uint4 __OVERLOADABLE__ abs(uint4);
^
cl_kernel.h:2359:24: note: candidate function
uint8 __OVERLOADABLE__ abs(uint8);
^
cl_kernel.h:2360:25: note: candidate function
uint16 __OVERLOADABLE__ abs(uint16);
^
cl_kernel.h:2361:25: note: candidate function
ulong2 __OVERLOADABLE__ abs(long2);
^
cl_kernel.h:2363:25: note: candidate function
ulong3 __OVERLOADABLE__ abs(long3);
^
cl_kernel.h:2365:25: note: candidate function
ulong4 __OVERLOADABLE__ abs(long4);
^
cl_kernel.h:2366:25: note: candidate function
ulong8 __OVERLOADABLE__ abs(long8);
^
cl_kernel.h:2367:26: note: candidate function
ulong16 __OVERLOADABLE__ abs(long16);
^
cl_kernel.h:2368:25: note: candidate function
ulong2 __OVERLOADABLE__ abs(ulong2);
^
cl_kernel.h:2370:25: note: candidate function
ulong3 __OVERLOADABLE__ abs(ulong3);
^
cl_kernel.h:2372:25: note: candidate function
ulong4 __OVERLOADABLE__ abs(ulong4);
^
cl_kernel.h:2373:25: note: candidate function
ulong8 __OVERLOADABLE__ abs(ulong8);
^
cl_kernel.h:2374:26: note: candidate function
ulong16 __OVERLOADABLE__ abs(ulong16);
^
CLError(code=-11, CL_BUILD_PROGRAM_FAILURE)
in broadcast at GPUArrays\src\abstractarray.jl:186
in acc_broadcast! at GPUArrays\src\backends\opencl\opencl.jl:168
in Type at Transpiler\src\clike/opencl\compilation.jl:79
in get! at base\dict.jl:449
in #27 at Transpiler\src\clike/opencl\compilation.jl:104
in #build! at base\<missing>
in #build!#113 at OpenCL\src\program.jl:101
in macro expansion at OpenCL\src\macros.jl:6