| 2295 | } |
| 2296 | |
| 2297 | void test_more_ops(const long nr, const long nc) |
| 2298 | { |
| 2299 | using namespace dlib::tt; |
| 2300 | print_spinner(); |
| 2301 | // We are going to make sure that the CPU implementation of these things matches |
| 2302 | // the CUDA implementation. |
| 2303 | |
| 2304 | tensor_rand rnd; |
| 2305 | |
| 2306 | resizable_tensor dest(nr,nc), src(nr,nc), dest2, src2; |
| 2307 | resizable_tensor srcb(nr,nc), srcc(nr,nc), srcb2, srcc2; |
| 2308 | |
| 2309 | |
| 2310 | rnd.fill_uniform(dest); |
| 2311 | rnd.fill_uniform(src); |
| 2312 | dest2 = dest; src2 = src; |
| 2313 | cuda::multiply(false, dest, dest, src); |
| 2314 | cpu::multiply(false, dest2, dest2, src2); |
| 2315 | DLIB_TEST(equal(mat(dest),mat(dest2))); |
| 2316 | cuda::multiply(true, dest, dest, src); |
| 2317 | cpu::multiply(true, dest2, dest2, src2); |
| 2318 | DLIB_TEST(equal(mat(dest),mat(dest2))); |
| 2319 | |
| 2320 | |
| 2321 | rnd.fill_uniform(dest); |
| 2322 | rnd.fill_uniform(src); |
| 2323 | dest2 = dest; src2 = src; |
| 2324 | cuda::affine_transform(dest, src, 2, 3); |
| 2325 | cpu::affine_transform(dest2, src2, 2, 3); |
| 2326 | DLIB_TEST(equal(mat(dest),mat(dest2))); |
| 2327 | |
| 2328 | rnd.fill_uniform(dest); |
| 2329 | rnd.fill_uniform(src); |
| 2330 | rnd.fill_uniform(srcb); |
| 2331 | dest2 = dest; src2 = src; srcb2 = srcb; |
| 2332 | cuda::affine_transform(dest, src, srcb, 2, 3, 4); |
| 2333 | cpu::affine_transform(dest2, src2, srcb2, 2, 3, 4); |
| 2334 | DLIB_TEST(equal(mat(dest),mat(dest2))); |
| 2335 | |
| 2336 | rnd.fill_uniform(dest); |
| 2337 | rnd.fill_uniform(src); |
| 2338 | rnd.fill_uniform(srcb); |
| 2339 | rnd.fill_uniform(srcc); |
| 2340 | dest2 = dest; src2 = src; srcb2 = srcb; srcc2 = srcc; |
| 2341 | cuda::affine_transform(dest, src, srcb, srcc, 2, 3, 4, 5); |
| 2342 | cpu::affine_transform(dest2, src2, srcb2, srcc2, 2, 3, 4, 5); |
| 2343 | DLIB_TEST(equal(mat(dest),mat(dest2))); |
| 2344 | |
| 2345 | cuda::affine_transform(dest, src, srcb, srcc, 2, 3, 4, 0); |
| 2346 | cpu::affine_transform(dest2, src2, srcb2, srcc2, 2, 3, 4, 0); |
| 2347 | DLIB_TEST(equal(mat(dest),mat(dest2))); |
| 2348 | |
| 2349 | cuda::affine_transform_range(0, dest.size(), dest, src, srcb, srcc, 2, 3, 4); |
| 2350 | cpu::affine_transform_range(0, dest2.size(), dest2, src2, srcb2, srcc2, 2, 3, 4); |
| 2351 | DLIB_TEST(equal(mat(dest),mat(dest2))); |
| 2352 | |
| 2353 | if (3 < dest.size()) |
| 2354 | { |
no test coverage detected