saymrwulf/pytorch
1
0
Fork 0
mirror of https://github.com/saymrwulf/pytorch.git synced 2026-05-14 20:57:59 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
pytorch/test
History
Xiao Wang f326f6a8a0 Remove dilation restriction on cuDNN ConvTranspose2d (#46290) 
Summary:
Close https://github.com/pytorch/pytorch/issues/31690

I have verified the functionality of ConvTranspose2d (with this PR) on roughly 32,000 random shapes on V100, A100, using cuDNN 8.0.4 and CUDA 11.1. The 32,000 shapes contain 4x8,000 of (fp16, fp32) x (nchw, nhwc) each.

The random shapes are sampled from
```jsonc
{
    "batch_size": {"low": 1, "high": 8},
    "in_channels": {"low": 16, "high": 128},
    "out_channels": {"low": 16, "high": 128},
    "height": {"low": 16, "high": 224},
    "stride": {"set": [[1, 1], [2, 2]]},
    "padding": {"set": [[0, 0]]},
    "output_padding": {"set": [[0, 0], [1, 1], [0, 1], [1, 0]]},
    "kernel_size": {"set": [[3, 3], [1, 1], [1, 3], [3, 1], [2, 2]]},
    "dilation": {"set": [[1, 1]]},
    "deterministic": {"set": [true, false]},
    "benchmark": {"set": [true, false]},
    "allow_tf32": {"set": [true, false]},
    "groups": {"set": [1, IN_CHANNELS]}
}
```
- Input `width` is the same as `height`.
- `groups` can be either 1, or the same as `in_channels` (grouped convolution). When `groups` is 1, `out_channels` is random; when `groups` is the same as `in_channels`, `out_channels` is also the same as `in_channels`

All of the checked shapes can be found in csv files here https://github.com/xwang233/code-snippet/tree/master/convtranspose2d-dilation/functionality-check-cudnn8.0.4.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/46290

Reviewed By: mruberry

Differential Revision: D24422091

Pulled By: ngimel

fbshipit-source-id: 9f0120f2995ae1575c0502f1b2742390d7937b24
2020-10-22 13:42:03 -07:00
..
backward_compatibility Enabling alias annotation checks for all operations during autograd tests (#46601) 2020-10-21 20:01:57 -07:00
benchmark_utils More Timer refinement (#46023) 2020-10-15 16:32:53 -07:00
bottleneck_test
cpp [NNC] IRSimplifier rules for Compare and Mod (#46412) 2020-10-19 19:37:09 -07:00
cpp_api_parity Support tuple inputs in NN Module test (#44853) 2020-09-28 22:05:05 -07:00
cpp_extensions Add barrier() at the end of init_process_group and new_group. (#45181) 2020-09-25 15:46:59 -07:00
custom_backend [JIT] Modify to_backend API so that it accepts wrapped modules (#43612) 2020-09-28 17:17:01 -07:00
custom_operator
distributed Pull in fairscale.nn.Pipe into PyTorch. (#44090) 2020-10-22 10:59:02 -07:00
distributions Support pytest for distribution testing (#45648) 2020-10-13 10:56:50 -07:00
error_messages
expect
fx [FX] Track use nodes in Node (#45775) 2020-10-07 00:15:04 -07:00
jit [JIT] adding torch.jit.isinstance support (#46062) 2020-10-20 16:47:49 -07:00
mobile Add named tuple's error message and workaround for RET failure (#46347) 2020-10-15 17:41:06 -07:00
namespace_b torch.package - a way to package models and code (#45015) 2020-09-22 21:21:21 -07:00
onnx Replace map(lambda constructs (#46462) 2020-10-22 09:50:22 -07:00
optim
package_a torch.package - a way to package models and code (#45015) 2020-09-22 21:21:21 -07:00
quantization [quant][graphmode][fx] Support quantizing FloatFunctional (#46634) 2020-10-22 01:21:17 -07:00
scripts
test_img
type_hint_tests
HowToWriteTestsUsingFileCheck.md
module_a.py torch.package - a way to package models and code (#45015) 2020-09-22 21:21:21 -07:00
print_test_stats.py Mark top 3 slowest tests as slow (#46068) 2020-10-08 21:10:03 -07:00
run_test.py Pull in fairscale.nn.Pipe into PyTorch. (#44090) 2020-10-22 10:59:02 -07:00
simulate_nccl_errors.py Remove py2 compatible future imports (#44735) 2020-09-16 12:55:57 -07:00
test_autograd.py detect inplace modifications of views earlier (fix #21875) (#46204) 2020-10-19 08:58:33 -07:00
test_bundled_images.py
test_bundled_inputs.py
test_complex.py
test_cpp_api_parity.py Always use NewModuleTest instead of ModuleTest. (#44745) 2020-09-17 07:36:39 -07:00
test_cpp_extensions_aot.py Fix exception chaining in test/ (#44193) 2020-09-14 14:20:16 -07:00
test_cpp_extensions_jit.py
test_cuda.py Minor cleaning of test_cuda.py (#46617) 2020-10-22 09:03:30 -07:00
test_cuda_primary_ctx.py
test_dataloader.py Mark top 3 slowest tests as slow (#46068) 2020-10-08 21:10:03 -07:00
test_determination.py Rename test_distributed to test_distributed_fork (#42932) 2020-09-08 23:13:37 -07:00
test_dispatch.py [WIP] Move catchAll to Math (#45939) 2020-10-16 16:17:16 -07:00
test_expecttest.py
test_foreach.py Revert "Push rocm to slow path (#46216)" (#46728) 2020-10-22 12:04:29 -07:00
test_function_schema.py Stricter backward compatibility check (#45773) 2020-10-06 01:28:48 -07:00
test_functional_autograd_benchmark.py
test_futures.py fix #45552 - adding add_done_callback(fn) to torch.futures.Future (#45675) 2020-10-13 07:47:36 -07:00
test_fx.py [FX] Make wrapped functions traceable (#46692) 2020-10-22 12:00:02 -07:00
test_fx_experimental.py add size_based_partition final (#46282) 2020-10-14 03:44:05 -07:00
test_indexing.py
test_jit.py Replace map(lambda constructs (#46462) 2020-10-22 09:50:22 -07:00
test_jit_cuda_fuser.py [nvFuser] Switching to CudaFusionGuard from BailOut for nvfuser - update 2 (#46452) 2020-10-19 15:44:31 -07:00
test_jit_disabled.py
test_jit_fuser.py Adjust TF32 tests (#44240) 2020-09-24 10:25:58 -07:00
test_jit_fuser_legacy.py Rename jobs and cli switches for testing GraphExecutor configurations to something a little bit more sensical. (#45715) 2020-10-05 11:43:28 -07:00
test_jit_fuser_te.py [numpy] torch.sin : support and promote integer inputs to float (#45733) 2020-10-22 01:58:57 -07:00
test_jit_legacy.py Rename jobs and cli switches for testing GraphExecutor configurations to something a little bit more sensical. (#45715) 2020-10-05 11:43:28 -07:00
test_jit_profiling.py Rename jobs and cli switches for testing GraphExecutor configurations to something a little bit more sensical. (#45715) 2020-10-05 11:43:28 -07:00
test_jit_py3.py [JIT] Add API for ignoring arbitrary module attributes (#45262) 2020-10-06 18:02:06 -07:00
test_jit_simple.py Rename jobs and cli switches for testing GraphExecutor configurations to something a little bit more sensical. (#45715) 2020-10-05 11:43:28 -07:00
test_jit_string.py
test_linalg.py Fix mistakes and increase clarity of norm documentation (#42696) 2020-10-10 14:12:43 -07:00
test_logging.py
test_metal.py [Metal] Add the Python binding for optimize_for_mobile (#46456) 2020-10-17 10:26:25 -07:00
test_mkldnn.py Remove py2 compatible future imports (#44735) 2020-09-16 12:55:57 -07:00
test_mobile_optimizer.py [pytorch] Replace "blacklist" in test/test_mobile_optimizer.py (#45512) 2020-09-30 10:43:59 -07:00
test_multiprocessing.py
test_multiprocessing_spawn.py Add exception classification to torch.multiprocessing.spawn (#45174) 2020-10-09 12:59:41 -07:00
test_namedtensor.py Fixed median nan propagation and implemented nanmedian (#45847) 2020-10-08 11:20:21 -07:00
test_namedtuple_return_api.py Fixed median nan propagation and implemented nanmedian (#45847) 2020-10-08 11:20:21 -07:00
test_native_functions.py Refactor scalar list APIs to use overloads (#45673) 2020-10-19 09:28:49 -07:00
test_nn.py Remove dilation restriction on cuDNN ConvTranspose2d (#46290) 2020-10-22 13:42:03 -07:00
test_numba_integration.py
test_op_aliases.py Migrate addr from the TH to Aten (CPU) (#44364) 2020-09-25 01:18:09 -07:00
test_openmp.py
test_ops.py Revert D24027761: Update backward definition for more operators and reenable tests in test_ops.py 2020-10-02 18:52:57 -07:00
test_optim.py Replace list(map(...)) constructs by list comprehensions (#46461) 2020-10-19 18:42:49 -07:00
test_overrides.py Fix incorrect signatures in get_testing_overrides, and add test for incorrect signatures (#45983) 2020-10-15 07:48:20 -07:00
test_package.py Fix version comparisons for Python 3.6, 3.10 and 4 (#32389) 2020-10-21 11:52:50 -07:00
test_profiler.py Source code level attribution in profiler (#43898) 2020-09-30 00:57:35 -07:00
test_pruning_op.py Remove py2 compatible future imports (#44735) 2020-09-16 12:55:57 -07:00
test_pytree.py Implement _broadcast_to_and_flatten(pytree, spec) (#46288) 2020-10-20 07:52:14 -07:00
test_quantization.py [reland][quant] Add FixedQParamsFakeQuantize module (#45538) (#46657) 2020-10-21 16:47:11 -07:00
test_serialization.py fix test_serialization not working with Windows. (#46120) 2020-10-12 15:18:46 -07:00
test_show_pickle.py
test_sparse.py Fixes bug in sspaddmm (#45113) (#45963) 2020-10-15 16:50:16 -07:00
test_spectral_ops.py Mark parts of spectral tests as slow (#46509) 2020-10-19 10:11:46 -07:00
test_static_runtime.py [StaticRuntime] Implement StaticRuntime::benchmark (#45639) 2020-10-06 20:54:43 -07:00
test_tensor_creation_ops.py Rocm skip test cases (#45782) 2020-10-05 15:12:25 -07:00
test_tensorboard.py Remove py2 compatible future imports (#44735) 2020-09-16 12:55:57 -07:00
test_tensorexpr.py [TensorExpr] Re-enable test for torch.cat, add a test for torch.cat being a single node in a fusion group. (#46447) 2020-10-16 20:26:48 -07:00
test_throughput_benchmark.py Remove py2 compatible future imports (#44735) 2020-09-16 12:55:57 -07:00
test_torch.py Replace map(lambda constructs (#46462) 2020-10-22 09:50:22 -07:00
test_type_hints.py Remove py2 compatible future imports (#44735) 2020-09-16 12:55:57 -07:00
test_type_info.py
test_type_promotion.py [fix] type promotion atan2 (#43466) 2020-09-23 22:23:05 -07:00
test_unary_ufuncs.py [numpy] torch.sin : support and promote integer inputs to float (#45733) 2020-10-22 01:58:57 -07:00
test_utils.py More Timer refinement (#46023) 2020-10-15 16:32:53 -07:00
test_vmap.py Batching rules for: new_zeros, new_empty (#46606) 2020-10-22 11:40:51 -07:00
test_vulkan.py
test_xnnpack_integration.py Mark top 3 slowest tests as slow (#46068) 2020-10-08 21:10:03 -07:00