DISABLED test_grad_scaling_autocast_fused_optimizers_Adam_cuda_float32 (__main__.TestCudaOptimsCUDA)

[pytorch-bot[bot]]

Platforms: linux

This test was disabled because it is failing in CI. See recent examples and the most recent trunk workflow logs.

Over the past 3 hours, it has been determined flaky in 8 workflow(s) with 24 failures and 8 successes.

Debugging instructions (after clicking on the recent samples link): DO NOT ASSUME THINGS ARE OKAY IF THE CI IS GREEN. We now shield flaky tests from developers so CI will thus be green but it will be harder to parse the logs. To find relevant log snippets:

1. Click on the workflow logs linked above
2. Click on the Test step of the job so that it is expanded. Otherwise, the grepping will not work.
3. Grep for test_grad_scaling_autocast_fused_optimizers_Adam_cuda_float32
4. There should be several instances run (as flaky tests are rerun in CI) from which you can study the logs.

Sample error message

``` Traceback (most recent call last): File "/var/lib/jenkins/workspace/test/test_cuda.py", line 4665, in test_grad_scaling_autocast_fused_optimizers scaler_control.step(opt_control) File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/amp/grad_scaler.py", line 457, in step retval = self._maybe_opt_step(optimizer, optimizer_state, *args, **kwargs) File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/amp/grad_scaler.py", line 351, in _maybe_opt_step if not sum(v.item() for v in optimizer_state["found_inf_per_device"].values()): File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/amp/grad_scaler.py", line 351, in if not sum(v.item() for v in optimizer_state["found_inf_per_device"].values()): RuntimeError: Host and device pointer dont match with cudaHostRegister. Please dont use this feature by setting PYTORCH_CUDA_ALLOC_CONF=use_cuda_host_register:False (default) To execute this test, run the following from the base repo dir: python test/test_cuda.py TestCudaOptimsCUDA.test_grad_scaling_autocast_fused_optimizers_Adam_cuda_float32 This message can be suppressed by setting PYTORCH_PRINT_REPRO_ON_FAILURE=0 ```

Test file path: test_cuda.py

cc @ptrblck @msaroufim @clee2000

[pytorch-bot[bot]]

Hello there! From the DISABLED prefix in this issue title, it looks like you are attempting to disable a test in PyTorch CI. The information I have parsed is below: * Test name: `test_grad_scaling_autocast_fused_optimizers_Adam_cuda_float32 (__main__.TestCudaOptimsCUDA)` * Platforms for which to skip the test: linux * Disabled by `pytorch-bot[bot]` Within ~15 minutes, `test_grad_scaling_autocast_fused_optimizers_Adam_cuda_float32 (__main__.TestCudaOptimsCUDA)` will be disabled in PyTorch CI for these platforms: linux. Please verify that your test name looks correct, e.g., `test_cuda_assert_async (__main__.TestCuda)`. To modify the platforms list, please include a line in the issue body, like below. The default action will disable the test for all platforms if no platforms list is specified. ``` Platforms: case-insensitive, list, of, platforms ``` We currently support the following platforms: asan, dynamo, inductor, linux, mac, macos, rocm, slow, win, windows. ### How to re-enable a test To re-enable the test globally, close the issue. To re-enable a test for only a subset of platforms, remove the platforms from the list in the issue body. This may take some time to propagate. To re-enable a test only for a PR, put `Fixes #135993` in the PR body and rerun the test jobs. Note that if a test is flaky, it maybe be difficult to tell if the test is still flaky on the PR.

[pytorch-bot[bot]]

Another case of trunk flakiness has been found here. The list of platforms [linux] appears to contain all the recently affected platforms [linux]. Either the change didn't propogate fast enough or disable bot might be broken.