[rocm] F.embedding reports invalid configuration argument #130806
Assignees
Labels
module: rocm
AMD GPU support for Pytorch
triaged
This issue has been looked at a team member, and triaged and prioritized into an appropriate module
Milestone
Comments
|
cc. @jeffdaily |
malfet
added
the
triaged
|
@xw285cornell Reproduced. Thanks for reporting. @jeffdaily : Here is the problem: The number 2147483648 looks quite crazy in hipLuanchKernel. |
|
will put out a PR to fix this soon. |
DiweiSun
pushed a commit
to DiweiSun/pytorch
that referenced
this issue
Jul 22, 2024
pytorch#130994) …with large index Fixes pytorch#130806 When an output size of 2147483648 (=131072*16384) is expected in the above issue, it throwed out the following error: RuntimeError: HIP error: invalid configuration argument What happened was that the second parameter passed to hipLaunchKernel was crazy {2147483648,1,1}. Found two issues in the Indexing.cu: 1: ptrdiff_t was used but it is signed int, outTotalSize >= 2147483648 can cause overflow when doing [this](https://github.com/pytorch/pytorch/blame/39493aa93419532957e6e5ee97cae842b53b8b59/aten/src/ATen/native/cuda/Indexing.cu#L1367): 2: On ROCm, std::min -> ::min did not work as expected when outTotalSize>=2147483648 As the result, 2147483648 was sent to hipLaunchKernel which the GPU does not support such a huge number since this number specifies the number of threads per block. The original code intended to set 128 threads per block, though this is debatable as the perf would not good for latest powerful GPUs (a TODO item to update for perf maybe?) , but at least it would not cause `invalid configuration argument` error. [Test] Run the same code snippet in the [issue](pytorch#130806), and print the output, its dim and numel(), which looks like below now: ``` output=tensor([[ 0.4044, -0.0244, -0.6865, ..., -0.7800, 0.1175, 1.6726], [-1.0866, -0.1609, 0.3538, ..., 1.9105, 0.7882, 1.1583], [-2.2079, 0.3736, 0.3610, ..., -0.2658, -0.0459, 1.3077], ..., [ 0.8753, -0.7482, -0.1978, ..., 0.9016, 1.1501, -0.5178], [-1.5845, -0.6277, 1.4520, ..., 0.5733, -2.1198, -0.0915], [-0.6310, -1.0239, -0.1910, ..., 0.4309, 0.1630, 0.3239]], device='cuda:0'), dim=2, numel=2147483648 ``` Added a large tensor unit test too. ``` /pytorch# pytest test/nn/test_embedding.py -k test_large_tensors ================================================================================== test session starts =================================================================================== platform linux -- Python 3.9.19, pytest-7.3.2, pluggy-1.4.0 rootdir: /dockerx/development/pytorch configfile: pytest.ini plugins: flakefinder-1.1.0, rerunfailures-14.0, xdist-3.3.1, xdoctest-1.1.0, cpp-2.3.0, hypothesis-5.35.1 collected 288 items / 287 deselected / 1 selected Running 1 items in this shard test/nn/test_embedding.py . [100%] =========================================================================== 1 passed, 287 deselected in 3.16s ============================================================================ ``` Pull Request resolved: pytorch#130994 Approved by: https://github.com/jeffdaily, https://github.com/xw285cornell
xuhancn
pushed a commit
to xuhancn/pytorch
that referenced
this issue
Jul 25, 2024
pytorch#130994) …with large index Fixes pytorch#130806 When an output size of 2147483648 (=131072*16384) is expected in the above issue, it throwed out the following error: RuntimeError: HIP error: invalid configuration argument What happened was that the second parameter passed to hipLaunchKernel was crazy {2147483648,1,1}. Found two issues in the Indexing.cu: 1: ptrdiff_t was used but it is signed int, outTotalSize >= 2147483648 can cause overflow when doing [this](https://github.com/pytorch/pytorch/blame/39493aa93419532957e6e5ee97cae842b53b8b59/aten/src/ATen/native/cuda/Indexing.cu#L1367): 2: On ROCm, std::min -> ::min did not work as expected when outTotalSize>=2147483648 As the result, 2147483648 was sent to hipLaunchKernel which the GPU does not support such a huge number since this number specifies the number of threads per block. The original code intended to set 128 threads per block, though this is debatable as the perf would not good for latest powerful GPUs (a TODO item to update for perf maybe?) , but at least it would not cause `invalid configuration argument` error. [Test] Run the same code snippet in the [issue](pytorch#130806), and print the output, its dim and numel(), which looks like below now: ``` output=tensor([[ 0.4044, -0.0244, -0.6865, ..., -0.7800, 0.1175, 1.6726], [-1.0866, -0.1609, 0.3538, ..., 1.9105, 0.7882, 1.1583], [-2.2079, 0.3736, 0.3610, ..., -0.2658, -0.0459, 1.3077], ..., [ 0.8753, -0.7482, -0.1978, ..., 0.9016, 1.1501, -0.5178], [-1.5845, -0.6277, 1.4520, ..., 0.5733, -2.1198, -0.0915], [-0.6310, -1.0239, -0.1910, ..., 0.4309, 0.1630, 0.3239]], device='cuda:0'), dim=2, numel=2147483648 ``` Added a large tensor unit test too. ``` /pytorch# pytest test/nn/test_embedding.py -k test_large_tensors ================================================================================== test session starts =================================================================================== platform linux -- Python 3.9.19, pytest-7.3.2, pluggy-1.4.0 rootdir: /dockerx/development/pytorch configfile: pytest.ini plugins: flakefinder-1.1.0, rerunfailures-14.0, xdist-3.3.1, xdoctest-1.1.0, cpp-2.3.0, hypothesis-5.35.1 collected 288 items / 287 deselected / 1 selected Running 1 items in this shard test/nn/test_embedding.py . [100%] =========================================================================== 1 passed, 287 deselected in 3.16s ============================================================================ ``` Pull Request resolved: pytorch#130994 Approved by: https://github.com/jeffdaily, https://github.com/xw285cornell
pytorchbot
pushed a commit
that referenced
this issue
Aug 13, 2024
#130994) …with large index Fixes #130806 When an output size of 2147483648 (=131072*16384) is expected in the above issue, it throwed out the following error: RuntimeError: HIP error: invalid configuration argument What happened was that the second parameter passed to hipLaunchKernel was crazy {2147483648,1,1}. Found two issues in the Indexing.cu: 1: ptrdiff_t was used but it is signed int, outTotalSize >= 2147483648 can cause overflow when doing [this](https://github.com/pytorch/pytorch/blame/39493aa93419532957e6e5ee97cae842b53b8b59/aten/src/ATen/native/cuda/Indexing.cu#L1367): 2: On ROCm, std::min -> ::min did not work as expected when outTotalSize>=2147483648 As the result, 2147483648 was sent to hipLaunchKernel which the GPU does not support such a huge number since this number specifies the number of threads per block. The original code intended to set 128 threads per block, though this is debatable as the perf would not good for latest powerful GPUs (a TODO item to update for perf maybe?) , but at least it would not cause `invalid configuration argument` error. [Test] Run the same code snippet in the [issue](#130806), and print the output, its dim and numel(), which looks like below now: ``` output=tensor([[ 0.4044, -0.0244, -0.6865, ..., -0.7800, 0.1175, 1.6726], [-1.0866, -0.1609, 0.3538, ..., 1.9105, 0.7882, 1.1583], [-2.2079, 0.3736, 0.3610, ..., -0.2658, -0.0459, 1.3077], ..., [ 0.8753, -0.7482, -0.1978, ..., 0.9016, 1.1501, -0.5178], [-1.5845, -0.6277, 1.4520, ..., 0.5733, -2.1198, -0.0915], [-0.6310, -1.0239, -0.1910, ..., 0.4309, 0.1630, 0.3239]], device='cuda:0'), dim=2, numel=2147483648 ``` Added a large tensor unit test too. ``` /pytorch# pytest test/nn/test_embedding.py -k test_large_tensors ================================================================================== test session starts =================================================================================== platform linux -- Python 3.9.19, pytest-7.3.2, pluggy-1.4.0 rootdir: /dockerx/development/pytorch configfile: pytest.ini plugins: flakefinder-1.1.0, rerunfailures-14.0, xdist-3.3.1, xdoctest-1.1.0, cpp-2.3.0, hypothesis-5.35.1 collected 288 items / 287 deselected / 1 selected Running 1 items in this shard test/nn/test_embedding.py . [100%] =========================================================================== 1 passed, 287 deselected in 3.16s ============================================================================ ``` Pull Request resolved: #130994 Approved by: https://github.com/jeffdaily, https://github.com/xw285cornell (cherry picked from commit 637ab85)
atalman
pushed a commit
that referenced
this issue
Aug 14, 2024
#133346) fix for launching kernel invalid config error when calling embedding … (#130994) …with large index Fixes #130806 When an output size of 2147483648 (=131072*16384) is expected in the above issue, it throwed out the following error: RuntimeError: HIP error: invalid configuration argument What happened was that the second parameter passed to hipLaunchKernel was crazy {2147483648,1,1}. Found two issues in the Indexing.cu: 1: ptrdiff_t was used but it is signed int, outTotalSize >= 2147483648 can cause overflow when doing [this](https://github.com/pytorch/pytorch/blame/39493aa93419532957e6e5ee97cae842b53b8b59/aten/src/ATen/native/cuda/Indexing.cu#L1367): 2: On ROCm, std::min -> ::min did not work as expected when outTotalSize>=2147483648 As the result, 2147483648 was sent to hipLaunchKernel which the GPU does not support such a huge number since this number specifies the number of threads per block. The original code intended to set 128 threads per block, though this is debatable as the perf would not good for latest powerful GPUs (a TODO item to update for perf maybe?) , but at least it would not cause `invalid configuration argument` error. [Test] Run the same code snippet in the [issue](#130806), and print the output, its dim and numel(), which looks like below now: ``` output=tensor([[ 0.4044, -0.0244, -0.6865, ..., -0.7800, 0.1175, 1.6726], [-1.0866, -0.1609, 0.3538, ..., 1.9105, 0.7882, 1.1583], [-2.2079, 0.3736, 0.3610, ..., -0.2658, -0.0459, 1.3077], ..., [ 0.8753, -0.7482, -0.1978, ..., 0.9016, 1.1501, -0.5178], [-1.5845, -0.6277, 1.4520, ..., 0.5733, -2.1198, -0.0915], [-0.6310, -1.0239, -0.1910, ..., 0.4309, 0.1630, 0.3239]], device='cuda:0'), dim=2, numel=2147483648 ``` Added a large tensor unit test too. ``` /pytorch# pytest test/nn/test_embedding.py -k test_large_tensors ================================================================================== test session starts =================================================================================== platform linux -- Python 3.9.19, pytest-7.3.2, pluggy-1.4.0 rootdir: /dockerx/development/pytorch configfile: pytest.ini plugins: flakefinder-1.1.0, rerunfailures-14.0, xdist-3.3.1, xdoctest-1.1.0, cpp-2.3.0, hypothesis-5.35.1 collected 288 items / 287 deselected / 1 selected Running 1 items in this shard test/nn/test_embedding.py . [100%] =========================================================================== 1 passed, 287 deselected in 3.16s ============================================================================ ``` Pull Request resolved: #130994 Approved by: https://github.com/jeffdaily, https://github.com/xw285cornell (cherry picked from commit 637ab85) Co-authored-by: hongxyan <hongxyan@amd.com>
pruthvistony
pushed a commit
to ROCm/pytorch
that referenced
this issue
Aug 15, 2024
pytorch#133346) fix for launching kernel invalid config error when calling embedding … (pytorch#130994) …with large index Fixes pytorch#130806 When an output size of 2147483648 (=131072*16384) is expected in the above issue, it throwed out the following error: RuntimeError: HIP error: invalid configuration argument What happened was that the second parameter passed to hipLaunchKernel was crazy {2147483648,1,1}. Found two issues in the Indexing.cu: 1: ptrdiff_t was used but it is signed int, outTotalSize >= 2147483648 can cause overflow when doing [this](https://github.com/pytorch/pytorch/blame/39493aa93419532957e6e5ee97cae842b53b8b59/aten/src/ATen/native/cuda/Indexing.cu#L1367): 2: On ROCm, std::min -> ::min did not work as expected when outTotalSize>=2147483648 As the result, 2147483648 was sent to hipLaunchKernel which the GPU does not support such a huge number since this number specifies the number of threads per block. The original code intended to set 128 threads per block, though this is debatable as the perf would not good for latest powerful GPUs (a TODO item to update for perf maybe?) , but at least it would not cause `invalid configuration argument` error. [Test] Run the same code snippet in the [issue](pytorch#130806), and print the output, its dim and numel(), which looks like below now: ``` output=tensor([[ 0.4044, -0.0244, -0.6865, ..., -0.7800, 0.1175, 1.6726], [-1.0866, -0.1609, 0.3538, ..., 1.9105, 0.7882, 1.1583], [-2.2079, 0.3736, 0.3610, ..., -0.2658, -0.0459, 1.3077], ..., [ 0.8753, -0.7482, -0.1978, ..., 0.9016, 1.1501, -0.5178], [-1.5845, -0.6277, 1.4520, ..., 0.5733, -2.1198, -0.0915], [-0.6310, -1.0239, -0.1910, ..., 0.4309, 0.1630, 0.3239]], device='cuda:0'), dim=2, numel=2147483648 ``` Added a large tensor unit test too. ``` /pytorch# pytest test/nn/test_embedding.py -k test_large_tensors ================================================================================== test session starts =================================================================================== platform linux -- Python 3.9.19, pytest-7.3.2, pluggy-1.4.0 rootdir: /dockerx/development/pytorch configfile: pytest.ini plugins: flakefinder-1.1.0, rerunfailures-14.0, xdist-3.3.1, xdoctest-1.1.0, cpp-2.3.0, hypothesis-5.35.1 collected 288 items / 287 deselected / 1 selected Running 1 items in this shard test/nn/test_embedding.py . [100%] =========================================================================== 1 passed, 287 deselected in 3.16s ============================================================================ ``` Pull Request resolved: pytorch#130994 Approved by: https://github.com/jeffdaily, https://github.com/xw285cornell (cherry picked from commit 637ab85) Co-authored-by: hongxyan <hongxyan@amd.com>
40 tasks
|
Verified with PyTorch2.4.1 final RC wheels on MI100/MI210: |
# for free
to join this conversation on GitHub.
Already have an account?
# to comment
Uh oh!
There was an error while loading. Please reload this page.
🐛 Describe the bug
F.embedding will crash with relatively large tensor input on the AMD GPU:
Versions
top of tree
cc @jeffdaily @sunway513 @jithunnair-amd @pruthvistony @ROCmSupport @dllehr-amd @jataylo @hongxiayang
The text was updated successfully, but these errors were encountered: