Cuda runtime error (8) : invalid device function at THCTensorMathPointwise.cu for latest build

dendisuhubdy · 2017-06-30T01:22:27Z

So I've tried to compile Pytorch locally for development purposes under the (dev) anaconda environment and another one under (lisa) which I install from the instructions from pytorch.org, I've got a recent warning when running the VAE code on Pytorch tutorials when running it on the newest version.

(dev) [SLURM] suhubdyd@kepler2:~/research/models/pytorch-models/vae$ python main.py 
THCudaCheck FAIL file=/u/suhubdyd/research/dl-frameworks/pytorch/torch/lib/THC/generated/../generic/THCTensorMathPointwise.cu line=247 error=8 : invalid device function
Traceback (most recent call last):
  File "main.py", line 138, in <module>
    train(epoch)
  File "main.py", line 108, in train
    recon_batch, mu, logvar = model(data)
  File "/u/suhubdyd/.conda/envs/dev/lib/python2.7/site-packages/torch/nn/modules/module.py", line 206, in __call__
    result = self.forward(*input, **kwargs)
  File "main.py", line 71, in forward
    mu, logvar = self.encode(x.view(-1, 784))
  File "main.py", line 54, in encode
    h1 = self.relu(self.fc1(x))
  File "/u/suhubdyd/.conda/envs/dev/lib/python2.7/site-packages/torch/nn/modules/module.py", line 206, in __call__
    result = self.forward(*input, **kwargs)
  File "/u/suhubdyd/.conda/envs/dev/lib/python2.7/site-packages/torch/nn/modules/linear.py", line 54, in forward
    return self._backend.Linear.apply(input, self.weight, self.bias)
  File "/u/suhubdyd/.conda/envs/dev/lib/python2.7/site-packages/torch/nn/_functions/linear.py", line 14, in forward
    output.add_(bias.expand_as(output))
RuntimeError: cuda runtime error (8) : invalid device function at /u/suhubdyd/research/dl-frameworks/pytorch/torch/lib/THC/generated/../generic/THCTensorMathPointwise.cu:247

while it doesn't matter for the stable distribution

(lisa) [SLURM] suhubdyd@kepler2:~/research/models/pytorch-models/vae$ python main.py 
Train Epoch: 1 [0/60000 (0%)]   Loss: 549.307800
Train Epoch: 1 [1280/60000 (2%)]        Loss: 311.565613
Train Epoch: 1 [2560/60000 (4%)]        Loss: 236.338989
Train Epoch: 1 [3840/60000 (6%)]        Loss: 225.371078
Train Epoch: 1 [5120/60000 (9%)]        Loss: 208.993668
Train Epoch: 1 [6400/60000 (11%)]       Loss: 206.427368
Train Epoch: 1 [7680/60000 (13%)]       Loss: 208.580597
Train Epoch: 1 [8960/60000 (15%)]       Loss: 201.324646
Train Epoch: 1 [10240/60000 (17%)]      Loss: 191.824326
Train Epoch: 1 [11520/60000 (19%)]      Loss: 195.214188
Train Epoch: 1 [12800/60000 (21%)]      Loss: 189.770447
Train Epoch: 1 [14080/60000 (23%)]      Loss: 173.119644
Train Epoch: 1 [15360/60000 (26%)]      Loss: 179.030197
Train Epoch: 1 [16640/60000 (28%)]      Loss: 170.247345
Train Epoch: 1 [17920/60000 (30%)]      Loss: 169.193451
Train Epoch: 1 [19200/60000 (32%)]      Loss: 162.828690
Train Epoch: 1 [20480/60000 (34%)]      Loss: 158.171326
Train Epoch: 1 [21760/60000 (36%)]      Loss: 158.530518
Train Epoch: 1 [23040/60000 (38%)]      Loss: 155.896255
Train Epoch: 1 [24320/60000 (41%)]      Loss: 158.835968
Train Epoch: 1 [25600/60000 (43%)]      Loss: 152.416977
Train Epoch: 1 [26880/60000 (45%)]      Loss: 153.593964
Train Epoch: 1 [28160/60000 (47%)]      Loss: 147.944260
Train Epoch: 1 [29440/60000 (49%)]      Loss: 148.223892
Train Epoch: 1 [30720/60000 (51%)]      Loss: 145.770905
Train Epoch: 1 [32000/60000 (53%)]      Loss: 144.410706
Train Epoch: 1 [33280/60000 (55%)]      Loss: 147.592163
Train Epoch: 1 [34560/60000 (58%)]      Loss: 149.320328

The text was updated successfully, but these errors were encountered:

pavanky · 2017-06-30T04:04:14Z

Did you compile pytorch on one machine and install it on another?

dendisuhubdy · 2017-06-30T04:26:14Z

@pavanky yes, I did a python setup.py install one machine, and I did a conda install pytorch torchvision cuda80 -c soumith on another, if that is what you mean.

pavanky · 2017-06-30T04:49:51Z

@dendisuhubdy The error you mentioned happens when you build a cuda application using the wrong gpu architecture. I am not sure how that might happen in your case. What GPU are you using?

dendisuhubdy · 2017-06-30T21:44:10Z

@pavanky Quadro P6000

ngimel · 2017-06-30T22:11:40Z

As @pavanky said, this error would appear if you compiled for one architecture, and deploy on another. It should not happen if you compile and run on the same machine (unless you have stale build directories where GPU architecture is somehow set incorrectly). Try removing you build directories (python setup.py clean) and recompiling. Also, make sure that you use CUDA 8, Pascal is not supported in earlier CUDA versions.

pavanky · 2017-06-30T22:15:30Z

@dendisuhubdy Try to get the latest version of CUDA 8. @ngimel correct me if I am wrong, but I think CUDA 8 pre-release (8.0.26) didn't support Pascal

ngimel · 2017-06-30T22:21:02Z

@pavanky, it supported Pascal, but as you've pointed out in another issue, compilation with CUDA 8 pre-release is still broken.

xuyifeng-nwpu · 2017-07-05T15:24:39Z

@pavanky @ngimel As you said, I encounted the same ploblem .
My host's GPU is 980ti, while my other server's GPUs are k80. When I transfer my host's docker file to server and run the torch code， the same error message show. Can you solve the ploblem between two different GPU？

soumith · 2017-07-05T15:34:26Z

@xuyifeng-nwpu you can set this environment variable before compiling PyTorch in your docker:

export TORCH_CUDA_ARCH_LIST="3.0;3.5;5.0;5.2+PTX;6.0;6.1"

It will take much longer to compile, but the docker will work everywhere.

xuyifeng-nwpu · 2017-07-06T00:47:18Z

@soumith Thank your reply.
1:My code was not PyTorch but Torch. I worry that the code of Torch cannot run in the new Pytorch envieroment, so I havn't install PyTorch. Do I need install the PyTorch? Is the Torch's code compatible with the PyTorch?
2:Where and when run the code export TORCH_CUDA_ARCH_LIST="3.0;3.5;5.0;5.2+PTX;6.0;6.1 ?
In the docker's container or the host through sudo gedit .bashrc?
What is compiling PyTorch? Is compiling PyTorch reinstalling the Torch ?
Forgive me as a novice.
3: K80's CC is 3.7. Do the numbur 3.7 add to code as
export TORCH_CUDA_ARCH_LIST="3.0;3.5;3.7;5.0;5.2+PTX;6.0;6.1

pavanky · 2017-07-06T02:13:50Z

@xuyifeng-nwpu you only need to export it when you are building Torch / PyTorch. This is not needed during the runtime. And yes you will need to have 3.7 for k80. If you only care about 980 and K80, you can probably just use export TORCH_CUDA_ARCH_LIST="3.7;5.2+PTX;" to reduce the build time.

dendisuhubdy · 2017-07-06T02:20:04Z

@pavanky @xuyifeng-nwpu @soumith anything on my side that I could do to help you guys?

xuyifeng-nwpu · 2017-07-06T02:25:37Z

@pavanky @dendisuhubdy @soumith thank your quick reply.
1:How to "building Torch" ?
Code1: Reinstalling the Torch (luarocks install torch) or
Code2: luarocks install cutorch luarocks install cunn?

2:K80 Tesla,3.7. 980ti : maxwell 5.2.
Is right the code "export TORCH_CUDA_ARCH_LIST="3.7;5.2+PTX'?
Do the word "ptx" represents Pascal, Tesla and Mmaxwell?

3:Where can I learn the grammar of 'TORCH_CUDA_ARCH_LIST'?

4:Where run the export code?
In the host or the docker's container?

Thanks again

soumith · 2017-07-13T06:28:44Z

as everyone on this thread pretty much said, @dendisuhubdy this is pretty much because you are compiling from source on one machine, and running the source-installation on another machine. My workaround is in the comments above. Closing the issue as there's no action to be taken from the pytorch devs side.

ysnan · 2018-11-19T05:24:55Z

@soumith Thank you for analysis the problem.
I also encounter similar problems with slightly different settings. I try to use nn.DataParallel to parallel computings two different GPUs.
I have two GPUs. One is GeForce GTX 1080 Ti with CC 6.1, and another is Quadro M5000 with CC 5.2.
With setting export TORCH_CUDA_ARCH_LIST="6.1;5.2+PTX; , the same error pops out. Could you please help to analyse the scenario within data parallel.

soumith · 2018-12-13T06:45:25Z

@YuesongNan even within DataParallel the situation should be the same. I think you might have multiple pytorch versions installed.

…3651cd (pytorch#38203) Summary: Pull Request resolved: pytorch#38203 Previous import was 79a7e0df7e86e0f32e7a05f563b24a566540c18b Included changes: - **[20b3e10e](onnx/onnx@20b3e10e)**: Add 'ignore_index' input in the spec for SoftmaxCrossEntropyLoss and NLLLoss. (pytorch#2680) <M. Zeeshan Siddiqui> - **[eae3eb8c](onnx/onnx@eae3eb8c)**: Use cmake GNUInstallDirs (pytorch#2661) <Gustavo Alvarez> - **[106821e9](onnx/onnx@106821e9)**: Update sequence test case so input is not scalar and splits are specified (pytorch#2675) <Scott McKay> - **[e094e101](onnx/onnx@e094e101)**: Remove unnecessary copies and std::move (pytorch#2684) <Changming Sun> - **[71145275](onnx/onnx@71145275)**: Update Batchnorm test (pytorch#2674) <Lara Haidar> - **[da13be2d](onnx/onnx@da13be2d)**: Rename OPTIONAL to OPTIONAL_VALUE (pytorch#2682) <Changming Sun> - **[2987fa06](onnx/onnx@2987fa06)**: Adding CI for ONNX Debug mode (Linux, OSX) (pytorch#2651) <Vinitra Swamy> - **[46fe392d](onnx/onnx@46fe392d)**: Update Pow input types in Opset 12 (pytorch#2666) <Lara Haidar> - **[ac1caf3b](onnx/onnx@ac1caf3b)**: Change type of label tensor to int32/int64 in SoftmaxCrossEntropyLoss spec. (pytorch#2667) <M. Zeeshan Siddiqui> - **[c2fefcbf](onnx/onnx@c2fefcbf)**: [Training] SG with Momentum Optimizer (pytorch#1959) <Wei-Sheng Chin> - **[8d15705e](onnx/onnx@8d15705e)**: [Training] Add Adagrad optimizer operator (pytorch#1955) <Wei-Sheng Chin> - **[94b01cdd](onnx/onnx@94b01cdd)**: Suppress a warning in unsqueeze (pytorch#2637) <Hong Xu> - **[0582d526](onnx/onnx@0582d526)**: Fix Greater/LessOrEqual function definition (pytorch#2645) <Takeshi Watanabe> - **[b852d819](onnx/onnx@b852d819)**: Increment version number to 1.7.0 (pytorch#2639) <Chin Huang> - **[ff4bb553](onnx/onnx@ff4bb553)**: Regenerate Min test data (pytorch#2644) <Takeshi Watanabe> Test Plan: ci Reviewed By: hl475 Differential Revision: D21493165 fbshipit-source-id: c5eb2f0143f1763166f46f118ec0985b41f759b3

…3651cd (#38203) Summary: Pull Request resolved: #38203 Previous import was 79a7e0df7e86e0f32e7a05f563b24a566540c18b Included changes: - **[20b3e10e](onnx/onnx@20b3e10e)**: Add 'ignore_index' input in the spec for SoftmaxCrossEntropyLoss and NLLLoss. (#2680) <M. Zeeshan Siddiqui> - **[eae3eb8c](onnx/onnx@eae3eb8c)**: Use cmake GNUInstallDirs (#2661) <Gustavo Alvarez> - **[106821e9](onnx/onnx@106821e9)**: Update sequence test case so input is not scalar and splits are specified (#2675) <Scott McKay> - **[e094e101](onnx/onnx@e094e101)**: Remove unnecessary copies and std::move (#2684) <Changming Sun> - **[71145275](onnx/onnx@71145275)**: Update Batchnorm test (#2674) <Lara Haidar> - **[da13be2d](onnx/onnx@da13be2d)**: Rename OPTIONAL to OPTIONAL_VALUE (#2682) <Changming Sun> - **[2987fa06](onnx/onnx@2987fa06)**: Adding CI for ONNX Debug mode (Linux, OSX) (#2651) <Vinitra Swamy> - **[46fe392d](onnx/onnx@46fe392d)**: Update Pow input types in Opset 12 (#2666) <Lara Haidar> - **[ac1caf3b](onnx/onnx@ac1caf3b)**: Change type of label tensor to int32/int64 in SoftmaxCrossEntropyLoss spec. (#2667) <M. Zeeshan Siddiqui> - **[c2fefcbf](onnx/onnx@c2fefcbf)**: [Training] SG with Momentum Optimizer (#1959) <Wei-Sheng Chin> - **[8d15705e](onnx/onnx@8d15705e)**: [Training] Add Adagrad optimizer operator (#1955) <Wei-Sheng Chin> - **[94b01cdd](onnx/onnx@94b01cdd)**: Suppress a warning in unsqueeze (#2637) <Hong Xu> - **[0582d526](onnx/onnx@0582d526)**: Fix Greater/LessOrEqual function definition (#2645) <Takeshi Watanabe> - **[b852d819](onnx/onnx@b852d819)**: Increment version number to 1.7.0 (#2639) <Chin Huang> - **[ff4bb553](onnx/onnx@ff4bb553)**: Regenerate Min test data (#2644) <Takeshi Watanabe> Test Plan: ci Reviewed By: hl475 Differential Revision: D21493165 fbshipit-source-id: 6863b289bfbf4235e36f0e2456ce44c776aaf164

…2385a6 Summary: Previous import was 79a7e0df7e86e0f32e7a05f563b24a566540c18b Included changes: - **[eae3eb8c](onnx/onnx@eae3eb8c)**: Use cmake GNUInstallDirs (pytorch#2661) <Gustavo Alvarez> - **[106821e9](onnx/onnx@106821e9)**: Update sequence test case so input is not scalar and splits are specified (pytorch#2675) <Scott McKay> - **[e094e101](onnx/onnx@e094e101)**: Remove unnecessary copies and std::move (pytorch#2684) <Changming Sun> - **[71145275](onnx/onnx@71145275)**: Update Batchnorm test (pytorch#2674) <Lara Haidar> - **[da13be2d](onnx/onnx@da13be2d)**: Rename OPTIONAL to OPTIONAL_VALUE (pytorch#2682) <Changming Sun> - **[2987fa06](onnx/onnx@2987fa06)**: Adding CI for ONNX Debug mode (Linux, OSX) (pytorch#2651) <Vinitra Swamy> - **[46fe392d](onnx/onnx@46fe392d)**: Update Pow input types in Opset 12 (pytorch#2666) <Lara Haidar> - **[ac1caf3b](onnx/onnx@ac1caf3b)**: Change type of label tensor to int32/int64 in SoftmaxCrossEntropyLoss spec. (pytorch#2667) <M. Zeeshan Siddiqui> - **[c2fefcbf](onnx/onnx@c2fefcbf)**: [Training] SG with Momentum Optimizer (pytorch#1959) <Wei-Sheng Chin> - **[8d15705e](onnx/onnx@8d15705e)**: [Training] Add Adagrad optimizer operator (pytorch#1955) <Wei-Sheng Chin> - **[94b01cdd](onnx/onnx@94b01cdd)**: Suppress a warning in unsqueeze (pytorch#2637) <Hong Xu> - **[0582d526](onnx/onnx@0582d526)**: Fix Greater/LessOrEqual function definition (pytorch#2645) <Takeshi Watanabe> - **[b852d819](onnx/onnx@b852d819)**: Increment version number to 1.7.0 (pytorch#2639) <Chin Huang> - **[ff4bb553](onnx/onnx@ff4bb553)**: Regenerate Min test data (pytorch#2644) <Takeshi Watanabe> Test Plan: ci Differential Revision: D21750299 fbshipit-source-id: ba52d96806b70fd4ff56a9743b5e56879ecf2b88

…2385a6 (#39089) Summary: Pull Request resolved: #39089 Previous import was 79a7e0df7e86e0f32e7a05f563b24a566540c18b Included changes: - **[eae3eb8c](onnx/onnx@eae3eb8c)**: Use cmake GNUInstallDirs (#2661) <Gustavo Alvarez> - **[106821e9](onnx/onnx@106821e9)**: Update sequence test case so input is not scalar and splits are specified (#2675) <Scott McKay> - **[e094e101](onnx/onnx@e094e101)**: Remove unnecessary copies and std::move (#2684) <Changming Sun> - **[71145275](onnx/onnx@71145275)**: Update Batchnorm test (#2674) <Lara Haidar> - **[da13be2d](onnx/onnx@da13be2d)**: Rename OPTIONAL to OPTIONAL_VALUE (#2682) <Changming Sun> - **[2987fa06](onnx/onnx@2987fa06)**: Adding CI for ONNX Debug mode (Linux, OSX) (#2651) <Vinitra Swamy> - **[46fe392d](onnx/onnx@46fe392d)**: Update Pow input types in Opset 12 (#2666) <Lara Haidar> - **[ac1caf3b](onnx/onnx@ac1caf3b)**: Change type of label tensor to int32/int64 in SoftmaxCrossEntropyLoss spec. (#2667) <M. Zeeshan Siddiqui> - **[c2fefcbf](onnx/onnx@c2fefcbf)**: [Training] SG with Momentum Optimizer (#1959) <Wei-Sheng Chin> - **[8d15705e](onnx/onnx@8d15705e)**: [Training] Add Adagrad optimizer operator (#1955) <Wei-Sheng Chin> - **[94b01cdd](onnx/onnx@94b01cdd)**: Suppress a warning in unsqueeze (#2637) <Hong Xu> - **[0582d526](onnx/onnx@0582d526)**: Fix Greater/LessOrEqual function definition (#2645) <Takeshi Watanabe> - **[b852d819](onnx/onnx@b852d819)**: Increment version number to 1.7.0 (#2639) <Chin Huang> - **[ff4bb553](onnx/onnx@ff4bb553)**: Regenerate Min test data (#2644) <Takeshi Watanabe> Test Plan: ci Reviewed By: hl475 Differential Revision: D21750299 fbshipit-source-id: c33ec1b1e0dc65d0187e78db96d749f9037aae9c

7117a7e patching nvfuser conv cudnn test numerics mismatch (#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (#2023) 6ac74d1 Fix sync map (#2047) f5bca33 Bank conflict checker improvements (#2032) d2ca7e3 Minor update on cp.async code generation. (#1901) d36cf61 Test file cleanup (#2040) 0b8e83f Allow non-root trivial reductions (#2037) a2dfe40 Fix vectorize size calculation (#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (#2025) 197221b removing ci workflow (#2034) 40e2703 Reduction rand like patch (#2031) bc77266 Add utility for checking bank conflict of shared memory (#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds (#2006)" (#2024) bca20c1 Cleanup trivial reduction workarounds (#2006) e4b6585 Trivial forwarding (#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (#1991) a4effa6 Enable output allocation cache (#2010) 35440b7 Patching bn inference (#2016) 0f9f0b4 Add matmul benchmark (#2007) 45045cd Enable tests previously disabled due to an aliasing bug (#2005) 967aa77 Contiguous indexing for View operations (#1990) a43cb20 Make inlining even more modular (#2004) dc45835 Test util cleanup (#2003) 3ca21eb More strict validation (#2000) a7a7d57 Fix build problem (#1999) fc235b0 Just fixes comments (#1998) 482386c cleanup (#1997) 4cbe0db Improve divisible split detection (#1970) 42ccc52 Minor build fix. (#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (#1994) 1d9858c Minor cleanup (#1992) f262d9c Add support for uniform RNG (#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (#1987) 634820c Add support for some empty fusion (#1981) eabe8d8 Segment self mapping fusions (#1954) e96aacf Enable Transpose operation (#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (#1969) b1bd32c Minor fix (#1967) bd93578 Enable transpose scheduler (#1927) b7a206e Move scheduler vectorize utilities into their own file (#1959) d9420e4 View scheduling (#1928) c668e13 Upstream push ci fixes (#1965) c40202b Fix dump effective bandwidth (#1962) 93505bc WAR on index mapping when exact and permissive maps differ (#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (#1946) f7bc341 Remove unused variables (#1955) df3393a Some cleanup (#1957) 7d1d7c8 TVDomainGuard factory (#1953) 357ba22 Fill allocation with nan on tests (#1956) 8eafc54 Fix detection of unmappable root domains (#1952) 90a51f2 Some indexing cleanups, Add eye support (#1940) ddc01e4 Exclude unsupported data types (#1951) 992e17c test the groups the same order as they are merged (#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (#1941) ac4de38 Merge pull request #1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (#1942) 89330aa Tensor factories must set the output shape as its input (#1939) [ghstack-poisoned]

7117a7e patching nvfuser conv cudnn test numerics mismatch (#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (#2023) 6ac74d1 Fix sync map (#2047) f5bca33 Bank conflict checker improvements (#2032) d2ca7e3 Minor update on cp.async code generation. (#1901) d36cf61 Test file cleanup (#2040) 0b8e83f Allow non-root trivial reductions (#2037) a2dfe40 Fix vectorize size calculation (#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (#2025) 197221b removing ci workflow (#2034) 40e2703 Reduction rand like patch (#2031) bc77266 Add utility for checking bank conflict of shared memory (#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds (#2006)" (#2024) bca20c1 Cleanup trivial reduction workarounds (#2006) e4b6585 Trivial forwarding (#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (#1991) a4effa6 Enable output allocation cache (#2010) 35440b7 Patching bn inference (#2016) 0f9f0b4 Add matmul benchmark (#2007) 45045cd Enable tests previously disabled due to an aliasing bug (#2005) 967aa77 Contiguous indexing for View operations (#1990) a43cb20 Make inlining even more modular (#2004) dc45835 Test util cleanup (#2003) 3ca21eb More strict validation (#2000) a7a7d57 Fix build problem (#1999) fc235b0 Just fixes comments (#1998) 482386c cleanup (#1997) 4cbe0db Improve divisible split detection (#1970) 42ccc52 Minor build fix. (#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (#1994) 1d9858c Minor cleanup (#1992) f262d9c Add support for uniform RNG (#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (#1987) 634820c Add support for some empty fusion (#1981) eabe8d8 Segment self mapping fusions (#1954) e96aacf Enable Transpose operation (#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (#1969) b1bd32c Minor fix (#1967) bd93578 Enable transpose scheduler (#1927) b7a206e Move scheduler vectorize utilities into their own file (#1959) d9420e4 View scheduling (#1928) c668e13 Upstream push ci fixes (#1965) c40202b Fix dump effective bandwidth (#1962) 93505bc WAR on index mapping when exact and permissive maps differ (#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (#1946) f7bc341 Remove unused variables (#1955) df3393a Some cleanup (#1957) 7d1d7c8 TVDomainGuard factory (#1953) 357ba22 Fill allocation with nan on tests (#1956) 8eafc54 Fix detection of unmappable root domains (#1952) 90a51f2 Some indexing cleanups, Add eye support (#1940) ddc01e4 Exclude unsupported data types (#1951) 992e17c test the groups the same order as they are merged (#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (#1941) ac4de38 Merge pull request #1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (#1942) 89330aa Tensor factories must set the output shape as its input (#1939) ghstack-source-id: 7ebfc36 Pull Request resolved: #87779

Syncing nvfuser devel branch to upstream master. https://github.com/csarofeen/pytorch/ Codegen changes include (TBD): Commits that's in this PR from the devel branch: ``` 7117a7e patching nvfuser conv cudnn test numerics mismatch (#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (#2023) 6ac74d1 Fix sync map (#2047) f5bca33 Bank conflict checker improvements (#2032) d2ca7e3 Minor update on cp.async code generation. (#1901) d36cf61 Test file cleanup (#2040) 0b8e83f Allow non-root trivial reductions (#2037) a2dfe40 Fix vectorize size calculation (#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (#2025) 197221b removing ci workflow (#2034) 40e2703 Reduction rand like patch (#2031) bc77266 Add utility for checking bank conflict of shared memory (#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds (#2006)" (#2024) bca20c1 Cleanup trivial reduction workarounds (#2006) e4b6585 Trivial forwarding (#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (#1991) a4effa6 Enable output allocation cache (#2010) 35440b7 Patching bn inference (#2016) 0f9f0b4 Add matmul benchmark (#2007) 45045cd Enable tests previously disabled due to an aliasing bug (#2005) 967aa77 Contiguous indexing for View operations (#1990) a43cb20 Make inlining even more modular (#2004) dc45835 Test util cleanup (#2003) 3ca21eb More strict validation (#2000) a7a7d57 Fix build problem (#1999) fc235b0 Just fixes comments (#1998) 482386c cleanup (#1997) 4cbe0db Improve divisible split detection (#1970) 42ccc52 Minor build fix. (#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (#1994) 1d9858c Minor cleanup (#1992) f262d9c Add support for uniform RNG (#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (#1987) 634820c Add support for some empty fusion (#1981) eabe8d8 Segment self mapping fusions (#1954) e96aacf Enable Transpose operation (#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (#1969) b1bd32c Minor fix (#1967) bd93578 Enable transpose scheduler (#1927) b7a206e Move scheduler vectorize utilities into their own file (#1959) d9420e4 View scheduling (#1928) c668e13 Upstream push ci fixes (#1965) c40202b Fix dump effective bandwidth (#1962) 93505bc WAR on index mapping when exact and permissive maps differ (#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (#1946) f7bc341 Remove unused variables (#1955) df3393a Some cleanup (#1957) 7d1d7c8 TVDomainGuard factory (#1953) 357ba22 Fill allocation with nan on tests (#1956) 8eafc54 Fix detection of unmappable root domains (#1952) 90a51f2 Some indexing cleanups, Add eye support (#1940) ddc01e4 Exclude unsupported data types (#1951) 992e17c test the groups the same order as they are merged (#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (#1941) ac4de38 Merge pull request #1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (#1942) 89330aa Tensor factories must set the output shape as its input (#1939) ``` [ghstack-poisoned]

7117a7e patching nvfuser conv cudnn test numerics mismatch (#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (#2023) 6ac74d1 Fix sync map (#2047) f5bca33 Bank conflict checker improvements (#2032) d2ca7e3 Minor update on cp.async code generation. (#1901) d36cf61 Test file cleanup (#2040) 0b8e83f Allow non-root trivial reductions (#2037) a2dfe40 Fix vectorize size calculation (#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (#2025) 197221b removing ci workflow (#2034) 40e2703 Reduction rand like patch (#2031) bc77266 Add utility for checking bank conflict of shared memory (#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds (#2006)" (#2024) bca20c1 Cleanup trivial reduction workarounds (#2006) e4b6585 Trivial forwarding (#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (#1991) a4effa6 Enable output allocation cache (#2010) 35440b7 Patching bn inference (#2016) 0f9f0b4 Add matmul benchmark (#2007) 45045cd Enable tests previously disabled due to an aliasing bug (#2005) 967aa77 Contiguous indexing for View operations (#1990) a43cb20 Make inlining even more modular (#2004) dc45835 Test util cleanup (#2003) 3ca21eb More strict validation (#2000) a7a7d57 Fix build problem (#1999) fc235b0 Just fixes comments (#1998) 482386c cleanup (#1997) 4cbe0db Improve divisible split detection (#1970) 42ccc52 Minor build fix. (#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (#1994) 1d9858c Minor cleanup (#1992) f262d9c Add support for uniform RNG (#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (#1987) 634820c Add support for some empty fusion (#1981) eabe8d8 Segment self mapping fusions (#1954) e96aacf Enable Transpose operation (#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (#1969) b1bd32c Minor fix (#1967) bd93578 Enable transpose scheduler (#1927) b7a206e Move scheduler vectorize utilities into their own file (#1959) d9420e4 View scheduling (#1928) c668e13 Upstream push ci fixes (#1965) c40202b Fix dump effective bandwidth (#1962) 93505bc WAR on index mapping when exact and permissive maps differ (#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (#1946) f7bc341 Remove unused variables (#1955) df3393a Some cleanup (#1957) 7d1d7c8 TVDomainGuard factory (#1953) 357ba22 Fill allocation with nan on tests (#1956) 8eafc54 Fix detection of unmappable root domains (#1952) 90a51f2 Some indexing cleanups, Add eye support (#1940) ddc01e4 Exclude unsupported data types (#1951) 992e17c test the groups the same order as they are merged (#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (#1941) ac4de38 Merge pull request #1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (#1942) 89330aa Tensor factories must set the output shape as its input (#1939) ghstack-source-id: f05ebc3 Pull Request resolved: #87779

Syncing nvfuser devel branch to upstream master. https://github.com/csarofeen/pytorch/ Codegen changes include (TBD): Commits that's in this PR from the devel branch: ``` 7117a7e patching nvfuser conv cudnn test numerics mismatch (#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (#2023) 6ac74d1 Fix sync map (#2047) f5bca33 Bank conflict checker improvements (#2032) d2ca7e3 Minor update on cp.async code generation. (#1901) d36cf61 Test file cleanup (#2040) 0b8e83f Allow non-root trivial reductions (#2037) a2dfe40 Fix vectorize size calculation (#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (#2025) 197221b removing ci workflow (#2034) 40e2703 Reduction rand like patch (#2031) bc77266 Add utility for checking bank conflict of shared memory (#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds (#2006)" (#2024) bca20c1 Cleanup trivial reduction workarounds (#2006) e4b6585 Trivial forwarding (#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (#1991) a4effa6 Enable output allocation cache (#2010) 35440b7 Patching bn inference (#2016) 0f9f0b4 Add matmul benchmark (#2007) 45045cd Enable tests previously disabled due to an aliasing bug (#2005) 967aa77 Contiguous indexing for View operations (#1990) a43cb20 Make inlining even more modular (#2004) dc45835 Test util cleanup (#2003) 3ca21eb More strict validation (#2000) a7a7d57 Fix build problem (#1999) fc235b0 Just fixes comments (#1998) 482386c cleanup (#1997) 4cbe0db Improve divisible split detection (#1970) 42ccc52 Minor build fix. (#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (#1994) 1d9858c Minor cleanup (#1992) f262d9c Add support for uniform RNG (#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (#1987) 634820c Add support for some empty fusion (#1981) eabe8d8 Segment self mapping fusions (#1954) e96aacf Enable Transpose operation (#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (#1969) b1bd32c Minor fix (#1967) bd93578 Enable transpose scheduler (#1927) b7a206e Move scheduler vectorize utilities into their own file (#1959) d9420e4 View scheduling (#1928) c668e13 Upstream push ci fixes (#1965) c40202b Fix dump effective bandwidth (#1962) 93505bc WAR on index mapping when exact and permissive maps differ (#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (#1946) f7bc341 Remove unused variables (#1955) df3393a Some cleanup (#1957) 7d1d7c8 TVDomainGuard factory (#1953) 357ba22 Fill allocation with nan on tests (#1956) 8eafc54 Fix detection of unmappable root domains (#1952) 90a51f2 Some indexing cleanups, Add eye support (#1940) ddc01e4 Exclude unsupported data types (#1951) 992e17c test the groups the same order as they are merged (#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (#1941) ac4de38 Merge pull request #1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (#1942) 89330aa Tensor factories must set the output shape as its input (#1939) ``` [ghstack-poisoned]

Syncing nvfuser devel branch to upstream master. https://github.com/csarofeen/pytorch/ Codegen changes include: * codegen improvement: i. allow non-root trivial reductions, allow empty/no-op fusion ii. fixes vectorization checks and size calculation iii. bank conflict handle improvement iv. enables transpose scheduler * misc: i. CI tests failure fixes ii. cpp tests file clean up iii. trivial forwarding supports added in codegen runtime iv. added factory methods support in codegen Commits that's in this PR from the devel branch: ``` 7117a7e patching nvfuser conv cudnn test numerics mismatch (#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (#2023) 6ac74d1 Fix sync map (#2047) f5bca33 Bank conflict checker improvements (#2032) d2ca7e3 Minor update on cp.async code generation. (#1901) d36cf61 Test file cleanup (#2040) 0b8e83f Allow non-root trivial reductions (#2037) a2dfe40 Fix vectorize size calculation (#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (#2025) 197221b removing ci workflow (#2034) 40e2703 Reduction rand like patch (#2031) bc77266 Add utility for checking bank conflict of shared memory (#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds (#2006)" (#2024) bca20c1 Cleanup trivial reduction workarounds (#2006) e4b6585 Trivial forwarding (#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (#1991) a4effa6 Enable output allocation cache (#2010) 35440b7 Patching bn inference (#2016) 0f9f0b4 Add matmul benchmark (#2007) 45045cd Enable tests previously disabled due to an aliasing bug (#2005) 967aa77 Contiguous indexing for View operations (#1990) a43cb20 Make inlining even more modular (#2004) dc45835 Test util cleanup (#2003) 3ca21eb More strict validation (#2000) a7a7d57 Fix build problem (#1999) fc235b0 Just fixes comments (#1998) 482386c cleanup (#1997) 4cbe0db Improve divisible split detection (#1970) 42ccc52 Minor build fix. (#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (#1994) 1d9858c Minor cleanup (#1992) f262d9c Add support for uniform RNG (#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (#1987) 634820c Add support for some empty fusion (#1981) eabe8d8 Segment self mapping fusions (#1954) e96aacf Enable Transpose operation (#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (#1969) b1bd32c Minor fix (#1967) bd93578 Enable transpose scheduler (#1927) b7a206e Move scheduler vectorize utilities into their own file (#1959) d9420e4 View scheduling (#1928) c668e13 Upstream push ci fixes (#1965) c40202b Fix dump effective bandwidth (#1962) 93505bc WAR on index mapping when exact and permissive maps differ (#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (#1946) f7bc341 Remove unused variables (#1955) df3393a Some cleanup (#1957) 7d1d7c8 TVDomainGuard factory (#1953) 357ba22 Fill allocation with nan on tests (#1956) 8eafc54 Fix detection of unmappable root domains (#1952) 90a51f2 Some indexing cleanups, Add eye support (#1940) ddc01e4 Exclude unsupported data types (#1951) 992e17c test the groups the same order as they are merged (#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (#1941) ac4de38 Merge pull request #1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (#1942) 89330aa Tensor factories must set the output shape as its input (#1939) ``` RUN_TORCHBENCH: nvfuser Differential Revision: [D40869846](https://our.internmc.facebook.com/intern/diff/D40869846) [ghstack-poisoned]

7117a7e patching nvfuser conv cudnn test numerics mismatch (#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (#2023) 6ac74d1 Fix sync map (#2047) f5bca33 Bank conflict checker improvements (#2032) d2ca7e3 Minor update on cp.async code generation. (#1901) d36cf61 Test file cleanup (#2040) 0b8e83f Allow non-root trivial reductions (#2037) a2dfe40 Fix vectorize size calculation (#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (#2025) 197221b removing ci workflow (#2034) 40e2703 Reduction rand like patch (#2031) bc77266 Add utility for checking bank conflict of shared memory (#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds (#2006)" (#2024) bca20c1 Cleanup trivial reduction workarounds (#2006) e4b6585 Trivial forwarding (#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (#1991) a4effa6 Enable output allocation cache (#2010) 35440b7 Patching bn inference (#2016) 0f9f0b4 Add matmul benchmark (#2007) 45045cd Enable tests previously disabled due to an aliasing bug (#2005) 967aa77 Contiguous indexing for View operations (#1990) a43cb20 Make inlining even more modular (#2004) dc45835 Test util cleanup (#2003) 3ca21eb More strict validation (#2000) a7a7d57 Fix build problem (#1999) fc235b0 Just fixes comments (#1998) 482386c cleanup (#1997) 4cbe0db Improve divisible split detection (#1970) 42ccc52 Minor build fix. (#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (#1994) 1d9858c Minor cleanup (#1992) f262d9c Add support for uniform RNG (#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (#1987) 634820c Add support for some empty fusion (#1981) eabe8d8 Segment self mapping fusions (#1954) e96aacf Enable Transpose operation (#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (#1969) b1bd32c Minor fix (#1967) bd93578 Enable transpose scheduler (#1927) b7a206e Move scheduler vectorize utilities into their own file (#1959) d9420e4 View scheduling (#1928) c668e13 Upstream push ci fixes (#1965) c40202b Fix dump effective bandwidth (#1962) 93505bc WAR on index mapping when exact and permissive maps differ (#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (#1946) f7bc341 Remove unused variables (#1955) df3393a Some cleanup (#1957) 7d1d7c8 TVDomainGuard factory (#1953) 357ba22 Fill allocation with nan on tests (#1956) 8eafc54 Fix detection of unmappable root domains (#1952) 90a51f2 Some indexing cleanups, Add eye support (#1940) ddc01e4 Exclude unsupported data types (#1951) 992e17c test the groups the same order as they are merged (#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (#1941) ac4de38 Merge pull request #1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (#1942) 89330aa Tensor factories must set the output shape as its input (#1939) RUN_TORCHBENCH: nvfuser ghstack-source-id: 3b21376 Pull Request resolved: #87779

Syncing nvfuser devel branch to upstream master. https://github.com/csarofeen/pytorch/ Codegen changes include: * codegen improvement: i. allow non-root trivial reductions, allow empty/no-op fusion ii. fixes vectorization checks and size calculation iii. bank conflict handle improvement iv. enables transpose scheduler * misc: i. CI tests failure fixes ii. cpp tests file clean up iii. trivial forwarding supports added in codegen runtime iv. added factory methods support in codegen Commits that's in this PR from the devel branch: ``` 7117a7e patching nvfuser conv cudnn test numerics mismatch (#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (#2023) 6ac74d1 Fix sync map (#2047) f5bca33 Bank conflict checker improvements (#2032) d2ca7e3 Minor update on cp.async code generation. (#1901) d36cf61 Test file cleanup (#2040) 0b8e83f Allow non-root trivial reductions (#2037) a2dfe40 Fix vectorize size calculation (#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (#2025) 197221b removing ci workflow (#2034) 40e2703 Reduction rand like patch (#2031) bc77266 Add utility for checking bank conflict of shared memory (#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds (#2006)" (#2024) bca20c1 Cleanup trivial reduction workarounds (#2006) e4b6585 Trivial forwarding (#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (#1991) a4effa6 Enable output allocation cache (#2010) 35440b7 Patching bn inference (#2016) 0f9f0b4 Add matmul benchmark (#2007) 45045cd Enable tests previously disabled due to an aliasing bug (#2005) 967aa77 Contiguous indexing for View operations (#1990) a43cb20 Make inlining even more modular (#2004) dc45835 Test util cleanup (#2003) 3ca21eb More strict validation (#2000) a7a7d57 Fix build problem (#1999) fc235b0 Just fixes comments (#1998) 482386c cleanup (#1997) 4cbe0db Improve divisible split detection (#1970) 42ccc52 Minor build fix. (#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (#1994) 1d9858c Minor cleanup (#1992) f262d9c Add support for uniform RNG (#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (#1987) 634820c Add support for some empty fusion (#1981) eabe8d8 Segment self mapping fusions (#1954) e96aacf Enable Transpose operation (#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (#1969) b1bd32c Minor fix (#1967) bd93578 Enable transpose scheduler (#1927) b7a206e Move scheduler vectorize utilities into their own file (#1959) d9420e4 View scheduling (#1928) c668e13 Upstream push ci fixes (#1965) c40202b Fix dump effective bandwidth (#1962) 93505bc WAR on index mapping when exact and permissive maps differ (#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (#1946) f7bc341 Remove unused variables (#1955) df3393a Some cleanup (#1957) 7d1d7c8 TVDomainGuard factory (#1953) 357ba22 Fill allocation with nan on tests (#1956) 8eafc54 Fix detection of unmappable root domains (#1952) 90a51f2 Some indexing cleanups, Add eye support (#1940) ddc01e4 Exclude unsupported data types (#1951) 992e17c test the groups the same order as they are merged (#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (#1941) ac4de38 Merge pull request #1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (#1942) 89330aa Tensor factories must set the output shape as its input (#1939) ``` RUN_TORCHBENCH: nvfuser Differential Revision: [D40869846](https://our.internmc.facebook.com/intern/diff/D40869846) [ghstack-poisoned]

Syncing nvfuser devel branch to upstream master. https://github.com/csarofeen/pytorch/ Codegen changes include: * codegen improvement: i. allow non-root trivial reductions, allow empty/no-op fusion ii. fixes vectorization checks and size calculation iii. bank conflict handle improvement iv. enables transpose scheduler * misc: i. CI tests failure fixes ii. cpp tests file clean up iii. trivial forwarding supports added in codegen runtime iv. added factory methods support in codegen Commits that's in this PR from the devel branch: ``` 7117a7e patching nvfuser conv cudnn test numerics mismatch (#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (#2023) 6ac74d1 Fix sync map (#2047) f5bca33 Bank conflict checker improvements (#2032) d2ca7e3 Minor update on cp.async code generation. (#1901) d36cf61 Test file cleanup (#2040) 0b8e83f Allow non-root trivial reductions (#2037) a2dfe40 Fix vectorize size calculation (#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (#2025) 197221b removing ci workflow (#2034) 40e2703 Reduction rand like patch (#2031) bc77266 Add utility for checking bank conflict of shared memory (#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds D856 (#2006)" (#2024) bca20c1 Cleanup trivial reduction workarounds (#2006) e4b6585 Trivial forwarding (#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (#1991) a4effa6 Enable output allocation cache (#2010) 35440b7 Patching bn inference (#2016) 0f9f0b4 Add matmul benchmark (#2007) 45045cd Enable tests previously disabled due to an aliasing bug (#2005) 967aa77 Contiguous indexing for View operations (#1990) a43cb20 Make inlining even more modular (#2004) dc45835 Test util cleanup (#2003) 3ca21eb More strict validation (#2000) a7a7d57 Fix build problem (#1999) fc235b0 Just fixes comments (#1998) 482386c cleanup (#1997) 4cbe0db Improve divisible split detection (#1970) 42ccc52 Minor build fix. (#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (#1994) 1d9858c Minor cleanup (#1992) f262d9c Add support for uniform RNG (#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (#1987) 634820c Add support for some empty fusion (#1981) eabe8d8 Segment self mapping fusions (#1954) e96aacf Enable Transpose operation (#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (#1969) b1bd32c Minor fix (#1967) bd93578 Enable transpose scheduler (#1927) b7a206e Move scheduler vectorize utilities into their own file (#1959) d9420e4 View scheduling (#1928) c668e13 Upstream push ci fixes (#1965) c40202b Fix dump effective bandwidth (#1962) 93505bc WAR on index mapping when exact and permissive maps differ (#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (#1946) f7bc341 Remove unused variables (#1955) df3393a Some cleanup (#1957) 7d1d7c8 TVDomainGuard factory (#1953) 357ba22 Fill allocation with nan on tests (#1956) 8eafc54 Fix detection of unmappable root domains (#1952) 90a51f2 Some indexing cleanups, Add eye support (#1940) ddc01e4 Exclude unsupported data types (#1951) 992e17c test the groups the same order as they are merged (#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (#1941) ac4de38 Merge pull request #1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (#1942) 89330aa Tensor factories must set the output shape as its input (#1939) ``` RUN_TORCHBENCH: nvfuser Differential Revision: [D40869846](https://our.internmc.facebook.com/intern/diff/D40869846) [ghstack-poisoned]

7117a7e patching nvfuser conv cudnn test numerics mismatch (#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (#2023) 6ac74d1 Fix sync map (#2047) f5bca33 Bank conflict checker improvements (#2032) d2ca7e3 Minor update on cp.async code generation. (#1901) d36cf61 Test file cleanup (#2040) 0b8e83f Allow non-root trivial reductions (#2037) a2dfe40 Fix vectorize size calculation (#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (#2025) 197221b removing ci workflow (#2034) 40e2703 Reduction rand like patch (#2031) bc77266 Add utility for checking bank conflict of shared memory (#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds (#2006)" (#2024) bca20c1 Cleanup trivial reduction workarounds (#2006) e4b6585 Trivial forwarding (#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (#1991) a4effa6 Enable output allocation cache (#2010) 35440b7 Patching bn inference (#2016) 0f9f0b4 Add matmul benchmark (#2007) 45045cd Enable tests previously disabled due to an aliasing bug (#2005) 967aa77 Contiguous indexing for View operations (#1990) a43cb20 Make inlining even more modular (#2004) dc45835 Test util cleanup (#2003) 3ca21eb More strict validation (#2000) a7a7d57 Fix build problem (#1999) fc235b0 Just fixes comments (#1998) 482386c cleanup (#1997) 4cbe0db Improve divisible split detection (#1970) 42ccc52 Minor build fix. (#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (#1994) 1d9858c Minor cleanup (#1992) f262d9c Add support for uniform RNG (#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (#1987) 634820c Add support for some empty fusion (#1981) eabe8d8 Segment self mapping fusions (#1954) e96aacf Enable Transpose operation (#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (#1969) b1bd32c Minor fix (#1967) bd93578 Enable transpose scheduler (#1927) b7a206e Move scheduler vectorize utilities into their own file (#1959) d9420e4 View scheduling (#1928) c668e13 Upstream push ci fixes (#1965) c40202b Fix dump effective bandwidth (#1962) 93505bc WAR on index mapping when exact and permissive maps differ (#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (#1946) f7bc341 Remove unused variables (#1955) df3393a Some cleanup (#1957) 7d1d7c8 TVDomainGuard factory (#1953) 357ba22 Fill allocation with nan on tests (#1956) 8eafc54 Fix detection of unmappable root domains (#1952) 90a51f2 Some indexing cleanups, Add eye support (#1940) ddc01e4 Exclude unsupported data types (#1951) 992e17c test the groups the same order as they are merged (#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (#1941) ac4de38 Merge pull request #1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (#1942) 89330aa Tensor factories must set the output shape as its input (#1939) RUN_TORCHBENCH: nvfuser ghstack-source-id: 424af8e Pull Request resolved: #87779

Syncing nvfuser devel branch to upstream master. https://github.com/csarofeen/pytorch/ Codegen changes include: * codegen improvement: i. allow non-root trivial reductions, allow empty/no-op fusion ii. fixes vectorization checks and size calculation iii. bank conflict handle improvement iv. enables transpose scheduler * misc: i. CI tests failure fixes ii. cpp tests file clean up iii. trivial forwarding supports added in codegen runtime iv. added factory methods support in codegen Commits that's in this PR from the devel branch: ``` 7117a7e patching nvfuser conv cudnn test numerics mismatch (#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (#2023) 6ac74d1 Fix sync map (#2047) f5bca33 Bank conflict checker improvements (#2032) d2ca7e3 Minor update on cp.async code generation. (#1901) d36cf61 Test file cleanup (#2040) 0b8e83f Allow non-root trivial reductions (#2037) a2dfe40 Fix vectorize size calculation (#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (#2025) 197221b removing ci workflow (#2034) 40e2703 Reduction rand like patch (#2031) bc77266 Add utility for checking bank conflict of shared memory (#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds (#2006)" (#2024) bca20c1 Cleanup trivial reduction workarounds (#2006) e4b6585 Trivial forwarding (#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (#1991) a4effa6 Enable output allocation cache (#2010) 35440b7 Patching bn inference (#2016) 0f9f0b4 Add matmul benchmark (#2007) 45045cd Enable tests previously disabled due to an aliasing bug (#2005) 967aa77 Contiguous indexing for View operations (#1990) a43cb20 Make inlining even more modular (#2004) dc45835 Test util cleanup (#2003) 3ca21eb More strict validation (#2000) a7a7d57 Fix build problem (#1999) fc235b0 Just fixes comments (#1998) 482386c cleanup (#1997) 4cbe0db Improve divisible split detection (#1970) 42ccc52 Minor build fix. (#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (#1994) 1d9858c Minor cleanup (#1992) f262d9c Add support for uniform RNG (#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (#1987) 634820c Add support for some empty fusion (#1981) eabe8d8 Segment self mapping fusions (#1954) e96aacf Enable Transpose operation (#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (#1969) b1bd32c Minor fix (#1967) bd93578 Enable transpose scheduler (#1927) b7a206e Move scheduler vectorize utilities into their own file (#1959) d9420e4 View scheduling (#1928) c668e13 Upstream push ci fixes (#1965) c40202b Fix dump effective bandwidth (#1962) 93505bc WAR on index mapping when exact and permissive maps differ (#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (#1946) f7bc341 Remove unused variables (#1955) df3393a Some cleanup (#1957) 7d1d7c8 TVDomainGuard factory (#1953) 357ba22 Fill allocation with nan on tests (#1956) 8eafc54 Fix detection of unmappable root domains (#1952) 90a51f2 Some indexing cleanups, Add eye support (#1940) ddc01e4 Exclude unsupported data types (#1951) 992e17c test the groups the same order as they are merged (#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (#1941) ac4de38 Merge pull request #1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (#1942) 89330aa Tensor factories must set the output shape as its input (#1939) ``` RUN_TORCHBENCH: nvfuser Differential Revision: [D40869846](https://our.internmc.facebook.com/intern/diff/D40869846) [ghstack-poisoned]

Syncing nvfuser devel branch to upstream master. https://github.com/csarofeen/pytorch/ Codegen changes include: * codegen improvement: i. allow non-root trivial reductions, allow empty/no-op fusion ii. fixes vectorization checks and size calculation iii. bank conflict handle improvement iv. enables transpose scheduler * misc: i. CI tests failure fixes ii. cpp tests file clean up iii. trivial forwarding supports added in codegen runtime iv. added factory methods support in codegen Commits that's in this PR from the devel branch: ``` 7117a7e patching nvfuser conv cudnn test numerics mismatch (#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (#2023) 6ac74d1 Fix sync map (#2047) f5bca33 Bank conflict checker improvements (#2032) d2ca7e3 Minor update on cp.async code generation. (#1901) d36cf61 Test file cleanup (#2040) 0b8e83f Allow non-root trivial reductions (#2037) a2dfe40 Fix vectorize size calculation (#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (#2025) 197221b removing ci workflow (#2034) 40e2703 Reduction rand like patch (#2031) bc77266 Add utility for checking bank conflict of shared memory (#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds (#2006)" (#2024) bca20c1 Cleanup trivial reduction workarounds (#2006) e4b6585 Trivial forwarding (#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (#1991) a4effa6 Enable output allocation cache (#2010) 35440b7 Patching bn inference (#2016) 0f9f0b4 Add matmul benchmark (#2007) 45045cd Enable tests previously disabled due to an aliasing bug (#2005) 967aa77 Contiguous indexing for View operations (#1990) a43cb20 Make inlining even more modular (#2004) dc45835 Test util cleanup (#2003) 3ca21eb More strict validation (#2000) a7a7d57 Fix build problem (#1999) fc235b0 Just fixes comments (#1998) 482386c cleanup (#1997) 4cbe0db Improve divisible split detection (#1970) 42ccc52 Minor build fix. (#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (#1994) 1d9858c Minor cleanup (#1992) f262d9c Add support for uniform RNG (#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (#1987) 634820c Add support for some empty fusion (#1981) eabe8d8 Segment self mapping fusions (#1954) e96aacf Enable Transpose operation (#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (#1969) b1bd32c Minor fix (#1967) bd93578 Enable transpose scheduler (#1927) b7a206e Move scheduler vectorize utilities into their own file (#1959) d9420e4 View scheduling (#1928) c668e13 Upstream push ci fixes (#1965) c40202b Fix dump effective bandwidth (#1962) 93505bc WAR on index mapping when exact and permissive maps differ (#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (#1946) f7bc341 Remove unused variables (#1955) df3393a Some cleanup (#1957) 7d1d7c8 TVDomainGuard factory (#1953) 357ba22 Fill allocation with nan on tests (#1956) 8eafc54 Fix detection of unmappable root domains (#1952) 90a51f2 Some indexing cleanups, Add eye support (#1940) ddc01e4 Exclude unsupported data types (#1951) 992e17c test the groups the same order as they are merged (#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (#1941) ac4de38 Merge pull request #1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (#1942) 89330aa Tensor factories must set the output shape as its input (#1939) ``` RUN_TORCHBENCH: nvfuser Differential Revision: [D40869846](https://our.internmc.facebook.com/intern/diff/D40869846) Pull Request resolved: #87779 Approved by: https://github.com/davidberard98

Syncing nvfuser devel branch to upstream master. https://github.com/csarofeen/pytorch/ Codegen changes include: * codegen improvement: i. allow non-root trivial reductions, allow empty/no-op fusion ii. fixes vectorization checks and size calculation iii. bank conflict handle improvement iv. enables transpose scheduler * misc: i. CI tests failure fixes ii. cpp tests file clean up iii. trivial forwarding supports added in codegen runtime iv. added factory methods support in codegen Commits that's in this PR from the devel branch: ``` 7117a7e patching nvfuser conv cudnn test numerics mismatch (pytorch#2048) 65af1a4 Inserting sync for redundant parallel types is already done at the (pytorch#2023) 6ac74d1 Fix sync map (pytorch#2047) f5bca33 Bank conflict checker improvements (pytorch#2032) d2ca7e3 Minor update on cp.async code generation. (pytorch#1901) d36cf61 Test file cleanup (pytorch#2040) 0b8e83f Allow non-root trivial reductions (pytorch#2037) a2dfe40 Fix vectorize size calculation (pytorch#2035) e040676 Use withPredicate to replace setPredicate to maintain Exprs immutable (pytorch#2025) 197221b removing ci workflow (pytorch#2034) 40e2703 Reduction rand like patch (pytorch#2031) bc77266 Add utility for checking bank conflict of shared memory (pytorch#2029) ddd1cf7 Add back FusionReductionWithTrivialReduction_CUDA (pytorch#2030) fbd97e5 Revert "Cleanup trivial reduction workarounds (pytorch#2006)" (pytorch#2024) bca20c1 Cleanup trivial reduction workarounds (pytorch#2006) e4b6585 Trivial forwarding (pytorch#1995) 1a0e355 Fix contiguity analysis of predicates to match updated contiguity. (pytorch#1991) a4effa6 Enable output allocation cache (pytorch#2010) 35440b7 Patching bn inference (pytorch#2016) 0f9f0b4 Add matmul benchmark (pytorch#2007) 45045cd Enable tests previously disabled due to an aliasing bug (pytorch#2005) 967aa77 Contiguous indexing for View operations (pytorch#1990) a43cb20 Make inlining even more modular (pytorch#2004) dc45835 Test util cleanup (pytorch#2003) 3ca21eb More strict validation (pytorch#2000) a7a7d57 Fix build problem (pytorch#1999) fc235b0 Just fixes comments (pytorch#1998) 482386c cleanup (pytorch#1997) 4cbe0db Improve divisible split detection (pytorch#1970) 42ccc52 Minor build fix. (pytorch#1996) fcf8c09 Cleanup of lower_utils.cpp: Isolate out GpuLower usage (pytorch#1989) 15f2f6d Move ConcretizedBroadcastDomains to shared_ptr in GpuLower. (pytorch#1988) 8f1c7f5 Minor cleanup lower_unroll.cpp (pytorch#1994) 1d9858c Minor cleanup (pytorch#1992) f262d9c Add support for uniform RNG (pytorch#1986) eb1dad1 Remove non-const functions, remove GpuLower instance on build, pass in ca_map. (pytorch#1987) 634820c Add support for some empty fusion (pytorch#1981) eabe8d8 Segment self mapping fusions (pytorch#1954) e96aacf Enable Transpose operation (pytorch#1882) 425dce2 Add a null scheduler that helps segmenting away no-op schedules (pytorch#1835) 306d4a6 Fix canScheduleCompileTime check of transpose scheduler (pytorch#1969) b1bd32c Minor fix (pytorch#1967) bd93578 Enable transpose scheduler (pytorch#1927) b7a206e Move scheduler vectorize utilities into their own file (pytorch#1959) d9420e4 View scheduling (pytorch#1928) c668e13 Upstream push ci fixes (pytorch#1965) c40202b Fix dump effective bandwidth (pytorch#1962) 93505bc WAR on index mapping when exact and permissive maps differ (pytorch#1960) 45e95fd Allow splitting inner-most ID to create virtual innermost ID in transpose scheduler (pytorch#1930) a3ecb33 Improve the comments at the beginning of index_compute.h (pytorch#1946) f7bc341 Remove unused variables (pytorch#1955) df3393a Some cleanup (pytorch#1957) 7d1d7c8 TVDomainGuard factory (pytorch#1953) 357ba22 Fill allocation with nan on tests (pytorch#1956) 8eafc54 Fix detection of unmappable root domains (pytorch#1952) 90a51f2 Some indexing cleanups, Add eye support (pytorch#1940) ddc01e4 Exclude unsupported data types (pytorch#1951) 992e17c test the groups the same order as they are merged (pytorch#1949) 208262b Move detection of self mapping IDs to IterDomainGraph from (pytorch#1941) ac4de38 Merge pull request pytorch#1945 from csarofeen/master_merge_0828 6310948 Add full, full_like, zeros, zeros_like, ones, ones_like (pytorch#1943) aab10bc Merge remote-tracking branch 'upstream/viable/strict' into HEAD 4c254c0 Fix arange when step is negative (pytorch#1942) 89330aa Tensor factories must set the output shape as its input (pytorch#1939) ``` RUN_TORCHBENCH: nvfuser Differential Revision: [D40869846](https://our.internmc.facebook.com/intern/diff/D40869846) Pull Request resolved: pytorch#87779 Approved by: https://github.com/davidberard98

soumith closed this as completed Jul 13, 2017

houseroad mentioned this issue May 10, 2020

Automatic update of fbcode/onnx to 20b3e10e6c3a9cdab90d2bb864d1c36d3e3651cd #38203

Closed

houseroad mentioned this issue May 27, 2020

Automatic update of fbcode/onnx to eae3eb8c61cf5ad27cc9a416dbdc5274982385a6 #39089

Closed

jjsjann123 pushed a commit to jjsjann123/pytorch that referenced this issue Oct 4, 2022

Remove unused variables (pytorch#1955)

f7bc341

jjsjann123 mentioned this issue Oct 26, 2022

[NVFuser] Upstream push 1026 #87779

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Cuda runtime error (8) : invalid device function at THCTensorMathPointwise.cu for latest build #1955

Cuda runtime error (8) : invalid device function at THCTensorMathPointwise.cu for latest build #1955

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Cuda runtime error (8) : invalid device function at THCTensorMathPointwise.cu for latest build #1955

Cuda runtime error (8) : invalid device function at THCTensorMathPointwise.cu for latest build #1955

Comments

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!