[reland][quant] QuantizedCUDA implementation #36936
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Summary: Closes #30813 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. [ghstack-poisoned]
jerryzh168
changed the title
Summary: Closes #30813 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. [ghstack-poisoned]
jerryzh168
added a commit
that referenced
this pull request
Apr 20, 2020
Summary: Closes #30813 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. ghstack-source-id: 7b5f047 Pull Request resolved: #36936
jerryzh168
changed the title
|
@jerryzh168 Sorry, haven't visited GitHub for some time. Unfortunately, I don't have Windows machines at hand. I can set up everything from scratch and investigate the problem, but that would take a day or so. Is it worth it? |
I think I have fixed it, I'll land again |
|
@jerryzh168 merged this pull request in 97d3a84. |
|
Unlanding. This appears to have broken CUDA 10 and in some cases CUDA 10.1 builds of PyTorch. While breaking CUDA 10 may be acceptable, how that's done should likely go through review and notice provided to engineers still using CUDA 10 that they'll need to upgrade. The CUDA 10.1 issue is reported as blocking development by some engineers using devfairs. The cause/scope of this build issue is unclear at this time. |
jerryzh168
added a commit
that referenced
this pull request
Apr 22, 2020
Summary: Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. [ghstack-poisoned]
jerryzh168
added a commit
that referenced
this pull request
Apr 22, 2020
Summary: Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. ghstack-source-id: 845ce0b Pull Request resolved: #37081
jerryzh168
added a commit
that referenced
this pull request
Apr 22, 2020
Summary: Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. [ghstack-poisoned]
jerryzh168
added a commit
that referenced
this pull request
Apr 22, 2020
Summary: Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. [ghstack-poisoned]
jerryzh168
added a commit
that referenced
this pull request
Apr 22, 2020
Summary: Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. ghstack-source-id: 447571e Pull Request resolved: #37081
jerryzh168
added a commit
that referenced
this pull request
Apr 22, 2020
Summary: Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. [ghstack-poisoned]
jerryzh168
added a commit
that referenced
this pull request
Apr 23, 2020
Summary: Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. [ghstack-poisoned]
jerryzh168
added a commit
that referenced
this pull request
Apr 23, 2020
Summary: Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. ghstack-source-id: c307ca5 Pull Request resolved: #37081
jerryzh168
added a commit
that referenced
this pull request
Apr 23, 2020
Summary: Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. Differential Revision: [D21206694](https://our.internmc.facebook.com/intern/diff/D21206694) [ghstack-poisoned]
jerryzh168
added a commit
that referenced
this pull request
Apr 23, 2020
Summary: Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. Differential Revision: [D21206694](https://our.internmc.facebook.com/intern/diff/D21206694) [ghstack-poisoned]
jerryzh168
added a commit
that referenced
this pull request
Apr 24, 2020
Summary: Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. Differential Revision: [D21206694](https://our.internmc.facebook.com/intern/diff/D21206694) [ghstack-poisoned]
jerryzh168
added a commit
that referenced
this pull request
Apr 24, 2020
Summary: Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. ghstack-source-id: 4e511be Pull Request resolved: #37081
facebook-github-bot
pushed a commit
that referenced
this pull request
Apr 24, 2020
Summary: Pull Request resolved: #37081 Closes #30813 Relanding of #35463 1. Tensor quantization logic(quantize_*) is moved to the aten/native/quantized. Previously all logic for tensor quantization lived in the aten/quantized/Quantizer.cpp file, and started to become complicated and hard to read. This problem should be addressed in refactoring PR. Still, I reworked this partially because I had to add tensor quantization logic for CUDA, and it was native to move everything to the aten/native/quantized. 2. Requirements to run CUDA_tensor_apply* was eased to process any tenser that lives on the CUDA device(QuantizedCUDA included). 3. All quantized data types now have a default constructor. NVCC refuses to compile any gpu_kernel or CUDA_tensor_apply* without them. 4. Minor changes in many files to register QuantizedCUDA backend. 5. test_quantized_tensor is extended to process QuantizedCUDA backend where possible. Test Plan: Imported from OSS Differential Revision: D21206694 Pulled By: jerryzh168 fbshipit-source-id: c7433aad9c095a34c57e6dddd128b5c5d9292373
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Stack from ghstack:
Summary:
Closes #30813
Relanding of #35463
Differential Revision: D21143025