Add basic torch.hash_tensor op #154149
Closed
Add basic torch.hash_tensor op #154149
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
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/154149
Note: Links to docs will display an error until the docs builds have been completed. ⏳ 1 Pending, 1 Unrelated FailureAs of commit 5c8132c with merge base f168cf4 ( UNSTABLE - The following job is marked as unstable, possibly due to flakiness on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
pytorch-bot
bot
added
the
module: cpu
mikaylagawarecki
added
module: python frontend
Attention! native_functions.yaml was changedIf you are adding a new function or defaulted argument to native_functions.yaml, you cannot use it from pre-existing Python frontend code until our FC window passes (two weeks). Split your PR into two PRs, one which adds the new C++ functionality, and one that makes use of it from Python, and land them two weeks apart. See https://github.com/pytorch/pytorch/wiki/PyTorch's-Python-Frontend-Backward-and-Forward-Compatibility-Policy#forwards-compatibility-fc for more info. Caused by: |
cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 jerryzh168 albanD [ghstack-poisoned]
pytorch-bot
bot
added
ciflow/inductor
module: cpu
cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 jerryzh168 albanD voznesenskym penguinwu EikanWang Guobing-Chen zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov [ghstack-poisoned]
[ghstack-poisoned]
[ghstack-poisoned]
[ghstack-poisoned]
[ghstack-poisoned]
[ghstack-poisoned]
[ghstack-poisoned]
mikaylagawarecki
added
the
release notes: python_frontend
Added `torch.hash_tensor` reduction function with a `mode` argument that defaults to multiply/shift via range then reduction with xor. The tensor is always viewed as int64 (applying padding as necessary) before the reduction so the result is always int64. I chose int64 rather than uint64 because e.g. a * x + b is not implemented for uint64 in torch on cuda. [ghstack-poisoned]
Added `torch.hash_tensor` reduction function with a `mode` argument that defaults to multiply/shift via range then reduction with xor. The tensor is always viewed as int64 (applying padding as necessary) before the reduction so the result is always int64. I chose int64 rather than uint64 because e.g. a * x + b is not implemented for uint64 in torch on cuda. [ghstack-poisoned]
Added `torch.hash_tensor` reduction function with a `mode` argument that defaults to multiply/shift via range then reduction with xor. The tensor is always viewed as int64 (applying padding as necessary) before the reduction so the result is always int64. I chose int64 rather than uint64 because e.g. a * x + b is not implemented for uint64 in torch on cuda. [ghstack-poisoned]
Added `torch.hash_tensor` reduction function with a `mode` argument that defaults to multiply/shift via range then reduction with xor. The tensor is always viewed as int64 (applying padding as necessary) before the reduction so the result is always int64. I chose int64 rather than uint64 because e.g. a * x + b is not implemented for uint64 in torch on cuda. [ghstack-poisoned]
Added `torch.hash_tensor` reduction function with a `mode` argument that defaults to reduction with xor. - The hash is always uint64. - Integers will be casted to uint64 before performing the xor_sum reduction - Floats will be upcasted to double and then bitcasted to uint64 before performing the xor_sum reduction [ghstack-poisoned]
Added `torch.hash_tensor` reduction function with a `mode` argument that defaults to reduction with xor. - The hash is always uint64. - Integers will be casted to uint64 before performing the xor_sum reduction - Floats will be upcasted to double and then bitcasted to uint64 before performing the xor_sum reduction [ghstack-poisoned]
albanD
reviewed
Jul 22, 2025
Comment on lines
+184
to
+185
| // return a double, otherwise uint64_t will be cast to double | ||
| // when accumulating and the result will be wrong |
There was a problem hiding this comment.
Let's say we have inputs
a: 3.14159, a_bits: 4614256650576692846
b: 1.61803, b_bits: 4609965778477721196
a ^ b: 9219082337818812418
If we return the result as int64_t, the next time it is used as an input, it will be cast to 9219082337818812418.0, which becomes 4890905006165143848 in bits, which would mess up the xor reduction here
Added `torch.hash_tensor` reduction function with a `mode` argument that defaults to reduction with xor. - The hash is always uint64. - Integers will be casted to uint64 before performing the xor_sum reduction - Floats will be upcasted to double and then bitcasted to uint64 before performing the xor_sum reduction [ghstack-poisoned]
Added `torch.hash_tensor` reduction function with a `mode` argument that defaults to reduction with xor. - The hash is always uint64. - Integers will be casted to uint64 before performing the xor_sum reduction - Floats will be upcasted to double and then bitcasted to uint64 before performing the xor_sum reduction [ghstack-poisoned]
Added `torch.hash_tensor` reduction function with a `mode` argument that defaults to reduction with xor. - The hash is always uint64. - Integers will be casted to uint64 before performing the xor_sum reduction - Floats will be upcasted to double and then bitcasted to uint64 before performing the xor_sum reduction [ghstack-poisoned]
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
yangw-dev
pushed a commit
that referenced
this pull request
Aug 1, 2025
Added `torch.hash_tensor` reduction function with a `mode` argument that defaults to reduction with xor. - The hash is always uint64. - Integers will be casted to uint64 before performing the xor_sum reduction - Floats will be upcasted to double and then bitcasted to uint64 before performing the xor_sum reduction Pull Request resolved: #154149 Approved by: https://github.com/albanD
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Added
torch.hash_tensorreduction function with amodeargument that defaults to reduction with xor.We don't provide an ordering aware hash function yet because we don't have an easy way to provide a fast cuda kernel for it
Stack from ghstack (oldest at bottom):
cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10 @jerryzh168 @voznesenskym @penguinwu @EikanWang @Guobing-Chen @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @chenyang78 @kadeng @muchulee8 @amjames @chauhang @aakhundov @coconutruben @Lucaskabela