[DTensor] implement histc #158298
[DTensor] implement histc #158298
Conversation
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/158298
Note: Links to docs will display an error until the docs builds have been completed. ❌ 2 Cancelled Jobs, 1 Unrelated FailureAs of commit e98bc45 with merge base 194539e ( CANCELLED JOBS - The following jobs were cancelled. Please retry:
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
ciflow/inductor
oncall: distributed
cc H-Huang awgu wanchaol fegin fduwjj wz337 d4l3k [ghstack-poisoned]
| out_full = out_dt.full_tensor() | ||
| self.assertEqual(global_bins, out_full) | ||
|
|
||
| # TODO: support backward |
There was a problem hiding this comment.
I think histc might not have a backward
There was a problem hiding this comment.
Well that would be convenient 😅
|
I did a test to confirm. If I call histc(tensor), then len(args_schema) was 1. e.g default values do not show up. If I called histc(tensor, min=1,max=2), len(args_schema) was 4. E.g even though 'bins' arg was not specified by the user, its default value got 'promoted' to look like it was passed by the user. So I think ==4 is correct, not too strict. I'm not an expert on how the argument passing works though, so let me know if I missed something. |
| # TODO what is this for? | ||
| # schema_info=RuntimeSchemaInfo(1), |
There was a problem hiding this comment.
bins, min, and max does not affect the strategy. So we can leave RuntimeSchemaInfo unspecified.
There was a problem hiding this comment.
actually, min, max do affect the strategy. bins does not affect the strategy. So i think i need to set it to 4?
i am glad you raised this comment becuase I forgot about my TODO.
There was a problem hiding this comment.
following up, i need to specify '2' to indicate the first non-tensor arg that needs to be hashed, not 4 to indicate the last.
cc H-Huang awgu wanchaol fegin fduwjj wz337 d4l3k [ghstack-poisoned]
cc H-Huang awgu wanchaol fegin fduwjj wz337 d4l3k [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 |
| self.assertEqual(comm_mode.get_total_counts(), 0) | ||
|
|
||
| out_full = out_dt.full_tensor() | ||
| self.assertEqual(global_bins, out_full) |
There was a problem hiding this comment.
can also try using the util self. _test_op_on_dtensor added in #158112
but this would change the comm_mode numbers because it adds an extra full_tensor().
Merge failedReason: 1 mandatory check(s) failed. The first few are: Dig deeper by viewing the failures on hud |
|
@pytorchbot merge -i |
Merge startedYour change will be merged while ignoring the following 2 checks: pull / cuda12.8-py3.10-gcc9-sm75 / test (pr_time_benchmarks, 1, 1, linux.g4dn.metal.nvidia.gpu, unstable), trunk / linux-jammy-rocm-py3.10 / build Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Merge failedReason: 1 jobs have failed, first few of them are: trunk / linux-jammy-rocm-py3.10 / test (default, 1, 2, linux.rocm.gpu.2) Details for Dev Infra teamRaised by workflow job |
|
@pytorchbot merge -f |
|
❌ 🤖 pytorchbot command failed: Try |
|
@pytorchbot merge -f"broken infra for rocm?" |
Merge startedYour change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Please use Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
5 participants
Stack from ghstack (oldest at bottom):
cc @H-Huang @awgu @wanchaol @fegin @fduwjj @wz337 @d4l3k