Skip to content

[Operator][Backend] Add nvcc flags for faster math and update Attention schedule#221

Merged
hjjq merged 20 commits intohidet-org:mainfrom
hjjq:attn
May 11, 2023
Merged

[Operator][Backend] Add nvcc flags for faster math and update Attention schedule#221
hjjq merged 20 commits intohidet-org:mainfrom
hjjq:attn

Conversation

@hjjq
Copy link
Copy Markdown
Collaborator

@hjjq hjjq commented May 9, 2023 •
edited
Loading

  • Make -ftz=true and prec-div=false default for all nvcc compiled kernels
  • Update Attention schedule template
  • Make repeat mapping explicit unroll by default when extent < 4
  • Fix erf test. Increase tolerance for pool test

@hjjq hjjq changed the title [Operator] Add fast_math and update Attention schedule [Operator] Add nvcc flags for faster math and update Attention schedule May 11, 2023
@hjjq hjjq changed the title [Operator] Add nvcc flags for faster math and update Attention schedule [Operator][Backend] Add nvcc flags for faster math and update Attention schedule May 11, 2023
@hjjq hjjq merged commit 971bd01 into hidet-org:main May 11, 2023
vadiklyutiy pushed a commit that referenced this pull request Jul 22, 2024
@BolinSNLHM @vadiklyutiy
[Operators] Registering torch.as_tensor (#235)
4add4b9 
Adding support for the operator `torch.as_tensor`, which was encountered
in #221

Also added more tests for `torch.argmax, torch.argmin` as discussed in
#234
vadiklyutiy pushed a commit that referenced this pull request Jul 22, 2024
@BolinSNLHM @vadiklyutiy
[Operator] Adding support to operators torch.Tensor.max and `torch.…
99b12b3 
…Tensor.new_full` (#238)

Adding two operators `torch.Tensor.max` and `torch.Tensor.new_full`
while attempting to compile models from TorchBench(mentioned in the
comments under #221).
vadiklyutiy pushed a commit that referenced this pull request Jul 23, 2024
@BolinSNLHM @vadiklyutiy
[Operators] Registering torch.as_tensor (#235)
540367b 
Adding support for the operator `torch.as_tensor`, which was encountered
in #221

Also added more tests for `torch.argmax, torch.argmin` as discussed in
#234
vadiklyutiy pushed a commit that referenced this pull request Jul 23, 2024
@BolinSNLHM @vadiklyutiy
[Operator] Adding support to operators torch.Tensor.max and `torch.…
c5912a4 
…Tensor.new_full` (#238)

Adding two operators `torch.Tensor.max` and `torch.Tensor.new_full`
while attempting to compile models from TorchBench(mentioned in the
comments under #221).
vadiklyutiy pushed a commit that referenced this pull request Dec 26, 2024
@BolinSNLHM
[Operators] Registering torch.as_tensor (#235)
ac87ccb 
Adding support for the operator `torch.as_tensor`, which was encountered
in #221

Also added more tests for `torch.argmax, torch.argmin` as discussed in
#234
vadiklyutiy pushed a commit that referenced this pull request Dec 26, 2024
@BolinSNLHM
[Operator] Adding support to operators torch.Tensor.max and `torch.…
d0aabaa 
…Tensor.new_full` (#238)

Adding two operators `torch.Tensor.max` and `torch.Tensor.new_full`
while attempting to compile models from TorchBench(mentioned in the
comments under #221).
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@hjjq