Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

KRNL Lowering for CPUs with Optimizations #192

Open
8 tasks
AlexandreEichenberger opened this issue Jun 26, 2020 · 3 comments
Open
8 tasks

KRNL Lowering for CPUs with Optimizations #192

AlexandreEichenberger opened this issue Jun 26, 2020 · 3 comments
Assignees
Labels
KRNL IR and lowering Support for lowering of KRNL IR to lower MLIR dialects.

Comments

@AlexandreEichenberger
Copy link
Collaborator

Optimizations for GEMM and CONV

GEMM:

  • Tiling
  • SIMD code gen
  • Memory Layouts
  • Good instructions mix for Z

CONV:

  • Tiling
  • SIMD code gen
  • Memory Layouts
  • Good instructions mix for Z
@AlexandreEichenberger AlexandreEichenberger self-assigned this Jun 26, 2020
@AlexandreEichenberger AlexandreEichenberger added the KRNL IR and lowering Support for lowering of KRNL IR to lower MLIR dialects. label Jun 26, 2020
@doru1004
Copy link
Collaborator

doru1004 commented Nov 6, 2020

@AlexandreEichenberger any updates on this?

@AlexandreEichenberger
Copy link
Collaborator Author

@doru1004 Next once we have all the dynamic shapes...

@Mookel
Copy link

Mookel commented Jun 10, 2021

Hi, all, I am just wondering is there any progress or update on this issue? 😀

cjvolzka added a commit to cjvolzka/onnx-mlir that referenced this issue Nov 15, 2023
* detect LayerNorm in presence of reciprocal and div of 1 (onnx#2609)

Signed-off-by: Alexandre Eichenberger <[email protected]>

* [NNPA] Use F16 as element type for zTensor (onnx#2611)

* Use f16 as element type for zTensor

Signed-off-by: Tung D. Le <[email protected]>

---------

Signed-off-by: Tung D. Le <[email protected]>

* Layernorm: convert instance norm and group norm to layer norm. (onnx#2595)

Signed-off-by: Alexandre Eichenberger <[email protected]>
Co-authored-by: Tung D. Le <[email protected]>

* Parse and set --mcpu in onnx-mlir-opt command (onnx#2614)

Signed-off-by: Tung D. Le <[email protected]>

* Import dim_param for model inputs and outputs (onnx#2616)

* Import dim_param for model inputs and outputs
* use argument attributes

Signed-off-by: Tung D. Le <[email protected]>

---------

Signed-off-by: Tung D. Le <[email protected]>
Co-authored-by: Alexandre Eichenberger <[email protected]>

* [DialectBuilder] add builder funcrions for ONNXSumOp and ONNXConvOp (onnx#2572)

The DialectBuilder class seems to be missing the function create the
ONNXSumOp and ONNXConOp nodes and check their shape.  This patch adds
the necessary functions.

Signed-off-by: Ashay Rane <[email protected]>
Signed-off-by: Alexandre Eichenberger <[email protected]>
Co-authored-by: Alexandre Eichenberger <[email protected]>

* [StableHLO] Lowers PadOp (constant mode) & GatherElements Op to StableHLO (onnx#2602)

* [Stablehlo] Pad constant mode & GatherElements to Stablehlo

Signed-off-by: chongsong.chen <[email protected]>
Signed-off-by: Yan Xu <[email protected]>
Co-authored-by: chongsong.chen <[email protected]>
Co-authored-by: Alexandre Eichenberger <[email protected]>

* [build] Add cmake option to enable/disable Java components build (onnx#2613)

* Add ONNX_MLIR_ENABLE_JAVA cmake option (default TRUE)

Signed-off-by: Boyana Norris <[email protected]>
Co-authored-by: Alexandre Eichenberger <[email protected]>

Co-authored-by: Alexandre Eichenberger <[email protected]>
Co-authored-by: Tung D. Le <[email protected]>
Co-authored-by: Ashay Rane <[email protected]>
Co-authored-by: Yan Xu <[email protected]>
Co-authored-by: chongsong.chen <[email protected]>
Co-authored-by: Boyana Norris <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
KRNL IR and lowering Support for lowering of KRNL IR to lower MLIR dialects.
Projects
None yet
Development

No branches or pull requests

3 participants