spcl/open-earth-compiler

Issues

• Nsight-compute fail when profile the kernel performance

  #50 opened 4 months ago by weilinquan
  1

• Compilation fails when CUDA is not in the usual path

  #49 opened 6 months ago by cbouilla
  1

• status of project

  #48 opened 2 years ago by rohany
  3

• stencil-shape-inference is not working with tridiagonal example

  #47 opened 2 years ago by meshtag
  3

• problem about compiling fastwaves.mlir

  #46 opened 3 years ago by Xinyu302
  2

• [mlir]the ptx compilation flow

  #45 opened 3 years ago by CONGCONGLEEE
  2

• problem about laplace example in readme

  #44 opened 3 years ago by lethean1
  2

• cmake error: too many arguments to function createGpuToLLVMConversionPass

  #43 opened 3 years ago by CONGCONGLEEE
  2

• String-based OpBuilder removed

  #42 opened 4 years ago by lvaz
  4

• if/else optimization

  #37 opened 4 years ago by gysit
  0

• explicit intermediate storage to prevent fusion of long stencil chains

  #36 opened 4 years ago by gysit
  1

• add stencil access that support dynamic offsets given some constant bound

  #38 opened 4 years ago by gysit
  1

• Interleaved unrolling

  #41 opened 4 years ago by muellch
  0

• subdomains

  #34 opened 4 years ago by gysit
  1

• add the possiblity to load and store from and to the same field

  #40 opened 4 years ago by gysit
  1

• Change stencil.assert to stencil.cast with a return value

  #33 opened 4 years ago by gysit
  0

• operation to access a lower dimensional slice

  #39 opened 4 years ago by gysit
  0

• Multi-dimensional Unrolling

  #12 opened 5 years ago by muellch
  1

• register scheduling optimization

  #8 opened 5 years ago by gysit
  1

• Instruction count difference compared to reference

  #25 opened 4 years ago by muellch
  1

• Multi-threading bug in compiler

  #22 opened 5 years ago by muellch
  3

• Vector loads for stencils

  #14 opened 5 years ago by muellch
  0

• Refactoring ideas for the stencil dialect

  #17 opened 4 years ago by gysit
  2

• Write outputs with different range

  #16 opened 4 years ago by gysit
  0

• Loop.if vs cmpf/select during canonicalization

  #18 opened 4 years ago by muellch
  1

• LLVM CSE vs (PTX/ROCM) CSE

  #23 opened 4 years ago by muellch
  0

• Performance comparison to dawn

  #24 opened 4 years ago by muellch
  0

• Support any unroll factor

  #30 opened 4 years ago by muellch
  0

• tridiagonal solvers

  #35 opened 4 years ago by gysit
  0

• implement additional stencils and analyze the performance of our compiler

  #6 opened 4 years ago by gysit
  0

• CMake needs to include CUDA directories

  #32 opened 4 years ago by Mogball
  5

• Temporary + Output Fields are Shifted

  #19 opened 5 years ago by gysit
  0

• update the argument list format of the stencil dialect to match the standards of mlir

  #7 opened 5 years ago by gysit
  0

• Add continuous integration

  #28 opened 5 years ago by jmgorius
  2

• test different cu jit compiler flags

  #9 opened 5 years ago by gysit
  1

• lower dimensional arrays

  #3 opened 5 years ago by gysit
  0

• Inlining when stencil.applies use the same inputs

  #13 opened 5 years ago by muellch
  0

• Test if 'if-else' statements are lowered correctly

  #11 opened 5 years ago by muellch
  0

• Abstracted Stencil DSL

  #10 opened 5 years ago by muellch
  0

• lower to parallel loop instead to affine loops

  #2 opened 5 years ago by gysit
  0

• loop unrolling

  #4 opened 5 years ago by gysit
  1

• build system issues

  #1 opened 5 years ago by gysit
  0

• support bare ptr calling convention in the GPU backend

  #5 opened 5 years ago by gysit
  1