Commits

Branch selector

 master

User selector

frgossen

Datepicker

All time

Commit History

Commits on Jan 15, 2022

• [MLIR][MHLO] Add a pass to collapse reductions to 1D or 2D equivalents

  Show description for 3ad9798
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  3ad9798

  Copy full SHA for 3ad9798

Commits on Jan 13, 2022

• [MLIR][KernelGen] Remove trailing whitespace in *.mlir test

  Show description for 3688e29
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  3688e29

  Copy full SHA for 3688e29

Commits on Jan 12, 2022

• [MHLO] Disable pattern-based broadcast propagation in favor of the new pass

  Show description for 8686e3c
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  8686e3c

  Copy full SHA for 8686e3c

• [MHLO] Signal pass failure in SymbolicShapeOptimizationPass

  Show description for 7d5248a
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  7d5248a

  Copy full SHA for 7d5248a

Commits on Jan 11, 2022

• [MHLO] Move side effect-free ops into assuming regions over side effecting ops

  Show description for 4429446
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  4429446

  Copy full SHA for 4429446

• [MLIR][MHLO] Move elementwise ops into assuming regions to unblock fusion

  Show description for 6ab370d
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  6ab370d

  Copy full SHA for 6ab370d

• Propagate broadcasts in the presence of late output dimensions

  Show description for 5d6bc7e
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  5d6bc7e

  Copy full SHA for 5d6bc7e

• [MLIR][MHLO] Make broadcast propagation optional in `merge-assuming-ops` pass

  Show description for 5cb1975
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  5cb1975

  Copy full SHA for 5cb1975

Commits on Dec 14, 2021

• Add block-global broadcast propagation pass

  Show description for afee3b5
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  afee3b5

  Copy full SHA for afee3b5

• Move merge assuming ops test (old broadcast propagation pass) into the mhlo dir

  Show description for 7ba5fbc
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  7ba5fbc

  Copy full SHA for 7ba5fbc

• Rename broadcast propagation pass to MergeAssumingOps

  Show description for ad2966d
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  ad2966d

  Copy full SHA for ad2966d

Commits on Dec 9, 2021

• [CHLO] Avoid redundant casts in CHLO legalization

  Show description for e751f5e
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  e751f5e

  Copy full SHA for e751f5e

Commits on Dec 6, 2021

• [ShapeAnalysis] Make use of static type information for `dynamic_reshape`

  Show description for bfd1597
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  bfd1597

  Copy full SHA for bfd1597

Commits on Dec 2, 2021

• [ShapeAnalysis] Extend shape analysis test pass

  Show description for 2636e03
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  2636e03

  Copy full SHA for 2636e03

• [ReshapeSimplifier] Replace dynamic_reshape with tensor_collapse_shape

  Show description for 75910e2
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  75910e2

  Copy full SHA for 75910e2

Commits on Dec 1, 2021

• [KernelGen][JIT] Require compilation with ptxas for internal builds

  Show description for a294cf3
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  a294cf3

  Copy full SHA for a294cf3

Commits on Nov 26, 2021

• [ShapeAnalysis] Eliminate more redundant shape constraints in real-world cases

  Show description for ed84f76
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  ed84f76

  Copy full SHA for ed84f76

Commits on Nov 25, 2021

• [ShapeAnalysis] Rename functions and rephrase comments in the presence of scalar analysis

  Show description for 614b15a
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  614b15a

  Copy full SHA for 614b15a

• [KernelGen][JIT] Eliminate dependency on fatbin and ptxas for JIT-compiled kernels

  Show description for f37b7b1
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  f37b7b1

  Copy full SHA for f37b7b1

Commits on Nov 24, 2021

• [KernelGen][JIT] Enable JIT-compiler OnesLike kernels

  Show description for adfda01
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  adfda01

  Copy full SHA for adfda01

• [MLIR][JIT] Add the missing JIT-compiled kernels for TruncatedDiv

  Show description for bcb7248
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  bcb7248

  Copy full SHA for bcb7248

• [KernelGen][JIT] Return a dummy memref in case JIT compilation fails

  Show description for d6c6444
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  d6c6444

  Copy full SHA for d6c6444

Commits on Nov 23, 2021

• [KernelGen] Remove outdated warning

  Show description for 17ffa92
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  17ffa92

  Copy full SHA for 17ffa92

• [KernelGen][JIT] Remove GPU architecture propagation

  Show description for 584da35
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  584da35

  Copy full SHA for 584da35

• [KernelGen][JIT] Determine the GPU architecture at runtime

  Show description for 93aa174
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  93aa174

  Copy full SHA for 93aa174

Commits on Nov 22, 2021

• [MLIR][JIT] Register tf.math.square for unsigned integer types

  Show description for b4d1680
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  b4d1680

  Copy full SHA for b4d1680

Commits on Nov 19, 2021

• Integrate LLVM at llvm/llvm-project@d61840c168a3

  Show description for 31676d1
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  31676d1

  Copy full SHA for 31676d1

• Integrate LLVM at llvm/llvm-project@913d78c40c37

  Show description for 2e5d3f3
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  2e5d3f3

  Copy full SHA for 2e5d3f3

Commits on Nov 18, 2021

• Integrate LLVM at llvm/llvm-project@274f12a44c60

  Show description for 4658796
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  4658796

  Copy full SHA for 4658796

Commits on Nov 17, 2021

• Eliminate redundant reshape constraints in the presence of a -1 dimension

  Show description for 2b65e0f
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  2b65e0f

  Copy full SHA for 2b65e0f

Commits on Nov 15, 2021

• Remove redundant shape constraints

  Show description for 8b7add3
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  8b7add3

  Copy full SHA for 8b7add3

• [MLIR][KernelGen] Add experimental JIT-compiled GPU kernels for tf.OnesLike

  Show description for 5100e35
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  5100e35

  Copy full SHA for 5100e35

Commits on Nov 11, 2021

• Make shape analysis aware of `shape.num_elements` operation

  Show description for 3026b6a
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  3026b6a

  Copy full SHA for 3026b6a

• [KernelGen][JIT] Enable remaining JIT-compiled GPU kernels

  Show description for d396d2a
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  d396d2a

  Copy full SHA for d396d2a

Commits on Nov 8, 2021

• [MLIR][KernelGen] Add JIT-compiled GPU kernels for the missing select_v2 dtypes

  Show description for 3ebb0e1
  

  frgossen

  authored and

  tensorflower-gardener

  committed
  3ebb0e1

  Copy full SHA for 3ebb0e1

Pagination

Previous

Next