Code

Linear Algebra API & Parser#364

Code

Open

mattlkf wants to merge 70 commits intotensor-compiler:mastertensor-compiler/taco:masterfrom mattlkf:mastermattlkf/taco:masterCopy head branch name to clipboard

+2,983-18

Commits

Commits on Oct 30, 2020

• Add in linalg parser that still uses IndexExpr notation

  weiya711

  committed
  83ad3c0

  Copy full SHA for 83ad3c0

Commits on Nov 3, 2020

• Add in linalg_notation files for future use

  weiya711

  committed
  4630aed

  Copy full SHA for 4630aed

• some debugging couts

  mattlkf

  committed
  8811eda

  Copy full SHA for 8811eda

• Merge branch 'master' into me-playing-with-her-parser

  mattlkf

  committed
  0d7a174

  Copy full SHA for 0d7a174

• pair programming - LinalgBase, Matrix, Vector dummies

  mattlkf

  committed
  561aa51

  Copy full SHA for 561aa51

Commits on Nov 5, 2020

• Add in shapes and format for matrix and vector API and also assignment stmts

  weiya711

  committed
  00ed53d

  Copy full SHA for 00ed53d

Commits on Nov 6, 2020

• Add in code to start rewriting from linalg notation to index notation

  weiya711

  committed
  0a5a547

  Copy full SHA for 0a5a547

Commits on Nov 7, 2020

• squash init order warning, add dummy at method for Matrix, use size_t dims

  mattlkf

  committed
  d3a6d83

  Copy full SHA for d3a6d83

Commits on Nov 8, 2020

• init-ed a tensorbase inside the linalgbase

  mattlkf

  committed
  da5dbc2

  Copy full SHA for da5dbc2

Commits on Nov 9, 2020

• poking around at the Tensor API

  mattlkf

  committed
  ef59d4b

  Copy full SHA for ef59d4b

• resolve merge

  mattlkf

  committed
  20f1450

  Copy full SHA for 20f1450

• declarations for passing in a TensorBase ptr to the LinalgExpr

  mattlkf

  committed
  e039952

  Copy full SHA for e039952

Commits on Nov 10, 2020

• hopeful signs - getting tensorBase

  mattlkf

  committed
  3e1ed50

  Copy full SHA for 3e1ed50

• can reference the TensorBase inside the rewrite

  mattlkf

  committed
  d5336fc

  Copy full SHA for d5336fc

• experiment with doing the assignment after the rewrite of the RHS

  mattlkf

  committed
  8d512fe

  Copy full SHA for 8d512fe

• set idxcount(0) and print tensorBase in <<

  mattlkf

  committed
  4576e3c

  Copy full SHA for 4576e3c

Commits on Nov 11, 2020

• Add in transpose() and elemMul() for linalg expr and also add more nodes in rewrite

  weiya711

  committed
  703e62e

  Copy full SHA for 703e62e

• Add in rewrite for LinalgLiteralNodes

  weiya711

  committed
  86e2da1

  Copy full SHA for 86e2da1

• Add in matrix multiplication and shape checks and vectors and scalars

  weiya711

  committed
  9ebc5ba

  Copy full SHA for 9ebc5ba

Commits on Nov 12, 2020

• demonstrate that insertion works, compute on actual data works too

  mattlkf

  committed
  8ce89c6

  Copy full SHA for 8ce89c6

• rename my simplest test to tensorbase

  mattlkf

  committed
  b98e3f5

  Copy full SHA for b98e3f5

• Add in transpose for vectors

  weiya711

  committed
  6fd2d0c

  Copy full SHA for 6fd2d0c

• resolve merge

  mattlkf

  committed
  fb63f98

  Copy full SHA for fb63f98

• resolve merge

  mattlkf

  committed
  8b539a1

  Copy full SHA for 8b539a1

• Add in implicit rewrite after assignment

  weiya711

  committed
  8cb0c14

  Copy full SHA for 8cb0c14

• Add in fixes after TensorBase merge and partial linalg_parser that uses LinalgExpr and LinalgBase

  weiya711

  committed
  1a8d7f0

  Copy full SHA for 1a8d7f0

• remove the more egregious spewing of couts

  mattlkf

  committed
  0e06201

  Copy full SHA for 0e06201

• add data to testcases involving vectors + add a write method for vector

  mattlkf

  committed
  d6f25c1

  Copy full SHA for d6f25c1

Commits on Nov 13, 2020

• use ASSERT_EQ in some tests

  mattlkf

  committed
  98aedd5

  Copy full SHA for 98aedd5

Commits on Nov 16, 2020

• Add in changes to fix linalg_parser

  weiya711

  committed
  c7b8c4a

  Copy full SHA for c7b8c4a

• Add in scalar support for linalg_parser

  weiya711

  committed
  8bbb892

  Copy full SHA for 8bbb892

Commits on Nov 19, 2020

• WIP blocking

  weiya711

  committed
  14ce8e9

  Copy full SHA for 14ce8e9

• Add in blocking code...still may be a little bit buggy. Has a lot of print statements

  weiya711

  committed
  34732e9

  Copy full SHA for 34732e9

Commits on Dec 8, 2020

• enable A(1,2) = 3 writes, and define custom cast operator for Scalars

  mattlkf

  committed
  7ee2315

  Copy full SHA for 7ee2315

Commits on Dec 9, 2020

• Access Matrix/Vector classes with IndexVars

  mattlkf

  committed
  b7ba7fa

  Copy full SHA for b7ba7fa

• assign IndexExpr to Scalar properly

  mattlkf

  committed
  ce9a605

  Copy full SHA for ce9a605

Commits on Dec 10, 2020

• quick test that printing and reading the ScalarAccess returned by A(1,1) works

  mattlkf

  committed
  3e8f4a6

  Copy full SHA for 3e8f4a6

Commits on Dec 15, 2020

• Merge branch 'master' into insert-better

  mattlkf

  committed
  6ad76a7

  Copy full SHA for 6ad76a7

• change indexexpr test to show mixed use of tensor and matrix in linalg expression

  mattlkf

  committed
  79c61a5

  Copy full SHA for 79c61a5

• fix warning about initialization out of order

  mattlkf

  committed
  711d7c8

  Copy full SHA for 711d7c8

• fixed another Matrix constructor

  mattlkf

  committed
  7bd3e89

  Copy full SHA for 7bd3e89

• fixed another Matrix constructor

  mattlkf

  committed
  b9ac34f

  Copy full SHA for b9ac34f

• fixed two more Matrix constructors

  mattlkf

  committed
  15a41d5

  Copy full SHA for 15a41d5

• fixed simplest Matrix constructor and commented out two Matrix constructors I think we should remove

  mattlkf

  committed
  1f4f574

  Copy full SHA for 1f4f574

• Fix minor mistake in test

  weiya711

  committed
  f6fe343

  Copy full SHA for f6fe343

• Rename indexVars for linalg compiler and remove blocking tests

  weiya711

  committed
  0feaa60

  Copy full SHA for 0feaa60

Commits on Dec 16, 2020

• remove unwanted constructors + remove optional colVec arg from simplest Vector constructor

  mattlkf

  committed
  1ed47a4

  Copy full SHA for 1ed47a4

• rename tensorbase test

  mattlkf

  committed
  238cbbf

  Copy full SHA for 238cbbf

• remove tensorapi test

  mattlkf

  committed
  e90083a

  Copy full SHA for e90083a

• removing some cruft

  mattlkf

  committed
  f070811

  Copy full SHA for f070811

• Change rewrite to use LinalgNotationRewriter

  weiya711

  committed
  e9b2cb5

  Copy full SHA for e9b2cb5

• get ASSERT_TENSOR_EQ working, update tests to use new ScalarAccess methods

  mattlkf

  committed
  40101c8

  Copy full SHA for 40101c8

• Fixes to rewrite using LinalgNotationRewriter

  weiya711

  committed
  02188df

  Copy full SHA for 02188df

Commits on Dec 17, 2020

• more tests cleanup

  mattlkf

  committed
  f311d36

  Copy full SHA for f311d36

• merge in master

  mattlkf

  committed
  e77c9a2

  Copy full SHA for e77c9a2

• remove unnecessary couts and large commented-out section for rewrite

  mattlkf

  committed
  9efe965

  Copy full SHA for 9efe965

• merge upstream taco

  mattlkf

  committed
  ee2c596

  Copy full SHA for ee2c596

• Fix some things with rewriter and Scalar LinalgBase constructor and add new compound test

  weiya711

  committed
  db3b5c0

  Copy full SHA for db3b5c0

Commits on Dec 18, 2020

• Remove bool useTensorBase from Scalar constructor

  weiya711

  committed
  470a1c5

  Copy full SHA for 470a1c5

• allow for vector ASSERT_TENSOR_EQ too

  mattlkf

  committed
  2ebb31f

  Copy full SHA for 2ebb31f

• Merge branch 'master' of github.com:mattlkf/taco

  mattlkf

  committed
  0a39d92

  Copy full SHA for 0a39d92

• add some Scalar tests (and allow for direct assignment of value to Scalar)

  mattlkf

  committed
  e8399dd

  Copy full SHA for e8399dd

• minor cleanup

  mattlkf

  committed
  542566a

  Copy full SHA for 542566a

Commits on Dec 22, 2020

• Add in test that sometimes fails

  weiya711

  committed
  ee18452

  Copy full SHA for ee18452

Commits on Jan 5, 2021

• WIP debugging linalg and tensor api

  weiya711

  committed
  0e39c8a

  Copy full SHA for 0e39c8a

Commits on Jan 14, 2021

• Make LinalgTensorBaseNode subclass of LinalgVarNode

  Show description for da4c9ef
  mattlkf

  committed
  da4c9ef

  Copy full SHA for da4c9ef

• test printing explicitly, fix extra newline when printing Linalg

  mattlkf

  committed
  1726246

  Copy full SHA for 1726246

• merge upstream master

  mattlkf

  committed
  951e840

  Copy full SHA for 951e840

• Fix command line flags for linalg to be correct

  weiya711

  committed
  e0fe427

  Copy full SHA for e0fe427

• Remove print statements in linalg parser

  weiya711

  committed
  d5fcc3d

  Copy full SHA for d5fcc3d