Code

Cocktailer Artifact#518

Code

Merged

xysmlx merged 199 commits intomicrosoft:cocktailer_artifactmicrosoft/nnfusion:cocktailer_artifactfrom heheda12345:masterheheda12345/nnfusion:masterCopy head branch name to clipboardApr 24, 2023

+226,978-891

Commits

Commits on Jul 15, 2021

• some util functions for onnx frontend

  xysmlx

  committed
  a488ac2

  Copy full SHA for a488ac2

• more datatype support in constant op

  xysmlx

  committed
  ae261c7

  Copy full SHA for ae261c7

• draft: onnx frontend support for if and loop op

  xysmlx

  committed
  73d3e6f

  Copy full SHA for 73d3e6f

• draft: if and loop op_define

  xysmlx

  committed
  6cff7e7

  Copy full SHA for 6cff7e7

• refactor GraphConvert of the ONNX frontend, support if and loop convert

  xysmlx

  committed
  ed0221a

  Copy full SHA for ed0221a

• draft: cuda kernel emitter for if and loop op (placeholder)

  xysmlx

  committed
  1a30651

  Copy full SHA for 1a30651

• update onnx frontend convert and shape inference for loop op

  xysmlx

  committed
  436d7d3

  Copy full SHA for 436d7d3

Commits on Jul 20, 2021

• fix output size bug in ONNX Loop op convert

  xysmlx

  committed
  a2863d7

  Copy full SHA for a2863d7

• Generic_op_define and ONNX opset_11 frontend converter for ScatterND op

  xysmlx

  committed
  b7b66b2

  Copy full SHA for b7b66b2

• Comment m_expression construction of generic_op to bypass translate function problem

  xysmlx

  committed
  ffae3f7

  Copy full SHA for ffae3f7

Commits on Aug 10, 2021

• Merge branch 'master' into control_flow

  xysmlx

  committed
  f771004

  Copy full SHA for f771004

Commits on Nov 17, 2021

• Merge master branch into control_flow branch

  xysmlx

  committed
  811a126

  Copy full SHA for 811a126

Commits on Nov 18, 2021

• disable ORT optimizations

  xysmlx

  committed
  a77ee7b

  Copy full SHA for a77ee7b

• fix bug for disabling ORT optimizations

  xysmlx

  committed
  d42be30

  Copy full SHA for d42be30

• fix bug for disabling ORT optimizations

  xysmlx

  committed
  0a29cc1

  Copy full SHA for 0a29cc1

Commits on Nov 27, 2021

• temp

  nox-410

  committed
  2f29e0d

  Copy full SHA for 2f29e0d

Commits on Nov 28, 2021

• Support -ftuning_list in kernel tuning pass

  xysmlx

  committed
  b0d4c56

  Copy full SHA for b0d4c56

• Merge remote-tracking branch 'origin/control_flow' into control_flow_1

  nox-410

  committed
  faedc9b

  Copy full SHA for faedc9b

Commits on Dec 2, 2021

• Implement If and Loop code gen

  nox-410

  committed
  6812438

  Copy full SHA for 6812438

• add mod operator

  xysmlx

  committed
  324f2cf

  Copy full SHA for 324f2cf

• Add recursion

  nox-410

  committed
  e45ce1e

  Copy full SHA for e45ce1e

• Merge remote-tracking branch 'origin/control_flow' into control_flow_1

  nox-410

  committed
  5a2955a

  Copy full SHA for 5a2955a

Commits on Dec 3, 2021

• add -fcodegen_pybind

  xiayuqing0622

  committed
  df4ea2e

  Copy full SHA for df4ea2e

• fix bug

  xiayuqing0622

  committed
  2d2a9c5

  Copy full SHA for 2d2a9c5

Commits on Dec 4, 2021

• Fix code for cudaEmitter

  nox-410

  committed
  35212e6

  Copy full SHA for 35212e6

• Fix recursion kernel name

  nox-410

  committed
  ef037e1

  Copy full SHA for ef037e1

• Remove unused param in control flow

  nox-410

  committed
  ee06cb9

  Copy full SHA for ee06cb9

Commits on Dec 6, 2021

• Create base class for controlflow emitter

  nox-410

  committed
  a72c89f

  Copy full SHA for a72c89f

• Recursion Op workspace allocation

  nox-410

  committed
  d67fa83

  Copy full SHA for d67fa83

Commits on Dec 7, 2021

• python patch

  heheda12345

  committed
  14536eb

  Copy full SHA for 14536eb

• add kernel_entry

  heheda12345

  committed
  51f5de2

  Copy full SHA for 51f5de2

• remove half

  heheda12345

  committed
  3ff6e40

  Copy full SHA for 3ff6e40

• allocate tensor in c

  heheda12345

  committed
  5eb9ef4

  Copy full SHA for 5eb9ef4

• torch::tensor for one output

  heheda12345

  committed
  3b62fc8

  Copy full SHA for 3b62fc8

• tmp fix

  xiayuqing0622

  committed
  912c126

  Copy full SHA for 912c126

• fix

  nox-410

  committed
  beafc0b

  Copy full SHA for beafc0b

• list of tensor

  heheda12345

  committed
  dbde8dd

  Copy full SHA for dbde8dd

• Merge branch 'master' of github.com:heheda12345/nnfusion

  

  Chen Zhang

  committed
  8016924

  Copy full SHA for 8016924

• fix bias dim

  heheda12345

  committed
  7b1bdb0

  Copy full SHA for 7b1bdb0

Commits on Dec 8, 2021

• apply shared memory

  nox-410

  committed
  7648008

  Copy full SHA for 7648008

• stderr

  heheda12345

  committed
  0023532

  Copy full SHA for 0023532

• pybind int64

  heheda12345

  committed
  b6e052a

  Copy full SHA for b6e052a

Commits on Dec 9, 2021

• Fix recursion

  nox-410

  committed
  15bdba5

  Copy full SHA for 15bdba5

• some parameter changes

  nox-410

  committed
  07a7f59

  Copy full SHA for 07a7f59

• Merge branch 'control_flow_2' of https://github.com/nox-410/nnfusion

  heheda12345

  committed
  80726df

  Copy full SHA for 80726df

• bypass reshape and broadcast

  nox-410

  committed
  1c6f23a

  Copy full SHA for 1c6f23a

• fix duplicate node

  nox-410

  committed
  e3cfae0

  Copy full SHA for e3cfae0

Commits on Dec 10, 2021

• bugfix

  nox-410

  committed
  6176f8a

  Copy full SHA for 6176f8a

• Bypass GatherV2 & merge code

  nox-410

  committed
  3c22fce

  Copy full SHA for 3c22fce

• Merge branch 'control_flow_2' of https://github.com/nox-410/nnfusion

  heheda12345

  committed
  5dd292e

  Copy full SHA for 5dd292e

• Adjust parameter

  nox-410

  committed
  e9e9bd2

  Copy full SHA for e9e9bd2

Commits on Dec 11, 2021

• Merge remote-tracking branch 'zc/master' into control_flow_finetune

  nox-410

  committed
  7a58506

  Copy full SHA for 7a58506

• Merge branch 'control_flow_2' of https://github.com/nox-410/nnfusion

  heheda12345

  committed
  b694f34

  Copy full SHA for b694f34

• fix a bug related with gatherV2

  nox-410

  committed
  a23d8aa

  Copy full SHA for a23d8aa

• Merge branch 'control_flow_2' into control_flow_finetune

  nox-410

  committed
  38cfaaa

  Copy full SHA for 38cfaaa

Commits on Dec 12, 2021

• fix reshape & broadcast bypass

  nox-410

  committed
  bec08a9

  Copy full SHA for bec08a9

• fix reshape & broadcast bypass

  nox-410

  committed
  5071d8d

  Copy full SHA for 5071d8d

Commits on Dec 13, 2021

• add threadfence

  nox-410

  committed
  8c57871

  Copy full SHA for 8c57871

• Merge branch 'control_flow_2' into control_flow_finetune

  nox-410

  committed
  bb4e4a0

  Copy full SHA for bb4e4a0

Commits on Dec 23, 2021

• Support multiple outputs

  nox-410

  committed
  0598667

  Copy full SHA for 0598667

• Merge branch 'control_flow_2' into control_flow_finetune

  nox-410

  committed
  4888a31

  Copy full SHA for 4888a31

• Add loop initialization

  nox-410

  committed
  8d992b2

  Copy full SHA for 8d992b2

• Fix extern result memory for Loop

  nox-410

  committed
  bb54c8a

  Copy full SHA for bb54c8a

Commits on Dec 25, 2021

• Support broadcast Matmul

  nox-410

  committed
  6b57878

  Copy full SHA for 6b57878

• Adjust parameters

  nox-410

  committed
  287b0be

  Copy full SHA for 287b0be

• Allow scalar float in torch codegen

  nox-410

  committed
  6242182

  Copy full SHA for 6242182

• Skip inplace analysis for subgraphs

  nox-410

  committed
  6c3a48d

  Copy full SHA for 6c3a48d

Commits on Dec 27, 2021

• Fix Reshape error in Control flow

  nox-410

  committed
  2016e0d

  Copy full SHA for 2016e0d

• Use injected SumOp

  nox-410

  committed
  2e43315

  Copy full SHA for 2e43315

Commits on Dec 28, 2021

• update Dot kernel cache

  nox-410

  committed
  3233154

  Copy full SHA for 3233154

Commits on Dec 29, 2021

• Fix controlflow inplace

  nox-410

  committed
  74059ad

  Copy full SHA for 74059ad

• Set memory reuse to false

  nox-410

  committed
  ebf0679

  Copy full SHA for ebf0679

• Merge branch 'control_flow_finetune' of https://github.com/nox-410/nnfusion

  heheda12345

  committed
  28f3b07

  Copy full SHA for 28f3b07

Commits on Jan 1, 2022

• grid.sync() & elementwise

  heheda12345

  committed
  11f9caf

  Copy full SHA for 11f9caf

• add scatternd op

  heheda12345

  committed
  5fd70b2

  Copy full SHA for 5fd70b2

Commits on Jan 7, 2022

• -fcheck_result

  heheda12345

  committed
  cad0688

  Copy full SHA for cad0688

Commits on Jan 8, 2022

• add control edge to loop graph & fix kernel fusion

  heheda12345

  committed
  a87cfb5

  Copy full SHA for a87cfb5

• concat with fewer resource

  heheda12345

  committed
  d68b2e7

  Copy full SHA for d68b2e7

Commits on Jan 9, 2022

• concat: no implace

  heheda12345

  committed
  94e78d4

  Copy full SHA for 94e78d4

• elementwise: general blockdim

  heheda12345

  committed
  4429553

  Copy full SHA for 4429553

• hack: add more dependency

  heheda12345

  committed
  2d8229b

  Copy full SHA for 2d8229b

• forward control edge

  heheda12345

  committed
  98ea31b

  Copy full SHA for 98ea31b

• remove useless barrier

  heheda12345

  committed
  8cf153c

  Copy full SHA for 8cf153c

Commits on Jun 16, 2022

• add return

  heheda12345

  committed
  7616e7c

  Copy full SHA for 7616e7c

Commits on Jul 22, 2022

• fix bug in conv-bn

  heheda12345

  committed
  2d435fa

  Copy full SHA for 2d435fa

Commits on Jul 25, 2022

• add roller

  heheda12345

  committed
  903c774

  Copy full SHA for 903c774

• debug tensor

  heheda12345

  committed
  fd0dc0d

  Copy full SHA for fd0dc0d

• add roller as submodule

  heheda12345

  committed
  847be21

  Copy full SHA for 847be21

• upate gitignore

  heheda12345

  committed
  db67a1a

  Copy full SHA for db67a1a

• change weight of inner graph to Constant op

  heheda12345

  committed
  55706ff

  Copy full SHA for 55706ff

• gnode cout

  heheda12345

  committed
  0b71dba

  Copy full SHA for 0b71dba

• identity op

  heheda12345

  committed
  626c66e

  Copy full SHA for 626c66e

• blockCudaEmitter: emit parameters from function sig

  heheda12345

  committed
  e2b7995

  Copy full SHA for e2b7995

Commits on Jul 26, 2022

• draft version of seperate kernel launch, conflict wtih postprocessing now

  heheda12345

  committed
  ba24cbb

  Copy full SHA for ba24cbb

• fix typo

  heheda12345

  committed
  ad3802d

  Copy full SHA for ad3802d

Commits on Jul 27, 2022

• two branch call finish

  heheda12345

  committed
  b3d0f8c

  Copy full SHA for b3d0f8c

Commits on Aug 11, 2022

• fix concat op

  heheda12345

  committed
  0e9e76a

  Copy full SHA for 0e9e76a

• update op frontend

  heheda12345

  committed
  7ed651b

  Copy full SHA for 7ed651b

• conv to CNHW initial support

  heheda12345

  committed
  9f72880

  Copy full SHA for 9f72880

• add concat and reshape op

  heheda12345

  committed
  e75617e

  Copy full SHA for e75617e

Commits on Aug 15, 2022

• add if op to conv_layout_pass and fix related bugs

  heheda12345

  committed
  6a1e0c5

  Copy full SHA for 6a1e0c5

• fix bug in share memory allocation of if op

  heheda12345

  committed
  a0bc620

  Copy full SHA for a0bc620

Commits on Aug 18, 2022

• fuse small kernels (not finish yet)

  heheda12345

  committed
  01ac748

  Copy full SHA for 01ac748

• fuse small kernels

  heheda12345

  committed
  d1f988b

  Copy full SHA for d1f988b

Commits on Aug 23, 2022

• reorder the kernels

  heheda12345

  committed
  2f189d0

  Copy full SHA for 2f189d0

Commits on Aug 25, 2022

• impl d2h

  heheda12345

  committed
  ac90748

  Copy full SHA for ac90748

Commits on Aug 29, 2022

• fix bug in elementwise kernel

  heheda12345

  committed
  ac6e50f

  Copy full SHA for ac6e50f

• main_test 100+100, print ref

  heheda12345

  committed
  7a7d75e

  Copy full SHA for 7a7d75e

Commits on Aug 30, 2022

• fuse then else

  heheda12345

  committed
  e6bd194

  Copy full SHA for e6bd194

Commits on Aug 31, 2022

• move subtract out of if

  heheda12345

  committed
  6675125

  Copy full SHA for 6675125

Commits on Sep 5, 2022

• loop in c

  heheda12345

  committed
  bb1210e

  Copy full SHA for bb1210e

Commits on Sep 7, 2022

• fix small bugs for skipnet

  heheda12345

  committed
  b5c0a4e

  Copy full SHA for b5c0a4e

Commits on Sep 14, 2022

• CPU-GPU hybrid: assign stage

  heheda12345

  committed
  f658b61

  Copy full SHA for f658b61

• CPU-GPU hybrid: add d2h and h2d gnode

  heheda12345

  committed
  da1aaeb

  Copy full SHA for da1aaeb

• CPU-GPU hybrid: dumplicate memory pool

  heheda12345

  committed
  41609fc

  Copy full SHA for 41609fc

• CPU-GPU hybrid: call by tensor with _cpu

  heheda12345

  committed
  92445f4

  Copy full SHA for 92445f4

Commits on Sep 15, 2022

• CPU-GPU hybrid: forward stage info in element-wise fusion pass

  heheda12345

  committed
  645e78f

  Copy full SHA for 645e78f

• remove debug code

  heheda12345

  committed
  ad057ed

  Copy full SHA for ad057ed

• CPU-GPU hybrid: copy cpu emitter from cuda emitter

  heheda12345

  committed
  78614ca

  Copy full SHA for 78614ca

• CPU-GPU hybrid: for (int tid=0; tid <...)

  heheda12345

  committed
  84dbfa2

  Copy full SHA for 84dbfa2

• CPU-GPU hybrid: put result op on CPU

  heheda12345

  committed
  56bfcbf

  Copy full SHA for 56bfcbf

Commits on Sep 16, 2022

• CPU-GPU hybrid: bmm & conv codegen, can run

  heheda12345

  committed
  35adaba

  Copy full SHA for 35adaba

• CPU-GPU hybrid: avoid run to_cpu_pass in inner graph

  heheda12345

  committed
  9d2b6f1

  Copy full SHA for 9d2b6f1

Commits on Sep 27, 2022

• add cpu op

  heheda12345

  committed
  e914e1c

  Copy full SHA for e914e1c

• fix bugs in recursion

  heheda12345

  committed
  ccd9270

  Copy full SHA for ccd9270

• inline recursion call

  heheda12345

  committed
  ef8d23b

  Copy full SHA for ef8d23b

• recursive with stack

  heheda12345

  committed
  0d9e94c

  Copy full SHA for 0d9e94c

Commits on Sep 28, 2022

• add be_state_buffer and state_base

  heheda12345

  committed
  9d270ec

  Copy full SHA for 9d270ec

• add be_state_buffer and state_base to more place

  heheda12345

  committed
  fe943c7

  Copy full SHA for fe943c7

• fast barrier codegen

  heheda12345

  committed
  9cac765

  Copy full SHA for 9cac765

Commits on Oct 4, 2022

• check control edge in operator << (gnode)

  heheda12345

  committed
  294fa47

  Copy full SHA for 294fa47

Commits on Oct 5, 2022

• optimize elementwise perf

  heheda12345

  committed
  e358bc0

  Copy full SHA for e358bc0

• add pipeline fail to compile commands

  heheda12345

  committed
  f9e1e80

  Copy full SHA for f9e1e80

Commits on Oct 6, 2022

• eliminate copy back of cond

  heheda12345

  committed
  b7e558d

  Copy full SHA for b7e558d

Commits on Oct 7, 2022

• add bool to dtypes

  heheda12345

  committed
  88ee7bc

  Copy full SHA for 88ee7bc

Commits on Oct 9, 2022

• add translate_v2 for identity op

  heheda12345

  committed
  4c0f3d2

  Copy full SHA for 4c0f3d2

• avoid inplace opt when gnode i/o contains result

  heheda12345

  committed
  68e3643

  Copy full SHA for 68e3643

• fix bug in conv layout pass

  heheda12345

  committed
  b4b4e1a

  Copy full SHA for b4b4e1a

• cast_pytorch_tensor: use data_ptr instead of storage.data_ptr

  heheda12345

  committed
  dc2320b

  Copy full SHA for dc2320b

Commits on Oct 10, 2022

• add fused_max_grid to loop

  heheda12345

  committed
  99d2529

  Copy full SHA for 99d2529

Commits on Oct 11, 2022

• add more cpu op

  heheda12345

  committed
  452d049

  Copy full SHA for 452d049

• add naive impls for breakdown exp

  heheda12345

  committed
  c50e66c

  Copy full SHA for c50e66c

• skip scalar op: reshape

  heheda12345

  committed
  47f392e

  Copy full SHA for 47f392e

Commits on Oct 18, 2022

• is_outmost_graph for blockfusion

  heheda12345

  committed
  e798c76

  Copy full SHA for e798c76

Commits on Oct 19, 2022

• hacked parallel recursion: assume all calls can be executed in parallel

  heheda12345

  committed
  deccf1c

  Copy full SHA for deccf1c

Commits on Oct 20, 2022

• tune recursion

  heheda12345

  committed
  95ee0d4

  Copy full SHA for 95ee0d4

Commits on Nov 11, 2022

• add reduce-memcpy blockop

  heheda12345

  committed
  3691741

  Copy full SHA for 3691741

• argmax kernel (not tested)

  heheda12345

  committed
  001c376

  Copy full SHA for 001c376

• support while op, can compile but loop cannot stop

  heheda12345

  committed
  5e213d0

  Copy full SHA for 5e213d0

• alloc cond tensor, fix bug in parameter mapping, can run bs=1

  heheda12345

  committed
  e343fe7

  Copy full SHA for e343fe7

Commits on Nov 12, 2022

• hardcode num_local_thread_sync in reduce.hpp because emit_function_body is called multiple times

  heheda12345

  committed
  8668be8

  Copy full SHA for 8668be8

• while in c

  heheda12345

  committed
  62fcf06

  Copy full SHA for 62fcf06

• fast barrier for single block

  heheda12345

  committed
  4b7373d

  Copy full SHA for 4b7373d

• fix bug in fast barrier of single block

  heheda12345

  committed
  18e60a3

  Copy full SHA for 18e60a3

Commits on Nov 14, 2022

• fix bug in argmax and element_fused, pass while_op

  heheda12345

  committed
  3dfae17

  Copy full SHA for 3dfae17

Commits on Nov 17, 2022

• enalbe & disable result d2d inplace

  heheda12345

  committed
  823fc2a

  Copy full SHA for 823fc2a

Commits on Nov 22, 2022

• support different schedule of if inside while op

  heheda12345

  committed
  9215c13

  Copy full SHA for 9215c13

Commits on Nov 28, 2022

• extend elementwise to support simple broadcast

  heheda12345

  committed
  865c613

  Copy full SHA for 865c613

• extend scatternd to support index array

  heheda12345

  committed
  08f8504

  Copy full SHA for 08f8504

• reshape memcpy block kernel

  heheda12345

  committed
  80cc929

  Copy full SHA for 80cc929

• softmax block kernel

  heheda12345

  committed
  652677f

  Copy full SHA for 652677f

• batchmatmul with broadcast

  heheda12345

  committed
  117cb7f

  Copy full SHA for 117cb7f

• small fix

  heheda12345

  committed
  e4acd37

  Copy full SHA for e4acd37

• manually set max_block_dim for bcast and elementwise

  heheda12345

  committed
  27bdc7d

  Copy full SHA for 27bdc7d

Commits on Dec 2, 2022

• sync for rocm

  heheda12345

  committed
  428edb6

  Copy full SHA for 428edb6

Commits on Mar 27, 2023

• merge rocm code

  heheda12345

  committed
  f235896

  Copy full SHA for f235896

Commits on Mar 28, 2023

• add IfSingle operator

  heheda12345

  committed
  ecf47cc

  Copy full SHA for ecf47cc

Commits on Mar 30, 2023

• reorganize parameters

  heheda12345

  committed
  547136d

  Copy full SHA for 547136d

Commits on Apr 3, 2023

• remove cudadevicereset

  heheda12345

  committed
  c8cce31

  Copy full SHA for c8cce31

• fix depunit bug in loop

  heheda12345

  committed
  60a287a

  Copy full SHA for 60a287a

• dump kerneldb requests

  heheda12345

  committed
  f5e6739

  Copy full SHA for f5e6739

• search unroll width

  heheda12345

  committed
  65c6f4f

  Copy full SHA for 65c6f4f

Commits on Apr 10, 2023

• fix blockfusion sync problem

  heheda12345

  committed
  f0c358d

  Copy full SHA for f0c358d

Commits on Apr 14, 2023

• wrap python part with ifdef

  heheda12345

  committed
  f138341

  Copy full SHA for f138341

Commits on Apr 17, 2023

• add __syncthreads() to cf kernels

  heheda12345

  committed
  845c52c

  Copy full SHA for 845c52c

Commits on Apr 18, 2023

• copy file from ControlFlow repo

  heheda12345

  committed
  ac6bd1f

  Copy full SHA for ac6bd1f

• change path

  heheda12345

  committed
  77fbf5b

  Copy full SHA for 77fbf5b

• fix bug in scripts

  heheda12345

  committed
  8266a71

  Copy full SHA for 8266a71

Commits on Apr 19, 2023

• add more guides

  heheda12345

  committed
  f2d7df3

  Copy full SHA for f2d7df3

• add more guides

  heheda12345

  committed
  246385a

  Copy full SHA for 246385a

• remove cudnn

  heheda12345

  committed
  e4e02fe

  Copy full SHA for e4e02fe

• install_grinder script

  heheda12345

  committed
  16713b3

  Copy full SHA for 16713b3

• remove cudnn in manual

  heheda12345

  committed
  9e3c80f

  Copy full SHA for 9e3c80f

Commits on Apr 20, 2023

• autotvm kernel

  heheda12345

  committed
  ec00f1e

  Copy full SHA for ec00f1e

• remove training code

  heheda12345

  committed
  2ba5cf4

  Copy full SHA for 2ba5cf4

• change permission

  heheda12345

  committed
  e568ef6

  Copy full SHA for e568ef6

• update kernels in manual impls

  heheda12345

  committed
  a6514e7

  Copy full SHA for a6514e7

Commits on Apr 21, 2023

• add rocm kerneldb script

  heheda12345

  committed
  afdae02

  Copy full SHA for afdae02

• copy roller rocm code

  heheda12345

  committed
  06a0ff2

  Copy full SHA for 06a0ff2

Commits on Apr 22, 2023

• first try of rocm kerneldb

  heheda12345

  committed
  186fe5b

  Copy full SHA for 186fe5b

• rocm reproduced

  heheda12345

  committed
  48d0651

  Copy full SHA for 48d0651

• remove grinder from filename

  heheda12345

  committed
  767c6f8

  Copy full SHA for 767c6f8

Commits on Apr 23, 2023

• kerneldb scripts

  heheda12345

  committed
  2632ad8

  Copy full SHA for 2632ad8

• finish rocm?

  heheda12345

  committed
  dd1796a

  Copy full SHA for dd1796a

• remove name 'grinder' from scripts

  heheda12345

  committed
  ec0594f

  Copy full SHA for ec0594f

• update gitignore

  heheda12345

  committed
  063996b

  Copy full SHA for 063996b

• small fix

  heheda12345

  committed
  b1a889b

  Copy full SHA for b1a889b

Commits on Apr 24, 2023

• rename project and remove some script

  heheda12345

  committed
  071ded7

  Copy full SHA for 071ded7

• update links

  heheda12345

  committed
  7d0c87a

  Copy full SHA for 7d0c87a