* Fix first link in compatibility matrix table
* Revert "Fix first link in compatibility matrix table"
This reverts commit 069c5c116a.
* Remove sticky header and unused css
* Remove container from hardware specs matrix
---------
Co-authored-by: Peter Jun Park <peter.park@amd.com>
* Regenerate changelog
* Add component changelogs and known issue
Fix RELEASE.md headings
Update pub datestamp for 6.1.2
Add AMDSMI and ROCm SMI to 6.1.2 template
Add rccl and rocBLAS
Update intro blurb and headings
Add ROCm SMI fix
Add missed heading to AMDSMI
Update datestamp and release version number
Update version and release number
Add known issue re: MI300X error detection
Words
Add issue link
Rm GitHub issue link
Move known issue down
Update ki wording
Remove "this issue has been investigated ... " from known issue
Fix changelog h1
* Reorg known issue, upcoming changes, remove rocDecode tested configurations
* Add fixes from review
* Add fixed issue link
* Fix heading
* Remove known issue
* Update the links for rocminfo and rocm-bandwidth-test
* Update the links for rocminfo and rocm-bandwidth-test
* Update the links for rocminfo and rocm-bandwidth-test
* Update links to intersphinx links
---------
Co-authored-by: Peter Jun Park <peter.park@amd.com>
* Add Fine Tuning LLMs how to guide
* Reorg and refactor Fine-tuning LLMs with ROCm
Update index and headings
Fix formatting and update toc
Split out content from index to overview.rst
Add metadata
Clean up overview
Add inference sections, fix rst errors, clean up single-gpu-fine-tuning
Combine fine-tuning and inference guides
Fix some links and formatting
Update toc and add formatting fixes
Add ck kernel fusion content
Update toc
Clean up model quantization and acceleration
Add CK images
Clean up profiling
Update triton kernel performance optimization
Update llm inference frameworks guide
Disable automatic number of figures and tables in Sphinx conf
Change tabs to spaces
Change heading to end with -ing
Add link fixes and heading updates
Add rocprof/Omniperf/Omnitrace section
Update profiling and debugging guide
Add formatting fixes
Satisfy spellcheck
Fix words
Delete unused file
Finish overview
Clean up first 4 sections
Multi-gpu fine-tuning guide: slight fixes
Update toc
Remove tabs
Formatting fixes
* Minor wording updates
* Add some clean-up
* Update profiling and debugging gudie
* Fix Omnitrace link
* Update ck kernel fusion with latest
* Update CK formatting
* Fix perfetto link syntax
* Fix typos and add blurbs
* Add fixes to Triton optimization doc
* Tabify saving adapters / models section
* Fix linting errors - spellcheck
Fix spelling and grammar
Satisfy linter
Update wording in profiling guide
Add fixes to satisfy linter
More fixes for linting in Triton guide
More linting fixes
Spellcheck in CK guide
* Improve triton guide
Fix linting errors and optics
* Add occupancy / vgpr table
Change some wording
* Re-add tunableop
* Add missing indent in _toc.yml
* Remove ckProfiler references
* Add links to resources
* Add refs in CK optimization guide
* Rename files and fix internal links
* Organize tuning guides
Reorg triton
* Add compute unit diagram
* Remove AutoAWQ
* Add higher res image for Perfetto trace example
* Update link text
* Update fig nums
* Update some formatting
* Update "Inductor"
* Change "Inductor" to TorchInductor
* Add link to official TorchInductor docs
* Regenerate changelog
* Add component changelogs and known issue
Fix RELEASE.md headings
Update pub datestamp for 6.1.2
Add AMDSMI and ROCm SMI to 6.1.2 template
Add rccl and rocBLAS
Update intro blurb and headings
Add ROCm SMI fix
Add missed heading to AMDSMI
Update datestamp and release version number
Update version and release number
Add known issue re: MI300X error detection
Words
Add issue link
Rm GitHub issue link
Move known issue down
Update ki wording
Remove "this issue has been investigated ... " from known issue
Fix changelog h1
Template with bash commands to update cmake with snap.
Use template for two components that want updated cmake with latest source on their default branches.
* Add Using ROCm for AI:wq
Add PyTorch Docker installation images
Split doc into subtopics
Add metadata
Clean up index
Clean up hugging face guide
Clean up installation guide
Fix rST formatting
Clean up install and train-a-model
Clean up MAD
Delete unused file
Add ref anchors and clean up MAD doc
Add formatting fixes
Update toc and section index
Format some code blocks
Remove install guide and update toc
Chop installation guide
Clean up deployment and hugging face sections
Change headings to end in -ing
Fix spelling in Training a model
Delete MAD and split out install content
Fix formatting
Change words to satisfy spellcheck linter
* Add review suggestions and add helpful links
Co-authored-by: Leo Paoletti <164940351+lpaoletti@users.noreply.github.com>
Add helpful links and add review suggestions
Remove fine-tuning link and links to D5 and MAGMA
Update docs/how-to/rocm-for-ai/deploy-your-model.rst
Co-authored-by: Young Hui - AMD <145490163+yhuiYH@users.noreply.github.com>
Update DeepSpeed link
Add subheading to ML framework installation and closing blurb to hugging face models guide
* Reorder topics
* removed docker and pointed ROCm deps to our existing builds
* removed vmImage tag for pool
* added pip to apt list and renamed from rocFFT to hipFFT
* fixed spelling mistakes in rocmDependencies
* added correct apt dep for pip
* removed leading slash in the cmake flags
* changed cxx_compiler to /rocm/bin/hipcc
* added llvm-project, ROCR-Runtime, clr, and rocminfo to rocm deps
* added rocFFT as a rocm dependency
* removed docker and added our builds for components
* removed rocFFT from rocm deps
* Fixed typo in rocFFT value
* added rocprofiler-register to rocFFT and fixed typo in the dependencies-rocm file
* changed cxx compiler to amdclang++
* fixed amdclang++ paths
* moving to faster machine
* added cmake module paths
* switched back to medium build
* added libopm-dev to apt deps
* added libomp-14-dev to apt deps
* added aomp as a rocm dep
* added aomp as a rocm dep
* added hipcc as the cxx_compiler
* reverted back to clang++ as the cxx_compiler
* removed unmentioned rocm deps from the readme
* removed docker
* added python3-pip as an apt dep
* fixed compiler paths
* added hipRAND as a rocm dep
* added print statements to see directory structure
* adding a print statement into /agent/_work/1/s/build/library
* added -Tensile_rocm_assembler as a build flag
* removed a broken script line
* added D to tensile rocm assembler
* added DROCM_PATH to build flags
* fixed typo
* changed build pool from medium to base
* changed build pool from base to low
* added env variables using josephs pr
* removed docker from hipBLASLt and added rocm dependencies that point to our builds
* added pip to the apt packages array
* changed cmake_cxx_compiler env var ro amdclang++
* changed cmake_cxx_compiler env var to amdclang++
* changed cmake_cxx_compiler env var to hipcc
* changed cmake_cxx_compiler env var to hipcc
* changed clang to amdclang
* changed all refs mentioning hipcc to amdclang
* changed cmake_cxx_compiler back to hipcc
* added a HIP_PATH env var based off Tensile/Source/FindHIP.cmake
* added hipcc to HIP_PATH
* added rocm-cmake to rocm deps
* added rocRAND as a rocm dep
* removed dcmake_module flag
* added libomp-dev as an apt dep
* added aomp as a rocm dep
* added clang as an apt dep
* reverted changes back to how they appear in develop since this branch will be submitted for review
* removed unecessary flags
* adding -DCMAKE_CXX_COMPILER=$(Agent.BuildDirectory)/rocm/llvm/bin/amdclang++ -DCMAKE_C_COMPILER=$(Agent.BuildDirectory)/rocm/llvm/bin/amdclang back to see if these are vital to a successful build
* removed newline character
* Disable aomp offload build for initial ci-build work
* Missing dependency for medium pool use of rocPRIM
* Latest rocBLAS source needs added ROCm dependencies
* Rename 'Tuning guides' to 'Hardware optimization'
* Move deep learning to Install section
* Change 'Hardware' to 'System' to align with index.md
* Satisfy spellcheck linter
* adding new framework install graphic with JAX
* Fix link to ROCm libraries list
* crop framework_install graphic
* Reset .wordlist.txt update
* Prettify deep learning framework installation page
* Change spacing in list of frameworks
---------
Co-authored-by: Young Hui <young.hui@amd.com>
aomp build is not triggered by changes to aomp repo, but by updates to llvm-project and ROCR-Runtime, so trigger definition can remain this ROCm/ROCm repo.
Instead of using docker and apt install of ROCm component dependencies, use tarballs from Azure Pipeline builds to enable updates of ROCm interdependencies without waiting for releases..
