Code

Merge llvm branch to master#932

Code

Draft

iomaganaris wants to merge 105 commits intomasterBlueBrain/nmodl:masterfrom magkanar/llvm-rebaseBlueBrain/nmodl:magkanar/llvm-rebaseCopy head branch name to clipboard

+12,796-1,002

Commits

Commits on Sep 15, 2022

• Disable python bindings for faster build

  Show description for d30c027
  

  pramodk

  authored and

  iomaganaris

  committed
  d30c027

• Integrate LLVM into CMake build system

  Show description for ebe6539
  

  pramodk

  authored and

  iomaganaris

  committed
  ebe6539

• Code infrastructure for LLVM code generation backend

  Show description for 8994f8e
  

  pramodk

  authored and

  iomaganaris

  committed
  8994f8e

• Azure CI fixes for LLVM build and README update

  Show description for a3f7891
  

  pramodk

  authored and

  iomaganaris

  committed
  a3f7891

• Print build status after cmake configure stage

  Show description for 3b17307
  

  pramodk

  authored and

  iomaganaris

  committed
  3b17307

• Adding test template for LLVM codegen

  Show description for cea865d
  

  pramodk

  authored and

  iomaganaris

  committed
  cea865d

• Initial LLVM codegen vistor routines (#457)

  Show description for 00e4ac0
  

  2 peopleauthored and

  iomaganaris

  committed
  00e4ac0

• FunctionBlock code generation and terminator checks (#470)

  Show description for 6623eb7
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  6623eb7

• Add option to run LLVM optimisation passes (#471)

  Show description for 13b129b
  

  pramodk

  authored and

  iomaganaris

  committed
  13b129b

• Add function call LLVM code generation (#477)

  Show description for 9c127f2
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  9c127f2

• Support for IndexedName codegen (#478)

  Show description for 470d54a
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  470d54a

• Improvements for code generation specific transformations (#483)

  Show description for 1674f3b
  

  pramodk

  authored and

  iomaganaris

  committed
  1674f3b

• nrn_state function generation in NMODL AST to help LLVM codegen (#484)

  Show description for f6d8b85
  

  pramodk

  and

  iomaganaris

  committed
  f6d8b85

• Running functions from MOD files via LLVM JIT (#482)

  Show description for 5e0fee0
  

  2 peopleauthored and

  iomaganaris

  committed
  5e0fee0

• Extended support for binary ops and refactoring (#489)

  Show description for 34bbaab
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  34bbaab

• Avoid converting LOCAL statement in all StatementBlocks (#492)

  Show description for 1f4c8dc
  

  pramodk

  authored and

  iomaganaris

  committed
  1f4c8dc

• Handle CodegenVarType type in JSON printer (#494)

  Show description for aa639de
  

  pramodk

  authored and

  iomaganaris

  committed
  aa639de

• Integrating LLVM helper into LLVM visitor (#497)

  Show description for a541c69
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  a541c69

• LLVM code generation for if/else statements (#499)

  Show description for 87baf0f
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  87baf0f

• Added error handling for values not in scope (#502)

  Show description for 05b721f
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  05b721f

• Added support for WHILE statement (#501)

  Show description for 5077c68
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  5077c68

• Create mechanism instance struct in LLVM IR (#507)

  Show description for 2223d00
  iomaganaris

  committed
  2223d00

• Printf support in LLVM IR codegen (#510)

  Show description for 9ca602e
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  9ca602e

• Fix issue error: ‘runtime_error’ is not a member of ‘std’ (#512)

  iomaganaris

  committed
  b3f6fa2

• Move code gen specific InstanceStruct node to codegen.yaml (#526)

  Show description for f15e3c5
  

  pramodk

  authored and

  iomaganaris

  committed
  f15e3c5

• * Improvements to codegen helper (Part I)

  Show description for f5dc06b
  

  pramodk

  authored and

  iomaganaris

  committed
  f5dc06b

• Addressing TODOs for Instance struct (#533) Part II

  Show description for 451fe17
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  451fe17

• Unit test for scalar state kernel generation in LLVM (#547)

  Show description for e1e8eab
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  e1e8eab

• Indexed name codegen improvements (#550)

  Show description for fd2053e
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  fd2053e

• Add InstanceStruct test data generation helper and unit test (#546)

  Show description for 57cb77d
  

  iomaganaris

  and

  pramodk

  committed
  57cb77d

• Add the remainder loop for vectorization of DERIVATIVE block (#534)

  Show description for 5768d68
  

  Nicolas Cornu

  authored and

  iomaganaris

  committed
  5768d68

• Always initialize return variable in function block (#554)

  Show description for 0497ee3
  

  Nicolas Cornu

  authored and

  iomaganaris

  committed
  0497ee3

• Running a kernel with NMODL-LLVM JIT (#549)

  Show description for 5989460
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  5989460

• Loop epilogue fix for LLVM visitor helper (#567)

  Show description for b5ceca6
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  b5ceca6

• Gather support and vectorisation fixes for LLVM code generation (#568)

  Show description for 9e2284b
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  9e2284b

• Verification and file utilities for LLVM IR codegen (#582)

  Show description for 0cb1cea
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  0cb1cea

• Add gather execution test (#591)

  

  georgemitenkov

  authored and

  iomaganaris

  committed
  c839e68

• Fixed loop allocations (#590)

  Show description for cffb50d
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  cffb50d

• Benchmarking LLVM code generation (#583)

  Show description for 481c728
  

  2 peopleauthored and

  iomaganaris

  committed
  481c728

• Minor benchmarking improvement (#593)

  Show description for 8b2d598
  

  pramodk

  authored and

  iomaganaris

  committed
  8b2d598

• Bug fix in codegen helper: delete LOCAL statement (#595)

  Show description for 2d67af2
  

  pramodk

  authored and

  iomaganaris

  committed
  2d67af2

• LLVM 13 compatibility and fixing void* type (#603)

  Show description for c884dd8
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  c884dd8

• Allow LOCAL variable inside StatementBlock for LLVM IR generation (#599)

  Show description for 3c38a2e
  

  pramodk

  authored and

  iomaganaris

  committed
  3c38a2e

• Update CI with LLVM v13 (trunk) (#605)

  Show description for 1f250ee
  

  pramodk

  authored and

  iomaganaris

  committed
  1f250ee

• Integrating vector maths library into LLVM codegen (#604)

  Show description for f6dee6e
  

  2 peopleauthored and

  iomaganaris

  committed
  f6dee6e

• Using shared libraries in LLVM JIT (#609)

  Show description for b50027a
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  b50027a

• Avoid local std::ofstream object causing segfault (#614)

  Show description for 4c9e1e1
  

  pramodk

  authored and

  iomaganaris

  committed
  4c9e1e1

• Refactoring of runners' infrastructure and dumping object files (#620)

  Show description for be984b8
  

  2 peopleauthored and

  iomaganaris

  committed
  be984b8

• Optimisation levels for benchmarking (#623)

  Show description for 3cf65cf
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  3cf65cf

• Adding function debug information (#628)

  Show description for 2356c19
  

  2 peopleauthored and

  iomaganaris

  committed
  2356c19

• Fixed using benchmarking_info in TestRunner (#631)

  

  georgemitenkov

  authored and

  iomaganaris

  committed
  92eadeb

• Fixed addition of SOLVE block to kernel's FOR loop (#636)

  Show description for dbccdaa
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  dbccdaa

• IR builder redesign for LLVM IR code generation pipeline (#634)

  Show description for e299af1
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  e299af1

• Fixed initialisation of CodegenAtomicStatement (#642)

  Show description for acbcd1b
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  acbcd1b

• Fix instance struct data generation for testing/benchmarking (#641)

  Show description for acfd3c3
  

  pramodk

  authored and

  iomaganaris

  committed
  acfd3c3

• Basic scatter support (#643)

  Show description for 0e09468
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  0e09468

• Benchmarking code re-organisation and minor improvements (#647)

  Show description for f7d00dd
  

  pramodk

  authored and

  iomaganaris

  committed
  f7d00dd

• Added attributes and metadata to LLVM IR compute kernels (#648)

  Show description for ac283f2
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  ac283f2

• Added loaded value to the stack (#655)

  Show description for 05e9cfa
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  05e9cfa

• Basic predication support for LLVM backend (#652)

  Show description for 9953758
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  9953758

• Improvements for LLVM code generation and benchmarking (#661)

  Show description for 7b45925
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  7b45925

• Fixed allocas insertion point for LLVM backend (#663)

  Show description for dcaff9a
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  dcaff9a

• Fast math flags for LLVM backend (#662)

  Show description for 9109139
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  9109139

• Avoid generating LLVM IR for Functions and Procedures if inlined (#664)

  iomaganaris

  committed
  fa5c7bf

• Fixed typo in benchmarking metrics (#665)

  

  georgemitenkov

  authored and

  iomaganaris

  committed
  30e53c7

• Remove only inlined blocks from AST based on symtab properties (#668)

  iomaganaris

  committed
  0362f66

• Use VarName on the RHS of assignment expression (#669)

  Show description for 480f26e
  

  pramodk

  authored and

  iomaganaris

  committed
  480f26e

• [LLVM] SLEEF and libsystem_m vector libraries support (#674)

  Show description for b907544
  

  2 peopleauthored and

  iomaganaris

  committed
  b907544

• [LLVM] Enhancements for optimization pipeline (#683)

  Show description for a1c4b0f
  

  2 peopleauthored and

  iomaganaris

  committed
  a1c4b0f

• [LLVM] Added saving to file utility (#685)

  Show description for 6506fcf
  

  2 peopleauthored and

  iomaganaris

  committed
  6506fcf

• [LLVM] Aliasing and cpu options for LLVM visitor and the benchmark (#686)

  Show description for d48bb20
  

  georgemitenkov

  and

  iomaganaris

  committed
  d48bb20

• Fix azure yaml pipeline from merge (#687)

  

  pramodk

  authored and

  iomaganaris

  committed
  c95dec9

• [LLVM] Support for newer versions of LLVM APIs

  Show description for 6f5e037
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  6f5e037

• Fix build issues for the rebased branch

  Show description for 58476bd
  

  pramodk

  authored and

  iomaganaris

  committed
  58476bd

• [LLVM] Allocate InstanceStruct on the GPU using cudaMallocManaged (#815)

  Show description for a30d52e
  iomaganaris

  committed
  a30d52e

• [LLVM][GPU] Separated CPU and GPU CLI options (#817)

  Show description for a755719
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  a755719

• [LLVM][refactoring] Added platform abstraction (#818)

  Show description for 412af4c
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  412af4c

• [LLVM][GPU] Added GPU-specific AST transformations (#819)

  Show description for 2d05aaa
  

  3 peoplecommitted
  2d05aaa

• [LLVM][GPU] Basic code generation for NVPTX backend (#820)

  Show description for 2e5e149
  

  2 peopleauthored and

  iomaganaris

  committed
  2e5e149

• Print kernel wrappers and nrn_init based on Instance Struct (#551)

  Show description for 6b9df33
  

  iomaganaris

  and

  pramodk

  committed
  6b9df33

• [LLVM][GPU] NVPTX specific passes for code generation (#833)

  Show description for 84a85d4
  

  georgemitenkov

  and

  iomaganaris

  committed
  84a85d4

• [LLVM] Code formatting changes (#838)

  Show description for afec6c6
  

  iomaganaris

  and

  Nicolas Cornu

  committed
  afec6c6

• [LLVM][GPU][+refactoring] Replacement of math intrinsics with library calls (#835)

  Show description for ac6d731
  

  georgemitenkov

  and

  iomaganaris

  committed
  ac6d731

• JIT invocation from python for benchmarks (#832)

  Show description for 5ffc590
  

  Omar Awile

  authored and

  iomaganaris

  committed
  5ffc590

• Fixes issue with debug printing of visitors (#854)

  Show description for af1ff70
  iomaganaris

  committed
  af1ff70

• Support for Breakpoint block (nrn_cur) for code generation (#645)

  Show description for 49a13af
  

  2 peopleauthored and

  iomaganaris

  committed
  49a13af

• [LLVM][GPU] Added CUDADriver to execute benchmark on GPU (#829)

  Show description for 8797f9b
  iomaganaris

  committed
  8797f9b

• [LLVM][GPU] Atomic updates support (#853)

  Show description for b13ca90
  

  georgemitenkov

  and

  iomaganaris

  committed
  b13ca90

• Replaced fmt literals with fmt::format

  iomaganaris

  committed
  ece5c47

• [LLVM][FIX] Float generation fix in LLVM helper visitor (#865)

  Show description for 0c8f566
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  0c8f566

• [CP-859] Code generation changes for "inline" scopmath solvers. (#859) (#868)

  Show description for a784b53
  

  iomaganaris

  and

  olupton

  committed
  a784b53

• [LLVM] Fixes compilation with LLVM codegen disabled (#867)

  Show description for 253f639
  iomaganaris

  committed
  253f639

• [CP-870] Cherry-pick sympy fix from master (#872)

  Show description for 77c4f74
  

  3 peoplecommitted
  77c4f74

• [LLVM][SIMD] Atomic updates support (#864)

  Show description for a5577bb
  

  georgemitenkov

  and

  iomaganaris

  committed
  a5577bb

• Install explicitly LLVM 13.0.1 in MacOS builds in Azure (#898)

  iomaganaris

  committed
  bf3c125

• Fix nrn_cur kernel code generation unit test (#892)

  Show description for 18856f4
  iomaganaris

  committed
  18856f4

• [LLVM][refactoring] Annotations pass and no more wrappers (#893)

  Show description for 44688e3
  

  georgemitenkov

  authored and

  iomaganaris

  committed
  44688e3

• [LLVM] Instantiate CodegenLLVMVisitor of PyJIT with fast math and debug symbols options (#912)

  Show description for d2211e0
  iomaganaris

  committed
  d2211e0

• Fix issues after rebase

  iomaganaris

  committed
  08555ee

• Update desired cmake version in github actions

  iomaganaris

  committed
  2188bb2

• Make Clang and CMake format happy

  iomaganaris

  committed
  958e9cd

• Disable llvm backend by default

  iomaganaris

  committed
  c5678e5

Commits on Sep 16, 2022

• Fix benchmark linking with visitors

  iomaganaris

  committed
  fee9a0b

• Setup sanitizer options for new test

  iomaganaris

  committed
  a967a3f

• Added supression for undefined behavior in std random number generator

  iomaganaris

  committed
  de5b144