xiaoleis-nv d03f35f9b6

[MLIR][NVVM] Fix the datatype error for nvvm.mma.sync when the operand is bf16 (#122664) ...

The PR fixes the datatype error for `nvvm.mma.sync` when the operand is
`bf16`. This operation originally requires the A/B type to be `f16x2`
for the `bf16` MMA. However, it violates the NVVM intrinsic
[[here](372044ee09/llvm/include/llvm/IR/IntrinsicsNVVM.td (L119))],
where the A/B operand type should be `i32`. This is a bug, and there are
no tests in MLIR that cover this datatype.

```
    // mma bf16 -> s32 @ m16n8k16/m16n8k8
    !eq(gft,"m16n8k16🅰️bf16") : !listsplat(llvm_i32_ty, 4),
    !eq(gft,"m16n8k16🅱️bf16") : !listsplat(llvm_i32_ty, 2),
    !eq(gft,"m16n8k8🅰️bf16") : !listsplat(llvm_i32_ty, 2),
    !eq(gft,"m16n8k8🅱️bf16") : [llvm_i32_ty],
```

This PR addresses this bug and adds tests to guarantee correctness.

Co-authored-by: Xiaolei Shi <xiaoleis@nvidia.com>

2025-01-13 15:03:05 +05:30

..

Import

[MLIR] Enable import of non self referential alias scopes (#121987)

2025-01-08 13:40:05 +01:00

nvvm

[MLIR][NVVM] Refactor tests in nvvmir.mlir (#119731)

2024-12-13 19:42:24 +05:30

amx.mlir

[MLIR][LLVM] Remove typed pointer remnants from target tests (#71210)

2023-11-03 21:21:45 +01:00

arm-neon-2d.mlir

…

arm-neon.mlir

[mlir][ArmNeon] Adds Arm Neon SMMLA, UMMLA, and USMMLA Intrinsics (#80511)

2024-02-06 14:12:40 -08:00

arm-sme-invalid.mlir

[mlir][ArmSME] Add arm_sme.intr.cnts(b|h|w|d) intrinsics (#77319)

2024-01-09 09:05:31 +00:00

arm-sme.mlir

[mlir][ArmSME] Support 2-way widening outer products (#78975)

2024-01-31 09:13:18 +00:00

arm-sve.mlir

[mlir][ArmSVE] Add arm_sve.psel operation (#95764)

2024-06-19 13:33:23 +01:00

attribute-alias-scopes.mlir

[MLIR] Enable import of non self referential alias scopes (#121987)

2025-01-08 13:40:05 +01:00

attribute-tbaa.mlir

…

comdat.mlir

…

commandline.mlir

[clang][flang][mlir] Reapply "Support -frecord-command-line option (#102975)"

2024-10-14 08:44:24 -06:00

data-layout.mlir

[MLIR] Add index bitwidth to the DataLayout (#85927)

2024-03-21 09:07:57 +01:00

disjoint.mlir

[mlir][LLVM] Add disjoint flag (#115855)

2024-11-15 13:48:01 +01:00

exact.mlir

[mlir][LLVM] Add exact flag (#115327)

2024-11-08 13:56:44 +01:00

external-func-dialect-attr.mlir

…

fp-math-function-attributes.mlir

[MLIR][LLVM] Add ftz and fuse FP ops related function attribute support (#97812)

2024-07-09 09:34:43 +01:00

frame-pointer.mlir

[MLIR] Add support for frame pointers in MLIR (#72145)

2023-12-05 11:52:13 +01:00

gpu.mlir

Populate the llvm::GlobalVariable ELF section, with the attribute from the ObjectAttrs (#117246)

2024-11-22 07:58:45 -08:00

ident.mlir

[flang][mlir] Add llvm.ident metadata when compiling with flang

2024-08-12 11:56:19 -06:00

invalid-module.mlir

…

llvmir-debug.mlir

[mlir][debug] Support DIGenericSubrange. (#113441)

2024-10-31 10:09:26 +00:00

llvmir-intrinsics.mlir

[mlir][LLVM] Add support for invariant group related intrinsics (#115877)

2024-11-13 16:11:40 +08:00

llvmir-invalid.mlir

[mlir][LLVMIR] Add operand bundle support for llvm.intr.assume (#112143)

2024-10-16 20:49:02 +08:00

llvmir-le-specific.mlir

[mlir][test] XFAIL little-endian-only tests on SPARC (#103726)

2024-08-15 10:02:41 +02:00

llvmir-phi-loc.mlir

[mlir][llvmir][debug] Correctly generate location for phi nodes. (#105534)

2024-08-22 10:08:04 +01:00

llvmir-types.mlir

[mlir][vector] Update docs for scalable vectors (#101842)

2024-08-06 11:52:55 +01:00

llvmir.mlir

[mlir][LLVM] Plumb range attributes on parameters and results through (#117801)

2024-11-27 12:31:51 -06:00

loop-metadata.mlir

[MLIR][LLVM] Add distinct identifier to DICompileUnit attribute (#77070)

2024-01-08 07:42:33 +01:00

nneg.mlir

[mlir][LLVM] Add nneg flag (#115498)

2024-11-11 14:01:50 +01:00

nsw_nuw.mlir

[mlir][LLVM] Add nsw and nuw flags to trunc (#115509)

2024-11-11 17:11:44 +01:00

nvvmir-invalid.mlir

[MLIR][NVVM] Refactor tests in nvvmir.mlir (#119731)

2024-12-13 19:42:24 +05:30

nvvmir.mlir

[MLIR][NVVM] Fix the datatype error for nvvm.mma.sync when the operand is bf16 (#122664)

2025-01-13 15:03:05 +05:30

omptarget-array-sectioning-host.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-byref-bycopy-generation-device.mlir

[mlir][OpenMP] - MLIR to LLVMIR translation support for delayed privatization of allocatables in omp.target ops (#116576)

2024-12-12 14:39:58 +01:00

omptarget-byref-bycopy-generation-host.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-constant-alloca-raise.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-constant-indexing-device-region.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-data-use-dev-ordering.mlir

[Flang][OpenMP] Derived type explicit allocatable member mapping (#113557)

2024-11-16 12:28:37 +01:00

omptarget-debug2.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-debug.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-declare-target-llvm-device.mlir

[mlir][OpenMP] - MLIR to LLVMIR translation support for delayed privatization of allocatables in omp.target ops (#116576)

2024-12-12 14:39:58 +01:00

omptarget-declare-target-llvm-host.mlir

[Offload][NFC] Remove omp_ prefix from offloading entries (#88071)

2024-04-09 15:50:15 -05:00

omptarget-depend-host-only.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-depend.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-fortran-common-block-host.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-llvm.mlir

[Flang][OpenMP] Update MapInfoFinalization to use BlockArgs Interface and modify use_device_ptr/addr to be order independent (#113919)

2024-11-14 15:47:37 +01:00

omptarget-nested-ptr-record-type-mapping-host.mlir

[OpenMP][MLIR] Descriptor explicit member map lowering changes (#113556)

2024-11-16 12:26:29 +01:00

omptarget-nested-record-type-mapping-host.mlir

[OpenMP][MLIR] Descriptor explicit member map lowering changes (#113556)

2024-11-16 12:26:29 +01:00

omptarget-nowait-llvm.mlir

[flang][OpenMP] Support target ... nowait (#111823)

2024-10-15 14:39:16 +02:00

omptarget-parallel-llvm.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-parallel-wsloop.mlir

[MLIR][OpenMP] Remove terminators from loop wrappers (#112229)

2024-10-15 11:28:39 +01:00

omptarget-record-type-mapping-host.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-record-type-with-ptr-member-host.mlir

[OpenMP][MLIR] Descriptor explicit member map lowering changes (#113556)

2024-11-16 12:26:29 +01:00

omptarget-region-device-llvm.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-region-host-only.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-region-llvm.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-region-parallel-llvm.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-target-cpu-features.mlir

[Flang][MLIR][OpenMP] Use function-attached target attributes for OpenMP lowering (#78291)

2024-02-02 13:16:36 +00:00

omptarget-target-inside-task.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

omptarget-teams-llvm.mlir

[MLIR][OpenMP] Skip host omp ops when compiling for the target device (#85239)

2024-04-05 09:25:28 -04:00

omptarget-threadprivate-device-lowering.mlir

[OpenMP][MLIR] Fix threadprivate lowering when compiling for target when target operations are in use (#119310)

2025-01-03 18:01:01 +01:00

omptarget-wsloop-collapsed.mlir

[MLIR][OpenMP] Remove terminators from loop wrappers (#112229)

2024-10-15 11:28:39 +01:00

omptarget-wsloop.mlir

[MLIR][OpenMP] Remove terminators from loop wrappers (#112229)

2024-10-15 11:28:39 +01:00

omptargetdata-nowait-llvm.mlir

[flang][OpenMP] Support target enter|update|exit .. nowait (#113305)

2024-10-23 10:48:54 +02:00

omptask_if_false.mlir

[mlir][OpenMP] - Honor dependencies in code-generation of the if clause in omp.task correctly (#90891)

2024-05-13 08:54:23 -05:00

opaque-ptr.mlir

Revert "[mlir][llvm] Fix elem type passing into getelementptr (#68136)"

2023-10-05 17:38:41 -07:00

openacc-llvm.mlir

[MLIR][acc] Introduce varType to acc data clause operations (#119007)

2024-12-09 15:14:48 -08:00

openmp-data-target-device.mlir

[MLIR][OpenMP] Remove terminators from loop wrappers (#112229)

2024-10-15 11:28:39 +01:00

openmp-firstprivate.mlir

[mlir][OpenMP] rewrite conversion of privatisation for omp.parallel (#111844)

2024-10-16 14:43:57 +01:00

openmp-llvm-invalid.mlir

[flang][openacc][openmp] Support implicit casting on the atomic interface (#114390)

2024-11-05 07:52:45 -08:00

openmp-llvm.mlir

[Flang] Add translation support for MutexInOutSet and InOutSet (#120715)

2024-12-26 15:02:09 +05:30

openmp-nested.mlir

[MLIR][OpenMP] Remove terminators from loop wrappers (#112229)

2024-10-15 11:28:39 +01:00

openmp-omp.private-dealloc.mlir

[MLIR][OpenMP] Extend omp.private materialization support: dealloc (#90841)

2024-05-03 08:59:01 +02:00

openmp-parallel-reduction-cleanup.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

openmp-parallel-reduction-init.mlir

[flang][OpenMP] Fix reduction init region block management (#122079)

2025-01-09 16:11:18 +01:00

openmp-parallel-reduction-multiblock.mlir

[flang][OpenMP] Fix reduction init region block management (#122079)

2025-01-09 16:11:18 +01:00

openmp-private.mlir

[flang][OpenMP] Fix reduction init region block management (#122079)

2025-01-09 16:11:18 +01:00

openmp-reduction-array-sections.mlir

[flang][OpenMP] Fix reduction init region block management (#122079)

2025-01-09 16:11:18 +01:00

openmp-reduction-byref.mlir

[mlir][OpenMP] Convert reduction alloc region to LLVMIR (#102524)

2024-08-22 14:11:51 +01:00

openmp-reduction-call.mlir

[MLIR][OpenMP] NFC: Uniformize OpenMP ops names (#85393)

2024-03-20 11:19:38 +00:00

openmp-reduction-init-arg.mlir

[flang][OpenMP] Fix reduction init region block management (#122079)

2025-01-09 16:11:18 +01:00

openmp-reduction-sections.mlir

[flang][OpenMP] Fix reduction init region block management (#122079)

2025-01-09 16:11:18 +01:00

openmp-reduction.mlir

[MLIR][OpenMP] Prevent composite omp.simd related crashes (#113680)

2024-10-29 17:05:12 +00:00

openmp-simd-aligned.mlir

[MLIR][OpenMP] Lowering aligned clause to LLVM IR for SIMD directive (#119536)

2025-01-03 16:22:38 +05:30

openmp-simd-private.mlir

[flang][OpenMP] Extend delayed privatization for omp.simd (#122156)

2025-01-12 07:46:58 +01:00

openmp-target-multiple-private.mlir

[mlir][OpenMP] - MLIR to LLVMIR translation support for delayed privatization of allocatables in omp.target ops (#116576)

2024-12-12 14:39:58 +01:00

openmp-target-private-allocatable.mlir

[mlir][OpenMP] - MLIR to LLVMIR translation support for delayed privatization of allocatables in omp.target ops (#116576)

2024-12-12 14:39:58 +01:00

openmp-target-private.mlir

[mlir][OpenMP] - MLIR to LLVMIR translation support for delayed privatization of allocatables in omp.target ops (#116576)

2024-12-12 14:39:58 +01:00

openmp-target-use-device-nested.mlir

[mlir][OpenMP] - MLIR to LLVMIR translation support for delayed privatization of allocatables in omp.target ops (#116576)

2024-12-12 14:39:58 +01:00

openmp-task-target-device.mlir

[MLIR][OpenMP] Normalize representation of entry block arg-defining clauses (#109809)

2024-10-01 16:18:36 +01:00

openmp-teams.mlir

[OpenMPIRBuilder][MLIR] Pass target-cpu and target-features to outlined functions (#80283)

2024-02-05 12:22:56 +00:00

openmp-todo.mlir

[flang][OpenMP] Extend delayed privatization for omp.simd (#122156)

2025-01-12 07:46:58 +01:00

openmp-wsloop-private-cond_br.mlir

[reapply (#118463)][OpenMP][OMPIRBuilder] Add delayed privatization support for wsloop (#119170)

2024-12-09 14:32:04 +01:00

openmp-wsloop-private.mlir

[reapply (#118463)][OpenMP][OMPIRBuilder] Add delayed privatization support for wsloop (#119170)

2024-12-09 14:32:04 +01:00

openmp-wsloop-reduction-cleanup.mlir

[OpenMP][OMPIRBuilder] Refactor reduction initialization logic into one util (#118447)

2024-12-05 05:23:49 +01:00

rocdl.mlir

[mlir] Support ROCDL::ReadlaneOp (#116593)

2024-11-19 17:36:30 -06:00

target-cpu.mlir

[MLIR][LLVM] Add explicit target_cpu attribute to llvm.func (#78287)

2024-01-17 14:55:02 +00:00

target-ext-type.mlir

[mlir][llvm] Use zeroinitializer for TargetExtType (#66510)

2023-09-18 17:49:36 +02:00

target-features.mlir

[mlir][llvm] Add llvm.target_features features attribute (#71510)

2023-12-05 11:29:31 +00:00

test.mlir

[mlir] support dialect attribute translation to LLVM IR (#75309)

2023-12-19 14:18:16 +01:00

tune-cpu.mlir

Add basic -mtune support (#98517)

2024-07-16 16:48:24 +01:00

vcix-rv32.mlir

[MLIR][VCIX] Support VCIX intrinsics in LLVMIR dialect (#75875)

2024-02-07 15:23:28 -05:00

vcix-rv64.mlir

[MLIR][VCIX] Support VCIX intrinsics in LLVMIR dialect (#75875)

2024-02-07 15:23:28 -05:00

vector-to-llvm-ir.mlir

…

vscale.mlir

Add support for MLIR to llvm vscale attribute (#67012)

2023-09-25 14:32:18 +01:00

x86vector.mlir

…