[llvm] [NVPTX] Add sparse MMA intrinsics (PR #150950)

[LLVM Continuous Integration via llvm-commits]

llvm-ci wrote:

LLVM Buildbot has detected a new failure on builder `llvm-nvptx-nvidia-win` running on `as-builder-8` while building `llvm` at step 7 "test-build-unified-tree-check-llvm".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/54/builds/12041

<details>
<summary>Here is the relevant piece of the build log for the reference</summary>

```
Step 7 (test-build-unified-tree-check-llvm) failure: test (failure)
******************** TEST 'LLVM :: CodeGen/NVPTX/wmma-ptx71-sm80.py' FAILED ********************
Exit Code: 4294967295

Command Output (stdout):
--
# RUN: at line 2
"C:\Python310\python.exe" C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\llvm-project\llvm\test\CodeGen\NVPTX\wmma-ptx71-sm80.py --ptx=71 --gpu-arch=80 > C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\test\CodeGen\NVPTX\Output\wmma-ptx71-sm80.py.tmp-ptx71-sm_80.ll
# executed command: 'C:\Python310\python.exe' 'C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\llvm-project\llvm\test\CodeGen\NVPTX\wmma-ptx71-sm80.py' --ptx=71 --gpu-arch=80
# RUN: at line 3
c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\bin\filecheck.exe C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\test\CodeGen\NVPTX\Output\wmma-ptx71-sm80.py.tmp-ptx71-sm_80.ll < C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\test\CodeGen\NVPTX\Output\wmma-ptx71-sm80.py.tmp-ptx71-sm_80.ll            --check-prefixes=INTRINSICS,M16N16,EXTGEOM,INT,SUBINT,MMA,ALTFLOAT,DOUBLE,PTX65MMA,PTX65LDMATRIX,PTX71MMA
# executed command: 'c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\bin\filecheck.exe' 'C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\test\CodeGen\NVPTX\Output\wmma-ptx71-sm80.py.tmp-ptx71-sm_80.ll' --check-prefixes=INTRINSICS,M16N16,EXTGEOM,INT,SUBINT,MMA,ALTFLOAT,DOUBLE,PTX65MMA,PTX65LDMATRIX,PTX71MMA
# RUN: at line 5
c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\bin\filecheck.exe C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\test\CodeGen\NVPTX\Output\wmma-ptx71-sm80.py.tmp-ptx71-sm_80.ll < C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\test\CodeGen\NVPTX\Output\wmma-ptx71-sm80.py.tmp-ptx71-sm_80.ll            --check-prefixes=INTRINSICS
# executed command: 'c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\bin\filecheck.exe' 'C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\test\CodeGen\NVPTX\Output\wmma-ptx71-sm80.py.tmp-ptx71-sm_80.ll' --check-prefixes=INTRINSICS
# RUN: at line 7
c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\bin\llc.exe < C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\test\CodeGen\NVPTX\Output\wmma-ptx71-sm80.py.tmp-ptx71-sm_80.ll -mtriple=nvptx64 -mcpu=sm_80 -mattr=+ptx71            | c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\bin\filecheck.exe C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\test\CodeGen\NVPTX\Output\wmma-ptx71-sm80.py.tmp-ptx71-sm_80.ll
# executed command: 'c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\bin\llc.exe' -mtriple=nvptx64 -mcpu=sm_80 -mattr=+ptx71
# executed command: 'c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\bin\filecheck.exe' 'C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\test\CodeGen\NVPTX\Output\wmma-ptx71-sm80.py.tmp-ptx71-sm_80.ll'
# RUN: at line 9
c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\bin\llc.exe < C:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\test\CodeGen\NVPTX\Output\wmma-ptx71-sm80.py.tmp-ptx71-sm_80.ll -mtriple=nvptx64 -mcpu=sm_80 -mattr=+ptx71            | c:/buildbot/latest-cuda/bin/ptxas.exe -arch=sm_60 -c - -arch=sm_80
# executed command: 'c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\bin\llc.exe' -mtriple=nvptx64 -mcpu=sm_80 -mattr=+ptx71
# executed command: c:/buildbot/latest-cuda/bin/ptxas.exe -arch=sm_60 -c - -arch=sm_80
# .---command stdout------------
# | ptxas warning : incompatible redefinition for option 'gpu-name', the last value of this option was used
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 38569; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 38609; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 38649; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 38689; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 38737; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 38785; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 38825; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 38865; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 38905; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 38945; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 38993; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39041; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39077; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39111; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39145; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39179; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39217; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39257; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39297; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39337; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39381; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39423; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39465; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39465; error   : Argument 6 of instruction 'Sparse mma': value '2' out of range, expected to be in range [0..1]
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39507; info    : Advisory: Modifier '.sp::ordered_metadata' should be used on instruction 'mma' instead of modifier '.sp' as it is expected to have substantially reduced performance on some future architectures
# | ptxas c:\buildbot\as-builder-8\llvm-nvptx-nvidia-win\build\lit-tmp-wqd1bo0y/tmpxft_000074e4_00000000-0_stdin, line 39507; error   : Argument 6 of instruction 'Sparse mma': value '3' out of range, expected to be in range [0..1]
...

```

</details>

https://github.com/llvm/llvm-project/pull/150950