[Issue]: Compilation error when trying different clang++ with libc++ library

[jdemirff]

debug_compilation.txt

Problem Description

Heyaa, we are encounting an interesting issue when we try to compile the AMDMiGraphX with our toolchain (clang++ & libc++). The clang++ version we are trying:

clang version 20.1.6 (https://github.com/llvm/llvm-project.git 47addd4540b4c393e478ba92bea2589e330c57fb)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /usr/bin

Before compiling we needed to change the symlink of the /usr/bin/ld to /usrc/bin/ld.bfd

Everything compiles for:

1. provided toolchain within rocm-6.4.1
2. our clang++ by not specifying libc++

The errors we are getting are particularly for some of the operators, and we are not even sure what the error is the error message is quite minimal:

In file included from /workdir/AMDMIGraphX/src/onnx/include/migraphx/onnx/op_parser.hpp:29:
In file included from /workdir/AMDMIGraphX/src/onnx/include/migraphx/onnx/onnx_parser.hpp:29:
In file included from /workdir/AMDMIGraphX/src/include/migraphx/program.hpp:29:
In file included from /workdir/AMDMIGraphX/src/include/migraphx/operation.hpp:35:
In file included from /workdir/AMDMIGraphX/src/include/migraphx/dyn_output.hpp:27:
In file included from /workdir/AMDMIGraphX/src/include/migraphx/shape.hpp:39:
In file included from /workdir/AMDMIGraphX/src/include/migraphx/serialize.hpp:28:
/workdir/AMDMIGraphX/src/include/migraphx/value.hpp:354:9: float_type, attr.floats());
      |                                               ^
 warningwarning: s'switch' missing 'default' label [-Wswitch-default]

and I see above for two more operator

Hardware related infos.

OS:
NAME="Ubuntu"
VERSION="22.04.2 LTS (Jammy Jellyfish)"

CPU: 
model name	: AMD Ryzen 7 8700G w/ Radeon 780M Graphics

GPU:
  Name:                    AMD Ryzen 7 8700G w/ Radeon 780M Graphics
  Marketing Name:          AMD Ryzen 7 8700G w/ Radeon 780M Graphics
  Name:                    gfx1103                            
  Marketing Name:                                             
      Name:                    amdgcn-amd-amdhsa--gfx1103         
      Name:                    amdgcn-amd-amdhsa--gfx11-generic 

Any help would be appreciated!

Operating System

22.04.2 LTS (Jammy Jellyfish)

CPU

AMD Ryzen 7 8700G w/ Radeon 780M Graphics

GPU

Other

Other

Radeon 780M Graphics

ROCm Version

ROCm 6.0.0

Steps to Reproduce

Use this particular clang++ version to compile:

clang version 20.1.6 (https://github.com/llvm/llvm-project.git 47addd4540b4c393e478ba92bea2589e330c57fb)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /usr/bin

(Optional for Linux users) Output of /opt/rocm/bin/rocminfo --support

ROCk module version 6.12.12 is loaded
=====================    
HSA System Attributes    
=====================    
Runtime Version:         1.15
Runtime Ext Version:     1.7
System Timestamp Freq.:  1000.000000MHz
Sig. Max Wait Duration:  18446744073709551615 (0xFFFFFFFFFFFFFFFF) (timestamp count)
Machine Model:           LARGE                              
System Endianness:       LITTLE                             
Mwaitx:                  DISABLED
XNACK enabled:           NO
DMAbuf Support:          YES
VMM Support:             YES

==========               
HSA Agents               
==========               
*******                  
Agent 1                  
*******                  
  Name:                    AMD Ryzen 7 8700G w/ Radeon 780M Graphics
  Uuid:                    CPU-XX                             
  Marketing Name:          AMD Ryzen 7 8700G w/ Radeon 780M Graphics
  Vendor Name:             CPU                                
  Feature:                 None specified                     
  Profile:                 FULL_PROFILE                       
  Float Round Mode:        NEAR                               
  Max Queue Number:        0(0x0)                             
  Queue Min Size:          0(0x0)                             
  Queue Max Size:          0(0x0)                             
  Queue Type:              MULTI                              
  Node:                    0                                  
  Device Type:             CPU                                
  Cache Info:              
    L1:                      32768(0x8000) KB                   
  Chip ID:                 0(0x0)                             
  ASIC Revision:           0(0x0)                             
  Cacheline Size:          64(0x40)                           
  Max Clock Freq. (MHz):   4200                               
  BDFID:                   0                                  
  Internal Node ID:        0                                  
  Compute Unit:            16                                 
  SIMDs per CU:            0                                  
  Shader Engines:          0                                  
  Shader Arrs. per Eng.:   0                                  
  WatchPts on Addr. Ranges:1                                  
  Memory Properties:       
  Features:                None
  Pool Info:               
    Pool 1                   
      Segment:                 GLOBAL; FLAGS: FINE GRAINED        
      Size:                    48960044(0x2eb122c) KB             
      Allocatable:             TRUE                               
      Alloc Granule:           4KB                                
      Alloc Recommended Granule:4KB                                
      Alloc Alignment:         4KB                                
      Accessible by all:       TRUE                               
    Pool 2                   
      Segment:                 GLOBAL; FLAGS: EXTENDED FINE GRAINED
      Size:                    48960044(0x2eb122c) KB             
      Allocatable:             TRUE                               
      Alloc Granule:           4KB                                
      Alloc Recommended Granule:4KB                                
      Alloc Alignment:         4KB                                
      Accessible by all:       TRUE                               
    Pool 3                   
      Segment:                 GLOBAL; FLAGS: KERNARG, FINE GRAINED
      Size:                    48960044(0x2eb122c) KB             
      Allocatable:             TRUE                               
      Alloc Granule:           4KB                                
      Alloc Recommended Granule:4KB                                
      Alloc Alignment:         4KB                                
      Accessible by all:       TRUE                               
    Pool 4                   
      Segment:                 GLOBAL; FLAGS: COARSE GRAINED      
      Size:                    48960044(0x2eb122c) KB             
      Allocatable:             TRUE                               
      Alloc Granule:           4KB                                
      Alloc Recommended Granule:4KB                                
      Alloc Alignment:         4KB                                
      Accessible by all:       TRUE                               
  ISA Info:                
*******                  
Agent 2                  
*******                  
  Name:                    gfx1103                            
  Uuid:                    GPU-XX                             
  Marketing Name:                                             
  Vendor Name:             AMD                                
  Feature:                 KERNEL_DISPATCH                    
  Profile:                 BASE_PROFILE                       
  Float Round Mode:        NEAR                               
  Max Queue Number:        128(0x80)                          
  Queue Min Size:          64(0x40)                           
  Queue Max Size:          131072(0x20000)                    
  Queue Type:              MULTI                              
  Node:                    1                                  
  Device Type:             GPU                                
  Cache Info:              
    L1:                      32(0x20) KB                        
    L2:                      2048(0x800) KB                     
  Chip ID:                 5567(0x15bf)                       
  ASIC Revision:           12(0xc)                            
  Cacheline Size:          128(0x80)                          
  Max Clock Freq. (MHz):   2900                               
  BDFID:                   28672                              
  Internal Node ID:        1                                  
  Compute Unit:            12                                 
  SIMDs per CU:            2                                  
  Shader Engines:          1                                  
  Shader Arrs. per Eng.:   2                                  
  WatchPts on Addr. Ranges:4                                  
  Coherent Host Access:    FALSE                              
  Memory Properties:       APU
  Features:                KERNEL_DISPATCH 
  Fast F16 Operation:      TRUE                               
  Wavefront Size:          32(0x20)                           
  Workgroup Max Size:      1024(0x400)                        
  Workgroup Max Size per Dimension:
    x                        1024(0x400)                        
    y                        1024(0x400)                        
    z                        1024(0x400)                        
  Max Waves Per CU:        32(0x20)                           
  Max Work-item Per CU:    1024(0x400)                        
  Grid Max Size:           4294967295(0xffffffff)             
  Grid Max Size per Dimension:
    x                        4294967295(0xffffffff)             
    y                        4294967295(0xffffffff)             
    z                        4294967295(0xffffffff)             
  Max fbarriers/Workgrp:   32                                 
  Packet Processor uCode:: 40                                 
  SDMA engine uCode::      21                                 
  IOMMU Support::          None                               
  Pool Info:               
    Pool 1                   
      Segment:                 GLOBAL; FLAGS: COARSE GRAINED      
      Size:                    24480020(0x1758914) KB             
      Allocatable:             TRUE                               
      Alloc Granule:           4KB                                
      Alloc Recommended Granule:2048KB                             
      Alloc Alignment:         4KB                                
      Accessible by all:       FALSE                              
    Pool 2                   
      Segment:                 GLOBAL; FLAGS: EXTENDED FINE GRAINED
      Size:                    24480020(0x1758914) KB             
      Allocatable:             TRUE                               
      Alloc Granule:           4KB                                
      Alloc Recommended Granule:2048KB                             
      Alloc Alignment:         4KB                                
      Accessible by all:       FALSE                              
    Pool 3                   
      Segment:                 GROUP                              
      Size:                    64(0x40) KB                        
      Allocatable:             FALSE                              
      Alloc Granule:           0KB                                
      Alloc Recommended Granule:0KB                                
      Alloc Alignment:         0KB                                
      Accessible by all:       FALSE                              
  ISA Info:                
    ISA 1                    
      Name:                    amdgcn-amd-amdhsa--gfx1103         
      Machine Models:          HSA_MACHINE_MODEL_LARGE            
      Profiles:                HSA_PROFILE_BASE                   
      Default Rounding Mode:   NEAR                               
      Default Rounding Mode:   NEAR                               
      Fast f16:                TRUE                               
      Workgroup Max Size:      1024(0x400)                        
      Workgroup Max Size per Dimension:
        x                        1024(0x400)                        
        y                        1024(0x400)                        
        z                        1024(0x400)                        
      Grid Max Size:           4294967295(0xffffffff)             
      Grid Max Size per Dimension:
        x                        4294967295(0xffffffff)             
        y                        4294967295(0xffffffff)             
        z                        4294967295(0xffffffff)             
      FBarrier Max Size:       32                                 
    ISA 2                    
      Name:                    amdgcn-amd-amdhsa--gfx11-generic   
      Machine Models:          HSA_MACHINE_MODEL_LARGE            
      Profiles:                HSA_PROFILE_BASE                   
      Default Rounding Mode:   NEAR                               
      Default Rounding Mode:   NEAR                               
      Fast f16:                TRUE                               
      Workgroup Max Size:      1024(0x400)                        
      Workgroup Max Size per Dimension:
        x                        1024(0x400)                        
        y                        1024(0x400)                        
        z                        1024(0x400)                        
      Grid Max Size:           4294967295(0xffffffff)             
      Grid Max Size per Dimension:
        x                        4294967295(0xffffffff)             
        y                        4294967295(0xffffffff)             
        z                        4294967295(0xffffffff)             
      FBarrier Max Size:       32                                 
*** Done ***             

Additional Information

If you can provide any steps to be able to debug further such error, would be great! thanks

[ppanchad-amd]

Hi @jdemirff. Internal ticket has been created to investigate this issue. Thanks!

[jdemirff]

@ppanchad-amd when I checked the error, I realised I didn't compile the dependencies with the same toolchain which is:

set(CMAKE_C_COMPILER /usr/bin/clang)
set(CMAKE_CXX_COMPILER /usr/bin/clang++)
set(CMAKE_CXX_FLAGS "-stdlib=libc++")
set(CMAKE_EXE_LINKER_FLAGS "-stdlib=libc++ -lc++abi -fuse-ld=/usr/bin/ld.bfd")

and this time I am getting the error while building the MLIR:

/workdir/AMDMIGraphX/depend/cget/build/tmp-4476c24f9c904b858fce2d8d873ac5ba/rocMLIR-88b9b7c78a1cde67c4ad5050eeeeeb47db2e4b4e/mlir/lib/Dialect/Rock/Pipelines/Pipelines.cpp:211:44: error: no viable overloaded '='
  211 |     extendToLLVMTypesOptions.extraTypeStrs = supportedFloats;
      |     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~~~~~~~
/workdir/AMDMIGraphX/depend/cget/build/tmp-4476c24f9c904b858fce2d8d873ac5ba/rocMLIR-88b9b7c78a1cde67c4ad5050eeeeeb47db2e4b4e/external/llvm-project/llvm/include/llvm/ADT/SmallVector.h:1241:16: note: candidate function not viable: no known conversion from 'SmallVector<[...], 1>' to 'const SmallVector<[...], 2>' for 1st argument
 1241 |   SmallVector &operator=(const SmallVector &RHS) {
      |                ^         ~~~~~~~~~~~~~~~~~~~~~~
/workdir/AMDMIGraphX/depend/cget/build/tmp-4476c24f9c904b858fce2d8d873ac5ba/rocMLIR-88b9b7c78a1cde67c4ad5050eeeeeb47db2e4b4e/external/llvm-project/llvm/include/llvm/ADT/SmallVector.h:1256:16: note: candidate function not viable: no known conversion from 'SmallVector<[...], 1>' to 'SmallVector<[...], 2>' for 1st argument
 1256 |   SmallVector &operator=(SmallVector &&RHS) {
      |                ^         ~~~~~~~~~~~~~~~~~
/workdir/AMDMIGraphX/depend/cget/build/tmp-4476c24f9c904b858fce2d8d873ac5ba/rocMLIR-88b9b7c78a1cde67c4ad5050eeeeeb47db2e4b4e/external/llvm-project/llvm/include/llvm/ADT/SmallVector.h:1274:16: note: candidate function not viable: expects an rvalue for 1st argument
 1274 |   SmallVector &operator=(SmallVectorImpl<T> &&RHS) {
      |                ^         ~~~~~~~~~~~~~~~~~~~~~~~~
/workdir/AMDMIGraphX/depend/cget/build/tmp-4476c24f9c904b858fce2d8d873ac5ba/rocMLIR-88b9b7c78a1cde67c4ad5050eeeeeb47db2e4b4e/external/llvm-project/llvm/include/llvm/ADT/SmallVector.h:1279:16: note: candidate function not viable: no known conversion from 'SmallVector<std::string, 1>' (aka 'SmallVector<basic_string<char>, 1>') to 'std::initializer_list<string>' for 1st argument
 1279 |   SmallVector &operator=(std::initializer_list<T> IL) {
      |                ^         ~~~~~~~~~~~~~~~~~~~~~~~~~~~
[100%] Built target obj.MLIRMIGraphXPipeline
[100%] Linking CXX static library ../../../../../lib/libMLIRMIGraphXPipeline.a
[100%] Built target MLIRMIGraphXPipeline
1 error generated.
gmake[2]: *** [mlir/lib/Dialect/Rock/Pipelines/CMakeFiles/obj.MLIRRockPipeline.dir/build.make:79: mlir/lib/Dialect/Rock/Pipelines/CMakeFiles/obj.MLIRRockPipeline.dir/Pipelines.cpp.o] Error 1
gmake[1]: *** [CMakeFiles/Makefile2:502109: mlir/lib/Dialect/Rock/Pipelines/CMakeFiles/obj.MLIRRockPipeline.dir/all] Error 2
gmake: *** [Makefile:156: all] Error 2
Command failed: ['/usr/local/bin/cmake', '--build', '/workdir/AMDMIGraphX/depend/cget/build/tmp-4476c24f9c904b858fce2d8d873ac5ba/build', '--config', 'Release', '--', '-j', '16']
Failed to build package ROCm/rocMLIR
Traceback (most recent call last):
  File "/opt/conda/envs/migraphx_env/bin/rbuild", line 8, in <module>
    sys.exit(cli())
             ~~~^^
  File "/opt/conda/envs/migraphx_env/lib/python3.13/site-packages/click/core.py", line 829, in __call__
    return self.main(*args, **kwargs)
           ~~~~~~~~~^^^^^^^^^^^^^^^^^
  File "/opt/conda/envs/migraphx_env/lib/python3.13/site-packages/click/core.py", line 782, in main
    rv = self.invoke(ctx)
  File "/opt/conda/envs/migraphx_env/lib/python3.13/site-packages/click/core.py", line 1259, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
                           ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^
  File "/opt/conda/envs/migraphx_env/lib/python3.13/site-packages/click/core.py", line 1066, in invoke
    return ctx.invoke(self.callback, **ctx.params)
           ~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/conda/envs/migraphx_env/lib/python3.13/site-packages/click/core.py", line 610, in invoke
    return callback(*args, **kwargs)
  File "/opt/conda/envs/migraphx_env/lib/python3.13/site-packages/rbuild/cli.py", line 299, in w
    f(make_builder, *args, **kwargs)
    ~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/conda/envs/migraphx_env/lib/python3.13/site-packages/rbuild/cli.py", line 312, in prepare
    b.prepare(init_with_define_flag=True)
    ~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/conda/envs/migraphx_env/lib/python3.13/site-packages/rbuild/cli.py", line 270, in prepare
    cg('install', *tokens, cwd=self.get_source_dir())
    ~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/conda/envs/migraphx_env/lib/python3.13/site-packages/rbuild/cli.py", line 225, in cget
    self.cmd(['cget', '-p', self.get_prefix()] + list(args), **kwargs)
    ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/conda/envs/migraphx_env/lib/python3.13/site-packages/rbuild/cli.py", line 222, in cmd
    subprocess.check_call(c, **kwargs)
    ~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
  File "/opt/conda/envs/migraphx_env/lib/python3.13/subprocess.py", line 419, in check_call
    raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '['cget', '-p', 'depend', 'install', '-f', 'requirements.txt']' returned non-zero exit status 1.

hopefully that will help further, the command I run to build is:

rbuild prepare -d depend -t toolchain_our_clang.cmake --cxx /usr/bin/clang++ --cc /usr/bin/clang

[tcgu-amd]

Hi @jdemirff, thanks for reaching out and working on this interesting project! Based on the error it seems to say that llvm::SmallVector is not assignment-compatible across different inline buffer sizes (the 1 and 2). This could be due to a stricter rule in compatibility enforced by libc++.

The culprit seems to be this line here https://github.com/ROCm/rocMLIR/blob/221ad7adb549ea2a39c8e15d07a7ada6e89cef88/mlir/lib/Dialect/Rock/Pipelines/Pipelines.cpp#L206

Let's change this from SmallVector<std::string, 1> supportedFloats = {"f16"}; to SmallVector<std::string, 2> supportedFloats = {"f16"}; and see what happens.

Thanks!

[jdemirff]

Hey @tcgu-amd , thanks a lot for the comment! While waiting for you, I did the following which let me successfully compile the dependencies -> https://github.com/jdemirff/rocMLIR/tree/extend-small-vector-usage
And but now I get the following errors when I try to compile the AMDMiGraphX:

[ 32%] Building CXX object src/tf/CMakeFiles/migraphx_tf.dir/parse_cast.cpp.o
grep: invalid option -- 'M'
Usage: grep [OPTION]... PATTERNS [FILE]...
Try 'grep --help' for more information.
[ 32%] Building CXX object src/targets/gpu/CMakeFiles/migraphx_device.dir/device/nonzero.cpp.o
grep: invalid option -- 'M'
Usage: grep [OPTION]... PATTERNS [FILE]...
Try 'grep --help' for more information.
[ 32%] Building CXX object src/targets/ref/CMakeFiles/migraphx_ref.dir/lowering.cpp.o
[ 32%] Building CXX object src/onnx/CMakeFiles/migraphx_onnx.dir/broadcast_qdq.cpp.o
[ 32%] Building CXX object src/tf/CMakeFiles/migraphx_tf.dir/parse_concat.cpp.o
[ 32%] Building CXX object src/targets/gpu/CMakeFiles/migraphx_device.dir/device/rnn_variable_seq_lens.cpp.o
[ 32%] Building CXX object src/targets/gpu/CMakeFiles/migraphx_device.dir/device/targets.cpp.o
[ 32%] Building CXX object src/onnx/CMakeFiles/migraphx_onnx.dir/checks.cpp.o
[ 32%] Building CXX object src/targets/gpu/CMakeFiles/migraphx_device.dir/device/topk.cpp.o
[ 32%] Building CXX object src/targets/gpu/CMakeFiles/migraphx_device.dir/device/prefix_scan_sum.cpp.o
[ 32%] Building CXX object src/targets/gpu/CMakeFiles/migraphx_device.dir/device/reverse.cpp.o
grep: invalid option -- 'M'
Usage: grep [OPTION]... PATTERNS [FILE]...
Try 'grep --help' for more information.
[ 32%] Building CXX object src/onnx/CMakeFiles/migraphx_onnx.dir/conv.cpp.o
[ 32%] Building CXX object src/onnx/CMakeFiles/migraphx_onnx.dir/map_activation_functions.cpp.o
grep: invalid option -- 'M'
Usage: grep [OPTION]... PATTERNS [FILE]...
Try 'grep --help' for more information.
grep: invalid option -- 'M'
Usage: grep [OPTION]... PATTERNS [FILE]...
Try 'grep --help' for more information.
[ 32%] Building CXX object src/tf/CMakeFiles/migraphx_tf.dir/parse_constant.cpp.o
[ 32%] Building CXX object src/onnx/CMakeFiles/migraphx_onnx.dir/onnx_parser.cpp.o
clang++: [ 32%] Building CXX object src/tf/CMakeFiles/migraphx_tf.dir/parse_expanddims.cpp.o
[ 32%] Building CXX object src/tf/CMakeFiles/migraphx_tf.dir/parse_gather.cpp.o
warning: no such include directory: '/include' [-Wmissing-include-dirs]
clang++: warning: no such include directory: '/include' [-Wmissing-include-dirs]
clang++: error: no input files
clang++: error: clang++: warning: no such include directory: '/include' [-Wmissing-include-dirs]
grep: invalid option -- 'M'
Usage: grep [OPTION]... PATTERNS [FILE]...
Try 'grep --help' for more information.
[ 32%] Building CXX object src/onnx/CMakeFiles/migraphx_onnx.dir/op_parser.cpp.o
clang++: error: no input files
grep: invalid option -- 'M'
Usage: grep [OPTION]... PATTERNS [FILE]...
Try 'grep --help' for more information.
clang++: warning: no such include directory: '/include' [-Wmissing-include-dirs]
clang++: error: no input files
[ 32%] Building CXX object src/onnx/CMakeFiles/migraphx_onnx.dir/onnx.cpp.o
grep: invalid option -- 'M'
Usage: grep [OPTION]... PATTERNS [FILE]...
Try 'grep --help' for more information.
no input files
make[2]: *** [src/targets/gpu/CMakeFiles/migraphx_device.dir/build.make:79: src/targets/gpu/CMakeFiles/migraphx_device.dir/device/argmax.cpp.o] Error 2
make[2]: *** Waiting for unfinished jobs....
make[2]: *** [src/targets/gpu/CMakeFiles/migraphx_device.dir/build.make:149: src/targets/gpu/CMakeFiles/migraphx_device.dir/device/multinomial.cpp.o] Error 2
make[2]: *** [src/targets/gpu/CMakeFiles/migraphx_device.dir/build.make:107: src/targets/gpu/CMakeFiles/migraphx_device.dir/device/contiguous.cpp.o] Error 2
make[2]: *** [src/targets/gpu/CMakeFiles/migraphx_device.dir/build.make:121: src/targets/gpu/CMakeFiles/migraphx_device.dir/device/fill.cpp.o] Error 2
grep: invalid option -- 'M'
Usage: grep [OPTION]... PATTERNS [FILE]...
Try 'grep --help' for more information.

not sure who is running grep and how 🤔 Any ideas?

[jdemirff]

Aaah I just realised could be the GPU_TARGETS command I was using: DGPU_TARGETS=$(/opt/rocm/bin/rocminfo | grep -o -m1 'gfx.*') now trying by specifying that to gfx1103

[jdemirff]

So as a next step of problems, I got the following error:

/opt/rocm/include/hip/amd_detail/amd_math_functions.h:47:8: error: no template named '__numeric_type'; did you mean '__hip::__numeric_type'?
   47 | struct __numeric_type<_Float16>
      |        ^
)

After reading that ROCm/clr#100 carefully, I commented in the mentioned lines as here (ROCm/clr@b8ba4cc) and carried on building the application.

The next error was the following:

/workdir/AMDMIGraphX/src/include/migraphx/verify.hpp:103:31: error: call to 'fabs' is ambiguous
  103 |         return std::fabs(x) < std::fabs(y);
      |                               ^~~~~~~~~

which I could easily fix by changing the line to:

return std::fabs(static_cast<double>(x)) < std::fabs(static_cast<double>(y));

than when I attempted to compile again got the following:

/usr/bin/ld.bfd: ../../../../lib/libmigraphx_gpu.so.2012000.0: undefined reference to `hipblaslt_ext::getAllAlgos(void*, hipblaslt_ext::GemmType, hipblasOperation_t, hipblasOperation_t, hipDataType, hipDataType, hipDataType, hipDataType, hipblasComputeType_t, std::__1::vector<_hipblasLtMatmulHeuristicResult_t, std::__1::allocator<_hipblasLtMatmulHeuristicResult_t> >&)'
/usr/bin/ld.bfd: ../../../../lib/libmigraphx_gpu.so.2012000.0: undefined reference to `hipblaslt_ext::getAlgosFromIndex(void*, std::__1::vector<int, std::__1::allocator<int> >&, std::__1::vector<_hipblasLtMatmulHeuristicResult_t, std::__1::allocator<_hipblasLtMatmulHeuristicResult_t> >&)'
clang++: error: linker command failed with exit code 1 (use -v to see invocation)
In file included from /workdir/AMDMIGraphX/test/ref/max.cpp:24:
In file included from /workdir/AMDMIGraphX/src/include/migraphx/instruction.hpp:27:
In file included from /workdir/AMDMIGraphX/src/include/migraphx/literal.hpp:27:
/workdir/AMDMIGraphX/src/include/migraphx/shape.hpp:391:9: warning: 'switch' missing 'default' label [-Wswitch-default]

which I assume is due to the hipBlas being compiled via stdlibc++ libraries rather than libc++. So what do you recommend for the next steps?

Do I need to compile hipBlas from source via libc++ too or is there any recommended workaround?

Cheers

[tcgu-amd]

Hi @jdemirff, nice work getting all these workarounds! I think it would be the best to compile all dependencies with the same libc++ library. Mixing libc++ and libstdc++ in general will likely not work.

By the way, github issues are typically used to track defects/bugs, and since this issue is not really a bug in MIGraphics, I am going to close it for now just so our bug tracker do not flag this. But don't worry we can still continue the discussion and I will continue to help as much as I can :)