facebookincubator / AITemplate

AITemplate is a Python framework which renders neural network into high performance CUDA/HIP C++ code. Specialized for FP16 TensorCore (NVIDIA GPU) and MatrixCore (AMD GPU) inference.
Apache License 2.0
4.55k stars 367 forks source link

Error to run examples on ROCm #91

Open duli2012 opened 1 year ago

duli2012 commented 1 year ago

Hello,

When I run the examples in the ROCm docker, I always get the following error: Traceback (most recent call last): File "examples/07_how_to_run_pt_model/how_to_run_pt_model.py", line 131, in verify_simple_model() File "examples/07_how_to_run_pt_model/how_to_run_pt_model.py", line 97, in verify_simple_model with compile_model( File "/usr/local/lib/python3.8/dist-packages/aitemplate/compiler/compiler.py", line 200, in compile_model compiler.transform.profile( File "/usr/local/lib/python3.8/dist-packages/aitemplate/compiler/transform/profile.py", line 79, in profile codegen.gen_profiler(sorted_graph, profiler_dir, dynamic_profiling_strategy) File "/usr/local/lib/python3.8/dist-packages/aitemplate/backend/codegen.py", line 71, in gen_profiler results.append(func.gen_profiler(workdir, dynamic_profiling_strategy)) File "/usr/local/lib/python3.8/dist-packages/aitemplate/compiler/ops/gemm_universal/gemm_common.py", line 509, in gen_profiler return func( TypeError: gemm_gen_profiler() takes 3 positional arguments but 4 were given

Could you please help to fix it?

Thanks.

Du

antinucleon commented 1 year ago

Current v0.1.1 code breaks ROCM. We are waiting AMD engineers to finish merging then tag it.

For now you can checkout v0.1 release tag to use all ROCM features.

On Tue, Nov 15, 2022 at 17:17 Du Li @.***> wrote:

Hello,

When I run the examples in the ROCm docker, I always get the following error: Traceback (most recent call last): File "examples/07_how_to_run_pt_model/how_to_run_pt_model.py", line 131, in verify_simple_model() File "examples/07_how_to_run_pt_model/how_to_run_pt_model.py", line 97, in verify_simple_model with compile_model( File "/usr/local/lib/python3.8/dist-packages/aitemplate/compiler/compiler.py", line 200, in compile_model compiler.transform.profile( File "/usr/local/lib/python3.8/dist-packages/aitemplate/compiler/transform/profile.py", line 79, in profile codegen.gen_profiler(sorted_graph, profiler_dir, dynamic_profiling_strategy) File "/usr/local/lib/python3.8/dist-packages/aitemplate/backend/codegen.py", line 71, in gen_profiler results.append(func.gen_profiler(workdir, dynamic_profiling_strategy)) File "/usr/local/lib/python3.8/dist-packages/aitemplate/compiler/ops/gemm_universal/gemm_common.py", line 509, in gen_profiler return func( TypeError: gemm_gen_profiler() takes 3 positional arguments but 4 were given

Could you please help to fix it?

Thanks.

Du

— Reply to this email directly, view it on GitHub https://github.com/facebookincubator/AITemplate/issues/91, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAJTLXRJSRP4PNUWNU264W3WIQY2JANCNFSM6AAAAAASBS6N7Y . You are receiving this because you are subscribed to this thread.Message ID: @.***>

-- Bing Xu

duli2012 commented 1 year ago

Thanks @antinucleon for the prompt response!

Du

EzioZz commented 11 months ago

I have found same problem while running example 03_bert. is this problem solved?

jiagaoxiang commented 10 months ago

Is this problem solved?

chenyang78 commented 10 months ago

Wondering if @fsx950223 could help? Thanks!

fsx950223 commented 10 months ago

Maybe there should be a link direct to rocm fork in AITemplate README.