diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml index f7faa1706f..bd68558bd4 100644 --- a/.github/workflows/test.yml +++ b/.github/workflows/test.yml @@ -350,7 +350,7 @@ jobs: - name: Check Device.DEFAULT and print some source run: | python -c "from tinygrad.ops import Device; assert Device.DEFAULT in ['LLVM','CLANG','CUDA','GPU'], Device.DEFAULT" - DEBUG=4 FORWARD_ONLY=1 python3 test/test_ops.py TestOps.test_add + DEBUG=5 PYTHONPATH=${{ github.workspace }} FORWARD_ONLY=1 python3 test/test_ops.py TestOps.test_add - name: Run pytest (not cuda) if: matrix.backend!='cuda' && matrix.backend!='ptx' && matrix.backend!='triton' run: python -m pytest -n=auto test/ -k '${{matrix.backend=='llvm'&&'not (test_nn.py and test_conv_transpose2d)'||'test'}}' -m 'not exclude_${{matrix.backend}}' --durations=20 diff --git a/tinygrad/runtime/ops_gpu.py b/tinygrad/runtime/ops_gpu.py index 13cc2853fb..26f790da4d 100644 --- a/tinygrad/runtime/ops_gpu.py +++ b/tinygrad/runtime/ops_gpu.py @@ -81,8 +81,8 @@ class CLProgram: elif CL.cl_ctxs[0].devices[0].name.startswith('gfx'): asm = early_exec(([ROCM_LLVM_PATH / "llvm-objdump", '-d', '-'], prg)) print('\n'.join([x for x in asm.decode('utf-8').split("\n") if 's_code_end' not in x])) - else: - # print the PTX for NVIDIA. TODO: probably broken for everything else + elif "NVIDIA" in CL.cl_ctxs[0].devices[0].name: + # print the PTX for NVIDIA. print(prg.decode('utf-8')) if argdtypes is not None: self.set_argdtypes(argdtypes)