tsupei commented on pull request #7111: URL: https://github.com/apache/tvm/pull/7111#issuecomment-747280452
@insop I ran the testcase in my local machine. However, some placeholders are not bound in dynamic cases. On the other hand, we cannot directly reuse the schedule defined in `_batch_matmul_implement` on [line 28](https://github.com/apache/tvm/blob/main/tests/python/topi/python/test_topi_batch_matmul.py), which may incurs some errors related to `Var` type. Maybe we need to define our own schedule for that dynamic testcase? ```bash tvm._ffi.base.TVMError: Traceback (most recent call last): [bt] (5) /home/jojo6174/tvm-installation/tvm/build/libtvm.so(TVMFuncCall+0x65) [0x7f05198a6765] [bt] (4) /home/jojo6174/tvm-installation/tvm/build/libtvm.so(+0xd19862) [0x7f0518d52862] [bt] (3) /home/jojo6174/tvm-installation/tvm/build/libtvm.so(tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const+0x2f0) [0x7f0518d51800] [bt] (2) /home/jojo6174/tvm-installation/tvm/build/libtvm.so(tvm::transform::ModulePassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const+0x1b7) [0x7f0518d52117] [bt] (1) /home/jojo6174/tvm-installation/tvm/build/libtvm.so(+0xf4aaac) [0x7f0518f83aac] [bt] (0) /home/jojo6174/tvm-installation/tvm/build/libtvm.so(+0xf479e2) [0x7f0518f809e2] Did you forget to bind? Variable `y` is directly accessed by host memory (it is not contained in a thread environment or in the function arguments. Variable `x` is directly accessed by host memory (it is not contained in a thread environment or in the function arguments. Variable `compute` is directly accessed by host memory (it is not contained in a thread environment or in the function arguments. Variable `compute` is directly accessed by host memory (it is not contained in a thread environment or in the function arguments. Variable `compute` is directly accessed by host memory (it is not contained in a thread environment or in the function arguments. File "/home/jojo6174/tvm-installation/tvm/src/tir/analysis/verify_memory.cc", line 202 RuntimeError: Memory verification failed with the following errors: PrimFunc([x, y, compute]) attrs={"global_symbol": "dense", "tir.noalias": (bool)1, "target": nvptx -keys=cuda,gpu -max_num_threads=1024 -mcpu=sm_61 -mtriple=nvptx64-nvidia-cuda -thread_warp_size=32} { for (b, 0, dynamic_batch_size) { for (i, 0, dynamic_M) { for (j, 0, dynamic_N) { compute[(((b*stride) + (i*stride)) + (j*stride))] = 0f for (k, 0, dynamic_K) { compute[(((b*stride) + (i*stride)) + (j*stride))] = (compute[(((b*stride) + (i*stride)) + (j*stride))] + (x[(((b*stride) + (i*stride)) + (k*stride))]*y[(((b*stride) + (j*stride)) + (k*stride))])) } } } } } ``` ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
