tvm
tvm copied to clipboard
apache
Open deep learning compiler stack for cpu, gpu and specialized accelerators
### This issue is to track progress for [Issue Tracking RFC](https://github.com/apache/tvm-rfcs/blob/main/rfcs/0093_Issue_Triage.md). This issue has these tasks: - [ ] Create labels described in the RFC. - [ ] Modify ISSUE_TEMPLATES...
There was a flaw in UMA backend.py that triggered an error when for passing target attributes. It is fixed. A link in the UMA tutorial was broken and fixed by...
This is a draft of implementation of 'set_output_zero_copy' method on VirtualMachine side. Brief description of approach. 1. There is python API function 'set_output' which save external outputs in VM outputs_...
Thanks for contributing to TVM! Please refer to guideline https://tvm.apache.org/docs/contribute/ for useful information and tips. After the pull request is submitted, please request code reviews from [Reviewers](https://github.com/apache/incubator-tvm/blob/master/CONTRIBUTORS.md#reviewers) by @ them...
This implements an initial Target Parser which aims to consolidate architecture feature detection from a few different places: * https://github.com/apache/tvm/blob/d2db9cb0d839e32778f461b77e59f6418282a511/python/tvm/topi/arm_cpu/arm_utils.py#L24-L70 * https://github.com/apache/tvm/blob/02fbaf0ed9120a8f95155e63de42459f230584aa/python/tvm/relay/qnn/op/legalizations.py#L350-L359 * https://github.com/apache/tvm/blob/b542724873140bb051492530d97a78b9b7b7983d/python/tvm/relay/op/strategy/arm_cpu.py#L232 A further patch will remove all...
When we invoke ``` relay_mod = relay.transform.ToMixedPrecision('bfloat16')(relay_mod) ``` pass, we are supposed to get a well-transformed model which contains different kinds of operators with auto mix precision. However, we have...
As mentioned in the title, the per store feature extractor of meta schedule seems not be able to handle TIR with one simply block of storing constant tensor. ### Actual...
It's difficult to know whether a PR will affect CI runtime without doing a purposeful investigation or hitting timeouts. Similar to the `tests_bot` (https://github.com/apache/tvm/blob/main/.github/workflows/tests_bot.yml) we could have a bot that...
In the current state of TVM, when we run hexagon tests on HDK boards and a test fails the session is not properly closed even after [deleting the python object](https://github.com/apache/tvm/blob/main/python/tvm/contrib/hexagon/session.py#L98)....
See https://ci.tlcpack.ai/blue/organizations/jenkins/tvm/detail/aarch64_frontend_tests/3/pipeline ``` [2022-09-10T11:36:45.046Z] tests/python/frontend/pytorch/qnn_test.py::test_serialized_modules LLVM ERROR: out of memory [2022-09-10T11:36:45.046Z] Fatal Python error: Aborted [2022-09-10T11:36:45.046Z] [2022-09-10T11:36:45.046Z] Thread 0x0000ffffa6c241f0 (most recent call first): [2022-09-10T11:36:45.046Z] File "/usr/local/lib/python3.7/dist-packages/execnet/gateway_base.py", line 400 in read...