tvm icon indicating copy to clipboard operation
tvm copied to clipboard

Open deep learning compiler stack for cpu, gpu and specialized accelerators

Results 636 tvm issues
Sort by recently updated
recently updated
newest added
trafficstars

The ArmChina team recently synchronized the community's new code and found that python's unittest was split from one folder into multiple folders and traversed multiple folders to run in the...

needs-triage
type:ci

Fix issue https://github.com/apache/tvm/issues/16510 Please help to review this patch, thanks @Hzfengsy @Johnson9009

### Expected behavior I try to use MetaScheduler to tuning matmul, and the dimensions of the matrix are m=8192, n=14336, k=8192. When n=8192, everything is ok, but once m or...

type: bug
needs-triage

### Expected behavior tir.PrimFunc.specialize() works as expected. ### Actual behavior ``` # from tvm.script import tir as T @T.prim_func def main(A: T.Buffer((4096, 4096), "float16"), B: T.Buffer((4096, 4096), "float16"), C: T.Buffer((4096,...

type: bug
needs-triage

Currently, graph executor loads params directly into data entries while those should be only 'views' of data from storage created with `TVMNDArray_CreateView`. When graph executor is being released, the data...

A few issues were found in building TVM 0.13.0 by using LLVM 18 in Ubuntu22.04, and the way presented in cmake/config.cmake file was tried to build the source code. llvm::CGFT_ObjectFile...

type: bug
needs-triage

Quantification is usually divided into two modes: per-channel or per-tensor. For per-channel zero_point and scale are 1d array and it's length same as tensor channel. For per-tensor zero_point and scale...

### Expected behavior tvm.error.InternalError: Traceback (most recent call last): File "/tvm/src/relay/backend/token_allocator.cc", line 41 InternalError: Check failed: (pval != nullptr) is false: Cannot allocate memory symbolic tensor shape [T.Any(), T.Any()] ###...

type: bug
needs-triage

Hello Everyone, I assume that this automated Checksum calculation might be helpful for somebody to detect Soft Faults during the calculation of a Conv2D operation. So I combined the knowledge...

Earlier implementation in Docker was download and install tvm using CUDA flags on: ``` # install tvm RUN git clone --recursive https://github.com/apache/incubator-tvm tvm && \ cd tvm && \ git...