There are many times a constant propagation action can create additional work where none would otherwise be needed. For example xla_hlo.iota will currently get expanded to a constant of 0-N values,...

There are a bunch of invariants spelled out in the FIRRTL spec that aren't enforced by the MLIR verifier. For example, it is invalid for the dest of a firrtl.connect to be smaller than the ...

The binary shift operation shall perform: shru - shall shift their left operand to the right by the number of bit positions given by the right operand. Fill the vacated bit positions with zeros shr...

The FIRRTL dialect has things like firrtl.xorr but there are no corresponding RTL dialect ops for them. The ops should be added and lowerings implemented.

The FIRRTL dialect has things like firrtl.lt but there are no corresponding RTL dialect ops for them. The ops should be added and lowerings implemented. Note that the FIRRTL ops encode the sign of ...

The firrtl dialect does some fancy things with SSA names: it pulls the 'name' attribute into the result of the operation, and correctly round trips this through the parser. The RTL ...

A few of the reference kernels in iree/hal/vmla/op_kernels_generic.h have particularly poor performance. While we expect the LLVM ahead-of-time backend to be more viable as a deployment target, hav...

Currently, this attribute is a boolean attribute. However, when false, we emit a FIFO, and when true, a sequential buffer. For these semantics, it would make more sense to have it as a unit attribute.

It would be useful to have the same type of einsum-like spec for reductions so we can more easily create problems. Atm this is still the old hardcoded way.

Currently, any index-typed values will be emitted as 64-bit wide signals. Unsurprisingly, testing shows that this can have a huge overhead in circuits. Eventually we want to have a proper (path- an...

Summary I run the code from the tutorial https://docs.oneflow.org/master/parallelism/05_ddp.html for 通过设置 SBP 做数据并行训练, but it turns out with 'MobileNetV2' object has no attribute &a...

你好，请问怎么装载 ONNX 模型，目前只看到 Oneflow->ONNX 工具，没有找到 ONNX->Oneflow 工具。

Is your feature request related to a problem? Please describe. GitHub actions builds are terribly slow due to the initiation of fresh llvm build process on each commit. Describe the solution you&am...

The existing code assumes the result tensor type is the same as input type for a few aten ops like log, exp, erf. See torch-mlir/lib/Dialect/Torch/Transforms/RefineTypes.cpp Lines 221 to 235 in 486...

@mikeurbach and I think it would be useful to have an import/export format for scheduling problem instances, e.g. for writing test cases and benchmarking independent from the concrete (and potentia...

Is your feature request related to a problem? Please describe. As of now, pixel elements are stored in a vector having floating point 32 bit elements while lowering dip.corr_2d (Ref : buddy-mlir/li...

Contribute to DzelilaMehanovic/web-programming-2020 development by creating an account on GitHub.

In the following IR, %optional could be replaced by %none, because the op torch.aten.arange.start implements the AllowsTypeRefinement trait. We could add a canonicalization pattern that replaces al...

Is your feature request related to a problem? Please describe. When we run the engine on the Azure test bed, we get the following among others: INFO[0000] Using XXXXXXXXXXXXXXXXXXXXX as subscriptio...

When revamping the shape library, I noticed that the shape function for torch.aten.native_layer_norm was not implemented correctly for the 2nd and 3rd results. I tried fixing the buggy shape functi...