lorenzo chelini

We need to disable opaque pointers, and we are good to go here. (Currently looking into how to disable them)

[WIP][DRAFT] Pack operation

7

Start revisiting the `linalgx.relayout` operation in lights of the discussions we had. The semantics I would like to have is: `tensor.expand_shape + tensor.transpose` TODO: 1. Dropping the affine map and...

Building the example (clang/test/Frontend/plugin-syntax-print-tokens.cpp)

4

Hi, I am trying to use the syntax plugin but I am hitting the following error: ``` [2882/2883] Running lit suite /home/lchelini/scratch/llvm-project-csp-orig/clang/test/Frontend llvm-lit: /home/lchelini/scratch/llvm-project-csp-orig/llvm/utils/lit/lit/llvm/config.py:342: note: using clang: /home/lchelini/scratch/llvm-project-csp-orig/build/bin/clang FAIL: Clang...

Handle `linalg.max`

1

Enable linalg to XSMM lowering for `linalg.max`.

When re-writing a batch matmul to matmul we tile fully along the batch dimension, however when the tensors are fully dynamic, the scf.forall parallelization introduces an affine min map that...

Enable running softmax with TPPs

3

To enable running softmax with TPPs we need more operations: 1. max/sum reduce op (%2 and %8) 2. sub operation (%5) We needs to support broadcast semantics in sub or...