[Mlir-commits] [mlir] [mlir][nvgpu] NVGPU Tutorials (PR #87065)

[Jacques Pienaar]

================
@@ -0,0 +1,319 @@
+# RUN: env SUPPORT_LIB=%mlir_cuda_runtime \
+# RUN:   %PYTHON %s | FileCheck %s
+
+# ===----------------------------------------------------------------------===//
+#  Chapter 4 : Multistage GEMM with Tensor Core
+# ===----------------------------------------------------------------------===//
+#
+# This program exemplifies a GEMM operation for `f32+=f16*f16`, utilizing the
+# Multistage method with a tile size of 128x128x64. The code completely
+# parallelizes the two outermost loops into thread blocks. It launches one Warp
+# Groups (128 threads in total) and allocates multiple slots/stage in the
+# shared memory. The program consists of three main parts: prologue, mainloop,
----------------
jpienaar wrote:

main loop ?

https://github.com/llvm/llvm-project/pull/87065