[AMDGPU] Add file headers for few files where it is missing.

Summary:
Added file headers for files which implement iterative lightweight scheduling
strategies. Which is basically an exercise which I undertook in order to get
used to LLVM development process.

Reviewers: arsenm, vpykhtin, cdevadas

Reviewed By: vpykhtin

Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, javed.absar, kerbowa, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D73417

llvm/lib/Target/AMDGPU/GCNIterativeScheduler.cpp

@@ -5,6 +5,11 @@
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
+///
+/// \file
+/// This file implements the class GCNIterativeScheduler.
+///
+//===----------------------------------------------------------------------===//
 
 #include "GCNIterativeScheduler.h"
 #include "AMDGPUSubtarget.h"

llvm/lib/Target/AMDGPU/GCNIterativeScheduler.h

@@ -5,6 +5,14 @@
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
+///
+/// \file
+/// This file defines the class GCNIterativeScheduler, which uses an iterative
+/// approach to find a best schedule for GCN architecture. It basically makes
+/// use of various lightweight schedules, scores them, chooses best one based on
+/// their scores, and finally implements the chosen one.
+///
+//===----------------------------------------------------------------------===//
 
 #ifndef LLVM_LIB_TARGET_AMDGPU_GCNITERATIVESCHEDULER_H
 #define LLVM_LIB_TARGET_AMDGPU_GCNITERATIVESCHEDULER_H

llvm/lib/Target/AMDGPU/GCNMinRegStrategy.cpp

@@ -5,6 +5,13 @@
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
+///
+/// \file
+/// This file defines and imlements the class GCNMinRegScheduler, which
+/// implements an experimental, simple scheduler whose main goal is to learn
+/// ways about consuming less possible registers for a region.
+///
+//===----------------------------------------------------------------------===//
 
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/SmallPtrSet.h"

llvm/lib/Target/AMDGPU/GCNRegPressure.cpp

@@ -5,6 +5,11 @@
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
+///
+/// \file
+/// This file implements the GCNRegPressure class.
+///
+//===----------------------------------------------------------------------===//
 
 #include "GCNRegPressure.h"
 #include "AMDGPUSubtarget.h"

llvm/lib/Target/AMDGPU/GCNRegPressure.h

@@ -5,6 +5,14 @@
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
+///
+/// \file
+/// This file defines the GCNRegPressure class, which tracks registry pressure
+/// by bookkeeping number of SGPR/VGPRs used, weights for large SGPR/VGPRs. It
+/// also implements a compare function, which compares different register
+/// pressures, and declares one with max occupance as winner.
+///
+//===----------------------------------------------------------------------===//
 
 #ifndef LLVM_LIB_TARGET_AMDGPU_GCNREGPRESSURE_H
 #define LLVM_LIB_TARGET_AMDGPU_GCNREGPRESSURE_H