From fc71284bcb131a3b1427e33d6e7f3037ffbca0ec Mon Sep 17 00:00:00 2001
From: hanishkvc <hanishkvc@gmail.com>
Date: Sat, 27 Jun 2020 13:03:55 +0530
Subject: [PATCH 15/15] Added a changelog entry, comment for intely logic

---
 Changelog                 |  1 +
 libavfilter/vf_fbdetile.c | 15 +++++++++++++++
 2 files changed, 16 insertions(+)

diff --git a/Changelog b/Changelog
index a60e7d2eb8..0e03491f6a 100644
--- a/Changelog
+++ b/Changelog
@@ -2,6 +2,7 @@ Entries are sorted chronologically from oldest to youngest within each release,
 releases are sorted from youngest to oldest.
 
 version <next>:
+- fbdetile cpu based framebuffer layout detiling video filter
 - AudioToolbox output device
 - MacCaption demuxer
 
diff --git a/libavfilter/vf_fbdetile.c b/libavfilter/vf_fbdetile.c
index 8ac054e944..5884d0a8f4 100644
--- a/libavfilter/vf_fbdetile.c
+++ b/libavfilter/vf_fbdetile.c
@@ -139,6 +139,21 @@ static void detile_intelx(AVFilterContext *ctx, int w, int h,
     }
 }
 
+/*
+ * Intel Legacy Tile-Y layout conversion support
+ *
+ * currently done in a simple dumb way. Two low hanging optimisations
+ * that could be readily applied are
+ *
+ * a) unrolling the inner for loop
+ *
+ * b) using simd based 128bit loading and storing along with prefetch
+ *    hinting. TOTHINK|CHECK: Does memcpy already does this if situation
+ *    is right?!
+ *
+ * Or I could even merge the two intel detiling logics into one, as
+ * the semantic and flow is the same for both logics.
+ */
 static void detile_intely(AVFilterContext *ctx, int w, int h,
                                 uint8_t *dst, int dstLineSize,
                           const uint8_t *src, int srcLineSize)
-- 
2.20.1

