From 3a238de99814ae05094d92b7ac62be70f6aba19c Mon Sep 17 00:00:00 2001
From: Martin Pulec <martin.pulec@cesnet.cz>
Date: Wed, 18 Sep 2024 12:11:42 +0200
Subject: [PATCH] vdisp/sdl: R10k conversion optimized

Create inner loop with fixed amount of iterations (16). This will allow
the compiler to unroll the inner loop and vectorize (16 iterations per
4 bytes is 512b allowing up to 512b instructions).

The eventual rest (%16 != 0) is computed per pixel as it used to be..
---
 src/video_display/sdl2.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/src/video_display/sdl2.c b/src/video_display/sdl2.c
index 517caebd1..4c4dd7f01 100644
--- a/src/video_display/sdl2.c
+++ b/src/video_display/sdl2.c
@@ -833,7 +833,17 @@ static struct video_frame *display_sdl2_getf(void *state)
 static void
 r10k_to_sdl2(size_t count, uint32_t *buf)
 {
+        enum {
+                LOOP_ITEMS = 16,
+        };
         unsigned int i = 0;
+        for (; i < count / LOOP_ITEMS; ++i) {
+                for (int j = 0; j < LOOP_ITEMS; ++j) {
+                        uint32_t val = htonl(*buf);
+                        *buf++       = val >> 2;
+                }
+        }
+        i *= LOOP_ITEMS;
         for (; i < count; ++i) {
                 uint32_t val = htonl(*buf);
                 *buf++       = val >> 2;