uxa: solid rects

The cost of performing relocations outweigh the advantages of using the
blitter for solids with lots of rectangles.

References:

  Bug 22127 - [UXA] 50% performance regression for XRenderFillRectangles
  https://bugs.freedesktop.org/show_bug.cgi?id=22127

By using the 3D pipeline we improve our performance by around 4x on
i945, measured by the jxbench microbenchmark, and a factor of 10x by
short-cutting to the 3D pipeline for blended rectangles.

Before, on a i945GME:
  19982.412060 Ops/s; rects (!); 15x15
  9599.131693 Ops/s; rects (!); 75x75
  3803.654743 Ops/s; rects (!); 250x250
  6836.743772 Ops/s; rects blended; 15x15
  1443.750000 Ops/s; rects blended; 75x75
  495.335821 Ops/s; rects blended; 250x250
  23247.933884 Ops/s; rects composition (!); 15x15
  10993.073048 Ops/s; rects composition (!); 75x75
  3595.905172 Ops/s; rects composition (!); 250x250

After:
  87271.145975 Ops/s; rects (!); 15x15
  32347.744361 Ops/s; rects (!); 75x75
  5884.177215 Ops/s; rects (!); 250x250
  73500.000000 Ops/s; rects blended; 15x15
  33580.882353 Ops/s; rects blended; 75x75
  5858.811749 Ops/s; rects blended; 250x250
  25582.317073 Ops/s; rects composition (!); 15x15
  6664.728682 Ops/s; rects composition (!); 75x75
  14965.909091 Ops/s; rects composition (!); 250x250 [suspicious]

This has no impact on Cairo, but I have a suspicion from watching xtrace
that Qt likes to blit thousands of 1x1 rectangles with the same colour.
However, we are still around 2-3x slower than the reported figures for
EXA!

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

uxa/uxa-priv.h

@@ -156,6 +156,7 @@ typedef struct {
 	BitmapToRegionProcPtr SavedBitmapToRegion;
 #ifdef RENDER
 	CompositeProcPtr SavedComposite;
+	CompositeRectsProcPtr SavedCompositeRects;
 	TrianglesProcPtr SavedTriangles;
 	GlyphsProcPtr SavedGlyphs;
 	TrapezoidsProcPtr SavedTrapezoids;
@@ -416,6 +417,13 @@ uxa_composite_rects(CARD8 op,
 		    PicturePtr pSrc,
 		    PicturePtr pDst, int nrect, uxa_composite_rect_t * rects);
 
+void
+uxa_solid_rects (CARD8		op,
+		 PicturePtr	dst,
+		 xRenderColor  *color,
+		 int		num_rects,
+		 xRectangle    *rects);
+
 void
 uxa_trapezoids(CARD8 op, PicturePtr pSrc, PicturePtr pDst,
 	       PictFormatPtr maskFormat, INT16 xSrc, INT16 ySrc,

uxa/uxa-render.c

@@ -828,6 +828,75 @@ uxa_acquire_mask(ScreenPtr screen,
 				    out_x, out_y);
 }
 
+void
+uxa_solid_rects (CARD8		op,
+		 PicturePtr	dst,
+		 xRenderColor  *color,
+		 int		num_rects,
+		 xRectangle    *rects)
+{
+	ScreenPtr screen = dst->pDrawable->pScreen;
+	uxa_screen_t *uxa_screen = uxa_get_screen(screen);
+	PixmapPtr pixmap;
+	int dst_x, dst_y;
+	PicturePtr src;
+	int error;
+
+	/* Using GEM, the relocation costs outweigh the advantages of the blitter */
+	if (num_rects == 1)
+		goto fallback;
+
+	if (dst->alphaMap)
+		goto fallback;
+
+	if (!uxa_screen->info->check_composite_texture)
+		goto fallback;
+
+	pixmap = uxa_get_offscreen_pixmap(dst->pDrawable, &dst_x, &dst_y);
+	if (!pixmap)
+		goto fallback;
+
+	if (op == PictOpClear)
+		color->red = color->green = color->blue = color->alpha = 0;
+	if (PICT_FORMAT_A(dst->format) == 0)
+		color->alpha = 0xffff;
+	if (color->alpha >= 0xff00 && op == PictOpOver)
+		op = PictOpSrc;
+
+	src = CreateSolidPicture(0, color, &error);
+	if (!src)
+		goto fallback;
+
+	if (!uxa_screen->info->check_composite(op, src, NULL, dst) ||
+	    !uxa_screen->info->check_composite_texture(screen, src)) {
+		FreePicture(src, 0);
+		goto fallback;
+	}
+
+	if (!uxa_screen->info->prepare_composite(op, src, NULL, dst, NULL, NULL, pixmap)) {
+		FreePicture(src, 0);
+		goto fallback;
+	}
+
+	while (num_rects--) {
+		uxa_screen->info->composite(pixmap,
+					    0, 0, 0, 0,
+					    rects->x + dst_x,
+					    rects->y + dst_y,
+					    rects->width,
+					    rects->height);
+		rects++;
+	}
+
+	uxa_screen->info->done_composite(pixmap);
+	FreePicture(src, 0);
+
+	return;
+
+fallback:
+	uxa_screen->SavedCompositeRects(op, dst, color, num_rects, rects);
+}
+
 static int
 uxa_try_driver_composite_rects(CARD8 op,
 			       PicturePtr pSrc,

uxa/uxa.c

@@ -388,6 +388,7 @@ static Bool uxa_close_screen(int i, ScreenPtr pScreen)
 #ifdef RENDER
 	if (ps) {
 		ps->Composite = uxa_screen->SavedComposite;
+		ps->CompositeRects = uxa_screen->SavedCompositeRects;
 		ps->Glyphs = uxa_screen->SavedGlyphs;
 		ps->Trapezoids = uxa_screen->SavedTrapezoids;
 		ps->AddTraps = uxa_screen->SavedAddTraps;
@@ -517,6 +518,9 @@ Bool uxa_driver_init(ScreenPtr screen, uxa_driver_t * uxa_driver)
 		uxa_screen->SavedComposite = ps->Composite;
 		ps->Composite = uxa_composite;
 
+		uxa_screen->SavedCompositeRects = ps->CompositeRects;
+		ps->CompositeRects = uxa_solid_rects;
+
 		uxa_screen->SavedGlyphs = ps->Glyphs;
 		ps->Glyphs = uxa_glyphs;