diff --git a/src/radeon_commonfuncs.c b/src/radeon_commonfuncs.c
index 92e10d2..7f204b7 100644
--- a/src/radeon_commonfuncs.c
+++ b/src/radeon_commonfuncs.c
@@ -548,6 +548,13 @@ static void FUNC_NAME(RADEONInit3DEngine)(ScrnInfoPtr pScrn)
 	OUT_ACCEL_REG(R300_RB3D_DSTCACHE_CTLSTAT, R300_DC_FLUSH_3D | R300_DC_FREE_3D);
 	FINISH_ACCEL();
 
+	BEGIN_ACCEL(2);
+	OUT_ACCEL_REG(R300_SC_SCISSOR0, ((0 << R300_SCISSOR_X_SHIFT) |
+					 (0 << R300_SCISSOR_Y_SHIFT)));
+	OUT_ACCEL_REG(R300_SC_SCISSOR1, ((8191 << R300_SCISSOR_X_SHIFT) |
+					 (8191 << R300_SCISSOR_Y_SHIFT)));
+	FINISH_ACCEL();
+
 	BEGIN_ACCEL(5);
 	OUT_ACCEL_REG(R300_SC_EDGERULE, 0xA5294A5);
 	if (IS_R300_3D) {
diff --git a/src/radeon_exa_render.c b/src/radeon_exa_render.c
index b5f0197..af0f715 100644
--- a/src/radeon_exa_render.c
+++ b/src/radeon_exa_render.c
@@ -1860,14 +1860,6 @@ static Bool FUNC_NAME(R300PrepareComposite)(int op, PicturePtr pSrcPicture,
 	FINISH_ACCEL();
     }
 
-    /* Clear out scissoring */
-    BEGIN_ACCEL(2);
-    OUT_ACCEL_REG(R300_SC_SCISSOR0, ((0 << R300_SCISSOR_X_SHIFT) |
-				     (0 << R300_SCISSOR_Y_SHIFT)));
-    OUT_ACCEL_REG(R300_SC_SCISSOR1, ((8191 << R300_SCISSOR_X_SHIFT) |
-				     (8191 << R300_SCISSOR_Y_SHIFT)));
-    FINISH_ACCEL();
-
     BEGIN_ACCEL(3);
 
     OUT_ACCEL_REG(R300_RB3D_COLOROFFSET0, dst_offset);
diff --git a/src/radeon_textured_videofuncs.c b/src/radeon_textured_videofuncs.c
index e324fd9..abde4a5 100644
--- a/src/radeon_textured_videofuncs.c
+++ b/src/radeon_textured_videofuncs.c
@@ -1475,44 +1475,6 @@ FUNC_NAME(RADEONDisplayTexturedVideo)(ScrnInfoPtr pScrn, RADEONPortPrivPtr pPriv
 	}
     }
 
-    /*
-     * Rendering of the actual polygon is done in two different
-     * ways depending on chip generation:
-     *
-     * < R300:
-     *
-     *     These chips can render a rectangle in one pass, so
-     *     handling is pretty straight-forward.
-     *
-     * >= R300:
-     *
-     *     These chips can accept a quad, but will render it as
-     *     two triangles which results in a diagonal tear. Instead
-     *     We render a single, large triangle and use the scissor
-     *     functionality to restrict it to the desired rectangle.
-     */
-
-    if (IS_R300_3D || IS_R500_3D) {
-	/*
-	 * Set up the scissor area to that of the output size.
-	 */
-
-	BEGIN_ACCEL(2);
-	if (IS_R300_3D) {
-	    /* R300 has an offset */
-	    OUT_ACCEL_REG(R300_SC_SCISSOR0, (((pPriv->drw_x + 1088) << R300_SCISSOR_X_SHIFT) |
-					     ((pPriv->drw_y + 1088) << R300_SCISSOR_Y_SHIFT)));
-	    OUT_ACCEL_REG(R300_SC_SCISSOR1, (((pPriv->drw_x + pPriv->dst_w + 1088 - 1) << R300_SCISSOR_X_SHIFT) |
-					     ((pPriv->drw_y + pPriv->dst_h + 1088 - 1) << R300_SCISSOR_Y_SHIFT)));
-	} else {
-	    OUT_ACCEL_REG(R300_SC_SCISSOR0, (((pPriv->drw_x) << R300_SCISSOR_X_SHIFT) |
-					     ((pPriv->drw_y) << R300_SCISSOR_Y_SHIFT)));
-	    OUT_ACCEL_REG(R300_SC_SCISSOR1, (((pPriv->drw_x + pPriv->dst_w - 1) << R300_SCISSOR_X_SHIFT) |
-					     ((pPriv->drw_y + pPriv->dst_h - 1) << R300_SCISSOR_Y_SHIFT)));
-	}
-	FINISH_ACCEL();
-    }
-
     FUNC_NAME(RADEONWaitForVLine)(pScrn, pPixmap,
 				  radeon_covering_crtc_num(pScrn,
 							   pPriv->drw_x,
@@ -1558,12 +1520,12 @@ FUNC_NAME(RADEONDisplayTexturedVideo)(ScrnInfoPtr pScrn, RADEONPortPrivPtr pPriv
 		     RADEON_CP_VC_CNTL_VTX_FMT_RADEON_MODE |
 		     (3 << RADEON_CP_VC_CNTL_NUM_SHIFT));
 	} else if (IS_R300_3D || IS_R500_3D) {
-	    BEGIN_RING(3 * vtx_count + 4);
+	    BEGIN_RING(4 * vtx_count + 4);
 	    OUT_RING(CP_PACKET3(R200_CP_PACKET3_3D_DRAW_IMMD_2,
-				3 * vtx_count));
-	    OUT_RING(RADEON_CP_VC_CNTL_PRIM_TYPE_TRI_LIST |
+				4 * vtx_count));
+	    OUT_RING(RADEON_CP_VC_CNTL_PRIM_TYPE_QUAD_LIST |
 		     RADEON_CP_VC_CNTL_PRIM_WALK_RING |
-		     (3 << RADEON_CP_VC_CNTL_NUM_SHIFT));
+		     (4 << RADEON_CP_VC_CNTL_NUM_SHIFT));
 	} else {
 	    BEGIN_RING(3 * vtx_count + 2);
 	    OUT_RING(CP_PACKET3(R200_CP_PACKET3_3D_DRAW_IMMD_2,
@@ -1574,7 +1536,7 @@ FUNC_NAME(RADEONDisplayTexturedVideo)(ScrnInfoPtr pScrn, RADEONPortPrivPtr pPriv
 	}
 #else /* ACCEL_CP */
 	if (IS_R300_3D || IS_R500_3D)
-	    BEGIN_ACCEL(2 + vtx_count * 3);
+	    BEGIN_ACCEL(2 + vtx_count * 4);
 	else
 	    BEGIN_ACCEL(1 + vtx_count * 3);
 
@@ -1584,9 +1546,9 @@ FUNC_NAME(RADEONDisplayTexturedVideo)(ScrnInfoPtr pScrn, RADEONPortPrivPtr pPriv
 					      RADEON_VF_RADEON_MODE |
 					      (3 << RADEON_VF_NUM_VERTICES_SHIFT)));
 	else if (IS_R300_3D || IS_R500_3D)
-	    OUT_ACCEL_REG(RADEON_SE_VF_CNTL, (RADEON_VF_PRIM_TYPE_TRIANGLE_LIST |
+	    OUT_ACCEL_REG(RADEON_SE_VF_CNTL, (RADEON_VF_PRIM_TYPE_QUAD_LIST |
 					      RADEON_VF_PRIM_WALK_DATA |
-					      (3 << RADEON_VF_NUM_VERTICES_SHIFT)));
+					      (4 << RADEON_VF_NUM_VERTICES_SHIFT)));
 	else
 	    OUT_ACCEL_REG(RADEON_SE_VF_CNTL, (RADEON_VF_PRIM_TYPE_RECTANGLE_LIST |
 					      RADEON_VF_PRIM_WALK_DATA |
@@ -1601,28 +1563,31 @@ FUNC_NAME(RADEONDisplayTexturedVideo)(ScrnInfoPtr pScrn, RADEONPortPrivPtr pPriv
 		VTX_OUT_FILTER((float)dstX,                                    (float)dstY,
 			(float)srcX / info->accel_state->texW[0],              (float)srcY / info->accel_state->texH[0],
 			(float)srcX + 0.5,                                     (float)srcY + 0.5);
-		VTX_OUT_FILTER((float)dstX,                                    (float)(dstY + dsth * 2),
-			(float)srcX / info->accel_state->texW[0],              (float)(srcY + srch * 2) / info->accel_state->texH[0],
-			(float)srcX + 0.5,                                     (float)(srcY + srch * 2) + 0.5);
-		VTX_OUT_FILTER((float)(dstX + dstw * 2),                       (float)dstY,
-			(float)(srcX + srcw * 2) / info->accel_state->texW[0], (float)srcY / info->accel_state->texH[0],
-			(float)(srcX + srcw * 2) + 0.5,                        (float)srcY + 0.5);
+		VTX_OUT_FILTER((float)dstX,                                    (float)(dstY + dsth),
+			(float)srcX / info->accel_state->texW[0],              (float)(srcY + srch) / info->accel_state->texH[0],
+			(float)srcX + 0.5,                                     (float)srcY + srch + 0.5);
+		VTX_OUT_FILTER((float)dstX + dstw,                             (float)dstY + dsth,
+			(float)(srcX + srcw) / info->accel_state->texW[0],     (float)(srcY + srch) / info->accel_state->texH[0],
+			(float)srcX + srcw + 0.5,                              (float)srcY + srch + 0.5);
+		VTX_OUT_FILTER((float)dstX + dstw,                             (float)dstY,
+			(float)(srcX + srcw) / info->accel_state->texW[0],     (float)srcY / info->accel_state->texH[0],
+			(float)srcX + srcw + 0.5,                              (float)srcY + 0.5);
 	} else {
 		if (IS_R300_3D || IS_R500_3D) {
 			/*
-			 * Render a big, scissored triangle. This means
-			 * doubling the triangle size and adjusting
-			 * texture coordinates.
+			 * Render a quad.
 			 */
 			VTX_OUT((float)dstX,                                           (float)dstY,
 				(float)srcX / info->accel_state->texW[0],              (float)srcY / info->accel_state->texH[0]);
-			VTX_OUT((float)dstX,                                           (float)(dstY + dsth * 2),
-				(float)srcX / info->accel_state->texW[0],              (float)(srcY + srch * 2) / info->accel_state->texH[0]);
-			VTX_OUT((float)(dstX + dstw * 2),                              (float)dstY,
-				(float)(srcX + srcw * 2) / info->accel_state->texW[0], (float)srcY / info->accel_state->texH[0]);
+			VTX_OUT((float)dstX,                                           (float)dstY + dsth,
+				(float)srcX / info->accel_state->texW[0],              (float)(srcY + srch) / info->accel_state->texH[0]);
+			VTX_OUT((float)dstX + dstw,                              (float)dstY + dsth,
+				(float)(srcX + srcw) / info->accel_state->texW[0], (float)(srcY + srch) / info->accel_state->texH[0]);
+			VTX_OUT((float)(dstX + dstw),                              (float)dstY,
+				(float)(srcX + srcw) / info->accel_state->texW[0], (float)srcY / info->accel_state->texH[0]);
 		} else {
 			/*
-			 * Just render a quad (using three coords).
+			 * Just render a rect (using three coords).
 			 */
 			VTX_OUT((float)dstX,                                       (float)(dstY + dsth),
 				(float)srcX / info->accel_state->texW[0],          (float)(srcY + srch) / info->accel_state->texH[0]);