diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index 7428c4305418e1bb4c01a1c156059f3c3347c2dd..cec46ebae5f795fbcbde6bd95c8fc73fb4803345 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -2518,6 +2518,13 @@ static void gfx_v7_0_ring_emit_ib(struct amdgpu_ring *ring,
 {
 	u32 header, control = 0;
 	u32 next_rptr = ring->wptr + 5;
+
+	/* drop the CE preamble IB for the same context */
+	if ((ring->type == AMDGPU_RING_TYPE_GFX) &&
+	    (ib->flags & AMDGPU_IB_FLAG_PREAMBLE) &&
+	    !ring->need_ctx_switch)
+		return;
+
 	if (ring->type == AMDGPU_RING_TYPE_COMPUTE)
 		control |= INDIRECT_BUFFER_VALID;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 48de9204ff5e2f0b11b6f1ae4fa5059c4898144a..fc8c46209db4e382f02333ca65d1f7c60e17608f 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -3647,6 +3647,13 @@ static void gfx_v8_0_ring_emit_ib(struct amdgpu_ring *ring,
 {
 	u32 header, control = 0;
 	u32 next_rptr = ring->wptr + 5;
+
+	/* drop the CE preamble IB for the same context */
+	if ((ring->type == AMDGPU_RING_TYPE_GFX) &&
+	    (ib->flags & AMDGPU_IB_FLAG_PREAMBLE) &&
+	    !ring->need_ctx_switch)
+		return;
+
 	if (ring->type == AMDGPU_RING_TYPE_COMPUTE)
 		control |= INDIRECT_BUFFER_VALID;
 
diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h
index ca0ea1efa3f4ca9c8dcd4147e07a6b4aa18d9134..fb428fe54186e24431cd8c87ef8c2bb9617e6de4 100644
--- a/include/uapi/drm/amdgpu_drm.h
+++ b/include/uapi/drm/amdgpu_drm.h
@@ -369,6 +369,9 @@ union drm_amdgpu_cs {
 /* GDS is used by this IB */
 #define AMDGPU_IB_FLAG_GDS	(1<<1)
 
+/* CE Preamble */
+#define AMDGPU_IB_FLAG_PREAMBLE (1<<2)
+
 struct drm_amdgpu_cs_chunk_ib {
 	/**
 	 * Handle of GEM object to be used as IB or 0 if it is already in