radeonsi: add a HUD query for getting an average GFX BO list size

Reviewed-by: Samuel Pitoiset <[email protected]>
Reviewed-by: Nicolai Hähnle <[email protected]>

7 files changed, 27 insertions, 0 deletions

diff --git a/src/gallium/drivers/radeon/r600_query.c b/src/gallium/drivers/radeon/r600_query.c
index 3308ad82c3d..db70878d1df 100644
--- a/src/gallium/drivers/radeon/r600_query.c
+++ b/src/gallium/drivers/radeon/r600_query.c
@@ -71,6 +71,7 @@ static enum radeon_value_id winsys_id_from_type(unsigned type)
 	case R600_QUERY_NUM_MAPPED_BUFFERS: return RADEON_NUM_MAPPED_BUFFERS;
 	case R600_QUERY_NUM_GFX_IBS: return RADEON_NUM_GFX_IBS;
 	case R600_QUERY_NUM_SDMA_IBS: return RADEON_NUM_SDMA_IBS;
+	case R600_QUERY_GFX_BO_LIST_SIZE: return RADEON_GFX_BO_LIST_COUNTER;
 	case R600_QUERY_NUM_BYTES_MOVED: return RADEON_NUM_BYTES_MOVED;
 	case R600_QUERY_NUM_EVICTIONS: return RADEON_NUM_EVICTIONS;
 	case R600_QUERY_NUM_VRAM_CPU_PAGE_FAULTS: return RADEON_NUM_VRAM_CPU_PAGE_FAULTS;
@@ -173,6 +174,12 @@ static bool r600_query_sw_begin(struct r600_common_context *rctx,
 		query->begin_result = rctx->ws->query_value(rctx->ws, ws_id);
 		break;
 	}
+	case R600_QUERY_GFX_BO_LIST_SIZE:
+		ws_id = winsys_id_from_type(query->b.type);
+		query->begin_result = rctx->ws->query_value(rctx->ws, ws_id);
+		query->begin_time = rctx->ws->query_value(rctx->ws,
+							  RADEON_NUM_GFX_IBS);
+		break;
 	case R600_QUERY_CS_THREAD_BUSY:
 		ws_id = winsys_id_from_type(query->b.type);
 		query->begin_result = rctx->ws->query_value(rctx->ws, ws_id);
@@ -318,6 +325,12 @@ static bool r600_query_sw_end(struct r600_common_context *rctx,
 		query->end_result = rctx->ws->query_value(rctx->ws, ws_id);
 		break;
 	}
+	case R600_QUERY_GFX_BO_LIST_SIZE:
+		ws_id = winsys_id_from_type(query->b.type);
+		query->end_result = rctx->ws->query_value(rctx->ws, ws_id);
+		query->end_time = rctx->ws->query_value(rctx->ws,
+							RADEON_NUM_GFX_IBS);
+		break;
 	case R600_QUERY_CS_THREAD_BUSY:
 		ws_id = winsys_id_from_type(query->b.type);
 		query->end_result = rctx->ws->query_value(rctx->ws, ws_id);
@@ -404,6 +417,10 @@ static bool r600_query_sw_get_result(struct r600_common_context *rctx,
 		return result->b;
 	}
 
+	case R600_QUERY_GFX_BO_LIST_SIZE:
+		result->u64 = (query->end_result - query->begin_result) /
+			      (query->end_time - query->begin_time);
+		return true;
 	case R600_QUERY_CS_THREAD_BUSY:
 	case R600_QUERY_GALLIUM_THREAD_BUSY:
 		result->u64 = (query->end_result - query->begin_result) * 100 /
@@ -1861,6 +1878,7 @@ static struct pipe_driver_query_info r600_driver_query_list[] = {
 	X("num-mapped-buffers",		NUM_MAPPED_BUFFERS,	UINT64, AVERAGE),
 	X("num-GFX-IBs",		NUM_GFX_IBS,		UINT64, AVERAGE),
 	X("num-SDMA-IBs",		NUM_SDMA_IBS,		UINT64, AVERAGE),
+	X("GFX-BO-list-size",		GFX_BO_LIST_SIZE,	UINT64, AVERAGE),
 	X("num-bytes-moved",		NUM_BYTES_MOVED,	BYTES, CUMULATIVE),
 	X("num-evictions",		NUM_EVICTIONS,		UINT64, CUMULATIVE),
 	X("VRAM-CPU-page-faults",	NUM_VRAM_CPU_PAGE_FAULTS, UINT64, CUMULATIVE),
diff --git a/src/gallium/drivers/radeon/r600_query.h b/src/gallium/drivers/radeon/r600_query.h
index 1f97e115906..1bbaa76e584 100644
--- a/src/gallium/drivers/radeon/r600_query.h
+++ b/src/gallium/drivers/radeon/r600_query.h
@@ -69,6 +69,7 @@ enum {
 	R600_QUERY_NUM_MAPPED_BUFFERS,
 	R600_QUERY_NUM_GFX_IBS,
 	R600_QUERY_NUM_SDMA_IBS,
+	R600_QUERY_GFX_BO_LIST_SIZE,
 	R600_QUERY_NUM_BYTES_MOVED,
 	R600_QUERY_NUM_EVICTIONS,
 	R600_QUERY_NUM_VRAM_CPU_PAGE_FAULTS,
diff --git a/src/gallium/drivers/radeon/radeon_winsys.h b/src/gallium/drivers/radeon/radeon_winsys.h
index e19fde6e431..247fff04ac5 100644
--- a/src/gallium/drivers/radeon/radeon_winsys.h
+++ b/src/gallium/drivers/radeon/radeon_winsys.h
@@ -91,6 +91,7 @@ enum radeon_value_id {
     RADEON_TIMESTAMP,
     RADEON_NUM_GFX_IBS,
     RADEON_NUM_SDMA_IBS,
+    RADEON_GFX_BO_LIST_COUNTER, /* number of BOs submitted in gfx IBs */
     RADEON_NUM_BYTES_MOVED,
     RADEON_NUM_EVICTIONS,
     RADEON_NUM_VRAM_CPU_PAGE_FAULTS,
diff --git a/src/gallium/winsys/amdgpu/drm/amdgpu_cs.c b/src/gallium/winsys/amdgpu/drm/amdgpu_cs.c
index c88be069d06..1b3ca6546b8 100644
--- a/src/gallium/winsys/amdgpu/drm/amdgpu_cs.c
+++ b/src/gallium/winsys/amdgpu/drm/amdgpu_cs.c
@@ -1225,6 +1225,9 @@ void amdgpu_cs_submit_ib(void *job, int thread_index)
          cs->flags[i] = (util_last_bit64(buffer->u.real.priority_usage) - 1) / 4;
       }
 
+      if (acs->ring_type == RING_GFX)
+         ws->gfx_bo_list_counter += cs->num_real_buffers;
+
       r = amdgpu_bo_list_create(ws->dev, cs->num_real_buffers,
                                 cs->handles, cs->flags,
                                 &cs->request.resources);
diff --git a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
index c4d52166fb7..2148c496d7d 100644
--- a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
+++ b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
@@ -140,6 +140,8 @@ static uint64_t amdgpu_query_value(struct radeon_winsys *rws,
       return ws->num_gfx_IBs;
    case RADEON_NUM_SDMA_IBS:
       return ws->num_sdma_IBs;
+   case RADEON_GFX_BO_LIST_COUNTER:
+      return ws->gfx_bo_list_counter;
    case RADEON_NUM_BYTES_MOVED:
       amdgpu_query_info(ws->dev, AMDGPU_INFO_NUM_BYTES_MOVED, 8, &retval);
       return retval;
diff --git a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.h b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.h
index 896a463ee6f..f011b8eeedf 100644
--- a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.h
+++ b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.h
@@ -65,6 +65,7 @@ struct amdgpu_winsys {
    uint64_t num_gfx_IBs;
    uint64_t num_sdma_IBs;
    uint64_t num_mapped_buffers;
+   uint64_t gfx_bo_list_counter;
 
    struct radeon_info info;
 
diff --git a/src/gallium/winsys/radeon/drm/radeon_drm_winsys.c b/src/gallium/winsys/radeon/drm/radeon_drm_winsys.c
index 2a6ce2ca216..8e43b68b229 100644
--- a/src/gallium/winsys/radeon/drm/radeon_drm_winsys.c
+++ b/src/gallium/winsys/radeon/drm/radeon_drm_winsys.c
@@ -631,6 +631,7 @@ static uint64_t radeon_query_value(struct radeon_winsys *rws,
     case RADEON_NUM_EVICTIONS:
     case RADEON_NUM_VRAM_CPU_PAGE_FAULTS:
     case RADEON_VRAM_VIS_USAGE:
+    case RADEON_GFX_BO_LIST_COUNTER:
         return 0; /* unimplemented */
     case RADEON_VRAM_USAGE:
         radeon_get_drm_value(ws->fd, RADEON_INFO_VRAM_USAGE,