drm/amdgpu: use LRU mapping policy for SDMA engines
authorAndres Rodriguez <andresx7@gmail.com>
Fri, 17 Mar 2017 18:41:21 +0000 (14:41 -0400)
committerAlex Deucher <alexander.deucher@amd.com>
Wed, 31 May 2017 20:49:04 +0000 (16:49 -0400)
Spreading the load across multiple SDMA engines can increase memory
transfer performance.

Signed-off-by: Andres Rodriguez <andresx7@gmail.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c

index 9a14c27..befc09b 100644 (file)
@@ -276,7 +276,6 @@ int amdgpu_queue_mgr_map(struct amdgpu_device *adev,
 
        switch (mapper->hw_ip) {
        case AMDGPU_HW_IP_GFX:
-       case AMDGPU_HW_IP_DMA:
        case AMDGPU_HW_IP_UVD:
        case AMDGPU_HW_IP_VCE:
        case AMDGPU_HW_IP_UVD_ENC:
@@ -284,6 +283,7 @@ int amdgpu_queue_mgr_map(struct amdgpu_device *adev,
        case AMDGPU_HW_IP_VCN_ENC:
                r = amdgpu_identity_map(adev, mapper, ring, out_ring);
                break;
+       case AMDGPU_HW_IP_DMA:
        case AMDGPU_HW_IP_COMPUTE:
                r = amdgpu_lru_map(adev, mapper, ring, out_ring);
                break;