diff options
| author | Pauli Nieminen <suokkos@gmail.com> | 2010-02-15 13:40:37 +0200 |
|---|---|---|
| committer | Michel Dänzer <michel@daenzer.net> | 2010-02-15 17:10:06 +0100 |
| commit | 78e7047c5235b09858b66dd3688d39aaa27d7589 (patch) | |
| tree | 9eebea9c29c1cf4b2b0648a64c220fdc4d2ad63a /src/radeon_video.c | |
| parent | 3ec25e59854b6b03ad763bc374d3475a50f562d8 (diff) | |
Allocate Xv buffers to GTT.
KMS doesn't have acceleration for upload to vram. memcpy/memmove to VRAM
directly is very slow (40M/s in benchmark) which causes visible problems
to video.
Allocating video buffer in GTT will give good performance (350-450M/s)
for memmove operation. This is nice performance boost for Xv under KMS.
Signed-off-by: Pauli Nieminen <suokkos@gmail.com>
Diffstat (limited to 'src/radeon_video.c')
| -rw-r--r-- | src/radeon_video.c | 9 |
1 files changed, 6 insertions, 3 deletions
diff --git a/src/radeon_video.c b/src/radeon_video.c index d7700fa0..92f65a6e 100644 --- a/src/radeon_video.c +++ b/src/radeon_video.c @@ -2949,7 +2949,8 @@ RADEONPutImage( } pPriv->video_offset = radeon_legacy_allocate_memory(pScrn, &pPriv->video_memory, (pPriv->doubleBuffer ? - (new_size * 2) : new_size), 64); + (new_size * 2) : new_size), 64, + RADEON_GEM_DOMAIN_GTT); if (pPriv->video_offset == 0) return BadAlloc; @@ -3179,7 +3180,8 @@ RADEONAllocateSurface( pitch = ((w << 1) + 15) & ~15; size = pitch * h; - offset = radeon_legacy_allocate_memory(pScrn, &surface_memory, size, 64); + offset = radeon_legacy_allocate_memory(pScrn, &surface_memory, size, 64, + RADEON_GEM_DOMAIN_GTT); if (offset == 0) return BadAlloc; @@ -3517,7 +3519,8 @@ RADEONPutVideo( pPriv->video_offset = radeon_legacy_allocate_memory(pScrn, &pPriv->video_memory, (pPriv->doubleBuffer ? - (new_size * 2) : new_size), 64); + (new_size * 2) : new_size), 64, + RADEON_GEM_DOMAIN_GTT); if (pPriv->video_offset == 0) return BadAlloc; |