diff options
| author | Bas Nieuwenhuizen <[email protected]> | 2017-03-12 14:12:19 +0100 |
|---|---|---|
| committer | Bas Nieuwenhuizen <[email protected]> | 2017-03-13 02:14:29 +0100 |
| commit | 7c282b3ca11aa1d9e05bd7fea19bbf464ac5e6fb (patch) | |
| tree | 9a2b77a2a63c409e73c1cf003a0ac8080b9f6b23 /src/amd/vulkan/radv_pipeline.c | |
| parent | e27fdbcb4c951bba0649cf60f2ae377f842267bb (diff) | |
radv: Store shaders in VRAM.
Less IFETCH latency on misses. Shader code is write once read many,
so GTT doesn't make much sense anyway.
If it turns out to fragment the CPU visible VRAM too much, we can upload with SDMA.
Signed-off-by: Bas Nieuwenhuizen <[email protected]>
Reviewed-by: Dave Airlie <[email protected]>
Diffstat (limited to 'src/amd/vulkan/radv_pipeline.c')
| -rw-r--r-- | src/amd/vulkan/radv_pipeline.c | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/amd/vulkan/radv_pipeline.c b/src/amd/vulkan/radv_pipeline.c index 723c32c4aaf..ce228df04a8 100644 --- a/src/amd/vulkan/radv_pipeline.c +++ b/src/amd/vulkan/radv_pipeline.c @@ -410,7 +410,7 @@ static void radv_fill_shader_variant(struct radv_device *device, S_00B848_FLOAT_MODE(variant->config.float_mode); variant->bo = device->ws->buffer_create(device->ws, binary->code_size, 256, - RADEON_DOMAIN_GTT, RADEON_FLAG_CPU_ACCESS); + RADEON_DOMAIN_VRAM, RADEON_FLAG_CPU_ACCESS); void *ptr = device->ws->buffer_map(variant->bo); memcpy(ptr, binary->code, binary->code_size); |