Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | vk: Implement scratch buffers to make spilling work | Kristian Høgsberg Kristensen | 2015-06-19 | 4 | -47/+72 |
| | |||||
* | vk: Undo relocs for scratch bos | Kristian Høgsberg Kristensen | 2015-06-19 | 3 | -95/+38 |
| | |||||
* | vk/allocator: Get rid of non-memfd path | Kristian Høgsberg Kristensen | 2015-06-19 | 1 | -20/+0 |
| | | | | We can just use modern valgrind now. | ||||
* | vk/headers: Make General State offsets relocations | Kristian Høgsberg Kristensen | 2015-06-19 | 3 | -38/+95 |
| | |||||
* | vk: Add missing gen7 RENDER_SURFACE_STATE struct | Kristian Høgsberg Kristensen | 2015-06-17 | 3 | -0/+209 |
| | |||||
* | vk: Compute CS exec mask and thread width max in pipeline | Kristian Høgsberg Kristensen | 2015-06-12 | 3 | -32/+25 |
| | | | | | | | | We compute the right mask and thread width max parameters as part of pipeline creation and set them accordingly at vkCmdDispatch() and vkCmdDispatchIndirect() time. These parameters depend only on the local group size and the dispatch width of the program so we can figure this out at pipeline create time. | ||||
* | vk: Set binding table layout for CS | Kristian Høgsberg Kristensen | 2015-06-12 | 1 | -0/+1 |
| | | | | We weren't setting the binding table layout for the backend compiler. | ||||
* | vk: Generate CS prog_data into the pipeline instance | Kristian Høgsberg Kristensen | 2015-06-12 | 1 | -6/+7 |
| | | | | | We were generating the prog_data into a local variable and never initializing the pipeline->cs_prog_data one. | ||||
* | vk: Document how depth/stencil formats work in anv_image_create() | Kristian Høgsberg Kristensen | 2015-06-11 | 1 | -9/+13 |
| | | | | | | | | | | This reverts commits e17ed04 * vk/image: Don't double-allocate stencil buffers 1ee2d1c * vk/image: Teach anv_image_choose_tile_mode about WMAJOR and instead adds a comment to describe the subtlety of how we create images for stencil only formats. | ||||
* | vk: Use compute pipeline layout when binding compute sets | Kristian Høgsberg Kristensen | 2015-06-11 | 1 | -1/+6 |
| | |||||
* | vk: Implement basic compute shader support | Kristian Høgsberg Kristensen | 2015-06-11 | 7 | -63/+393 |
| | |||||
* | vk: Emit PIPELINE_SELECT on demand | Kristian Høgsberg Kristensen | 2015-06-11 | 2 | -3/+10 |
| | |||||
* | vk: Stop asserting we have a fragment shader | Kristian Høgsberg Kristensen | 2015-06-11 | 1 | -2/+0 |
| | | | | Even for graphics, this is not a requirement, we can have a depth-only output pipeline. | ||||
* | vk: Defer setting viewport dynamic state | Kristian Høgsberg Kristensen | 2015-06-11 | 2 | -11/+12 |
| | | | | We can't emit this until we've done a 3D pipeline select. | ||||
* | vk: Disable shader stages in the graphics pipeline batch | Kristian Høgsberg Kristensen | 2015-06-11 | 2 | -22/+22 |
| | | | | | We need to move this into the graphics pipeline batch so we don't emit it for compute pipelines. | ||||
* | vk: Don't emit STATE_SIP | Kristian Høgsberg Kristensen | 2015-06-11 | 1 | -1/+0 |
| | | | | We don't have a SIP kernel and don't enable exceptions. | ||||
* | vk: Compile fragment shader after VS and GS | Kristian Høgsberg Kristensen | 2015-06-11 | 1 | -12/+15 |
| | | | | Just moving code around to do shader stages in the natual order. | ||||
* | vk/entrypoints: Don't print every single function call | Jason Ekstrand | 2015-06-11 | 1 | -13/+0 |
| | |||||
* | vk: Remove left-over trp call | Kristian Høgsberg Kristensen | 2015-06-11 | 1 | -2/+0 |
| | |||||
* | vk: Set maximum point size range | Kristian Høgsberg Kristensen | 2015-06-11 | 1 | -1/+3 |
| | | | | | We set both minimum and maximum point size to 0 in 3DSTATE_CLIP, which will clip away all points. | ||||
* | vk: Use generated headers with fixed point support | Kristian Høgsberg Kristensen | 2015-06-11 | 5 | -283/+283 |
| | | | | | The generated headers now convert float in the template struct to the correct fixed point format. | ||||
* | vk: Regenerate headers with __gen_validate_value() | Kristian Høgsberg Kristensen | 2015-06-11 | 2 | -0/+14 |
| | |||||
* | vk/formats: Refactor format properties code | Jason Ekstrand | 2015-06-11 | 1 | -17/+33 |
| | | | | | Along with the refactor, we now do the right thing when we hit an unsupported format: Set the flags to 0 and return VK_SUCCESS. | ||||
* | vk/image: Add a bunch of asserts | Jason Ekstrand | 2015-06-10 | 1 | -0/+26 |
| | |||||
* | vk: Add a couple vk_error calls | Jason Ekstrand | 2015-06-10 | 2 | -2/+2 |
| | |||||
* | vk/private: Add a non-fatal assert | Jason Ekstrand | 2015-06-10 | 1 | -0/+10 |
| | |||||
* | vk/cmd: Add an initial implementation of PipelineBarrier | Jason Ekstrand | 2015-06-10 | 1 | -1/+113 |
| | | | | | We may want to do something more inteligent here later such as actually handling image layout transitions. However, this should do for now. | ||||
* | vk/emit: Use valgrind to validate every packed field | Jason Ekstrand | 2015-06-10 | 2 | -1/+8 |
| | |||||
* | vk: Add valgrind checks in various emit functions | Jason Ekstrand | 2015-06-09 | 2 | -0/+7 |
| | | | | | | The check in batch_bo_finish should catch any undefined values in the batch but isn't that great for debugging. The checks in the various emit functions will help get better granularity. | ||||
* | vk: Move the valgrind include and VG() macro to private.h | Jason Ekstrand | 2015-06-09 | 3 | -12/+8 |
| | |||||
* | vk/image: Don't double-allocate stencil buffers | Chad Versace | 2015-06-09 | 1 | -1/+1 |
| | | | | | If the main surface has format S8_UINT, then don't allocate the auxiliary stencil surface. | ||||
* | vk/image: Teach anv_image_choose_tile_mode about WMAJOR | Chad Versace | 2015-06-09 | 1 | -1/+8 |
| | |||||
* | vk/util: Add anv_abortf(), anv_abortfv() | Chad Versace | 2015-06-09 | 2 | -0/+22 |
| | | | | Convenience functions to print an error message then abort. | ||||
* | vk: Define anv_noreturn macro | Chad Versace | 2015-06-09 | 1 | -0/+1 |
| | |||||
* | vk/image: Factor tile mode selection into separate function | Chad Versace | 2015-06-09 | 1 | -15/+19 |
| | | | | Because it will eventually need to get smarter. | ||||
* | vk/device: Actually allow destruction | Jason Ekstrand | 2015-06-09 | 1 | -3/+0 |
| | |||||
* | vk/cmd_buffer: Properly initialize/reset dynamic states | Jason Ekstrand | 2015-06-09 | 1 | -0/+6 |
| | |||||
* | vk/pipeline: Zero out the depth-stencil state when not in use | Jason Ekstrand | 2015-06-09 | 1 | -5/+10 |
| | |||||
* | vk/device: Use anv_CreateDynamicViewportState instead of the vk one | Jason Ekstrand | 2015-06-09 | 1 | -1/+1 |
| | |||||
* | vk/device: Dedent the vkCreateDynamicViewportState call | Jason Ekstrand | 2015-06-09 | 1 | -19/+19 |
| | |||||
* | vk/util: Annotate anv_finishme() as printflike | Chad Versace | 2015-06-09 | 2 | -2/+3 |
| | |||||
* | vk: Define anv_printflike() macro | Chad Versace | 2015-06-09 | 1 | -0/+2 |
| | |||||
* | vk/image: Stop hardcoding alignment of stencil surfaces | Chad Versace | 2015-06-09 | 1 | -1/+1 |
| | | | | Look up the alignment from anv_tile_info_table. | ||||
* | vk/image: Rewrite tile info table | Chad Versace | 2015-06-09 | 1 | -24/+31 |
| | | | | | | - Reduce the number of table lookups in anv_image_create from 4 to 1. - Add field for surface alignment. - Shorten field names tile_width, tile_height -> width, height. | ||||
* | vk/image: Delete an old comment | Chad Versace | 2015-06-09 | 1 | -2/+0 |
| | |||||
* | vk/compiler: Free the GL errors data | Jason Ekstrand | 2015-06-09 | 1 | -0/+1 |
| | |||||
* | vk/compiler: Free more of prog_data when tearing down a pipeline | Jason Ekstrand | 2015-06-09 | 1 | -2/+6 |
| | |||||
* | vk/queue: Embed the queue in and allocate it with the device | Jason Ekstrand | 2015-06-09 | 2 | -35/+55 |
| | |||||
* | vk/device: Free border color states when we have valgrind | Jason Ekstrand | 2015-06-09 | 1 | -0/+10 |
| | |||||
* | vk/device: Destroy all batch buffers | Jason Ekstrand | 2015-06-09 | 1 | -2/+2 |
| | | | | | Due to a copy+paste error, we were destroying all but the first batch or surface state buffer. Now we destroy them all. |