OSDN Git Service
Jason Ekstrand [Fri, 4 Dec 2015 19:08:37 +0000 (11:08 -0800)]
vk: Update the README for 0.210.1
Kristian Høgsberg [Tue, 1 Dec 2015 23:39:30 +0000 (15:39 -0800)]
vk: Turn on Bay Trail, Cherryview and Broxton support
Kristian Høgsberg Kristensen [Fri, 4 Dec 2015 07:58:05 +0000 (23:58 -0800)]
vk: Map uncached, coherent memory as write-combine
This gives us the required characteristics for the memory type.
Kristian Høgsberg Kristensen [Fri, 4 Dec 2015 07:09:09 +0000 (23:09 -0800)]
vk: Expose two memory types for non-LLC GPUs
We're required to expose a host-visible, coherent memory type. On big
core GPUs that share, LLC, we can expose one such memory type that's
also cached. However, on non-LLC GPUs we can't both be cached and
coherent. Thus, we expose both the required coherent type and the cached
but non-coherent combination.
Kristian Høgsberg [Tue, 1 Dec 2015 23:37:12 +0000 (15:37 -0800)]
vk: clflush all state for non-LLC GPUs
Kristian Høgsberg [Thu, 3 Dec 2015 20:09:33 +0000 (12:09 -0800)]
vk: Set I915_CACHING_NONE for userptr BOs when !llc
Regular objects are created I915_CACHING_CACHED on LLC platforms and
I915_CACHING_NONE on non-LLC platforms. However, userptr objects are
always created as I915_CACHING_CACHED, which on non-LLC means
snooped. That can be useful but comes with a bit of overheard. Since
we're eplicitly clflushing and don't want the overhead we need to turn
it off.
Kristian Høgsberg [Tue, 1 Dec 2015 23:25:07 +0000 (15:25 -0800)]
vk: Implement vkFlushMappedMemoryRanges()
We'll do a runtime switch on device->info.has_llc for now.
Jason Ekstrand [Fri, 4 Dec 2015 02:07:58 +0000 (18:07 -0800)]
nir/spirv: Update to SPIR-V version 1.0
Chad Versace [Thu, 3 Dec 2015 21:07:06 +0000 (13:07 -0800)]
anv/gen9: Fix SURFACE_STATE halign and valign
Pre-Skylake, RENDER_SUFFACE_STATE.SurfaceVerticalAlignment is in units
of surface samples. A surface sample is equivalent to a pixel in all
surfaces except interleaved multisample surfaces.
In Skylake, it is in units of surface elements. A surface element is
equivalent to a surface sample except for compressed formats, in which
case the element is a compression block.
Chad Versace [Thu, 3 Dec 2015 16:40:47 +0000 (08:40 -0800)]
anv: Embed isl_surf into anv_surface
This reduces struct anv_surface to just two members: an offset and the
embedded isl_surf.
Chad Versace [Thu, 3 Dec 2015 02:27:35 +0000 (18:27 -0800)]
anv/image: Drop assertions on SURFTYPE extent limits
In anv_image_create(), stop asserting that VkImageCreateInfo::extent
does not exceed the hardware limits for the given SURFTYPE. The
assertions were incorrect because they did not take into account the
hardware gen. Anyways, these types of assertions belong in isl, not
anvil.
Chad Versace [Thu, 3 Dec 2015 00:46:16 +0000 (16:46 -0800)]
anv/image: Use isl to calculate surface layout
Remove the surface layout calculations in anv_image_make_surface(). Let
isl_surf_init() do the heavy lifting.
Fixes 8 Crucible tests and regresses none. (hw=Broadwell and
crucible@
33d91ec).
Chad Versace [Sat, 14 Nov 2015 00:01:35 +0000 (16:01 -0800)]
isl: Implement isl_surf_init() for gen4-gen9
This is a big code push. The patch is about 3000 lines.
Function isl_surf_init() calculates the physical layout of a surface.
The implementation is "complete" (but untested) for all 1D, 2D, 3D, and
cube surfaces for gen4 through gen9, except:
* gen9 1D surfaces
* gen9 Ys multisampled surfaces
* auxiliary surfaces (such as hiz, mcs, ccs)
Chad Versace [Wed, 2 Dec 2015 17:14:48 +0000 (09:14 -0800)]
isl: Rename legacy Y tiling to ISL_TILING_Y0
Rename legacy Y tiling from ISL_TILING_Y to ISL_TILING_Y0 in order to
clearly distinguish it from Yf and Ys. Using ISL_TILING_Y to denote
legacy Y tiling would lead to confusion with i965, because i965 uses
I195_TILE_Y to denote *any* Y tiling.
Chad Versace [Thu, 3 Dec 2015 02:41:08 +0000 (18:41 -0800)]
anv/image: Vulkan's depthPitch is in bytes, not rows
Fix for VkGetImageSubresourceLayout.
Jason Ekstrand [Thu, 3 Dec 2015 23:23:33 +0000 (15:23 -0800)]
anv/device: Give a version of 0.210.1 in apiVersion
Jason Ekstrand [Thu, 3 Dec 2015 00:28:36 +0000 (16:28 -0800)]
vk: Add new WSI support and bump the API to 0.210.1
Jason Ekstrand [Thu, 3 Dec 2015 06:06:22 +0000 (22:06 -0800)]
anv/entrypoints: Run the headers through the preprocessor first
This allows us to filter based on preprocessor directives. We could build
a partial preprocessor into the generator, but we would likely get it
wrong. This allows us to filter out, for instance, windows-specific WSI
stuff.
Jason Ekstrand [Thu, 3 Dec 2015 21:27:00 +0000 (13:27 -0800)]
vk/0.210.0: Advertise version 0.210.0
Jason Ekstrand [Thu, 3 Dec 2015 21:25:12 +0000 (13:25 -0800)]
vk/0.210.0: Update queries to the new API
Jason Ekstrand [Thu, 3 Dec 2015 19:05:52 +0000 (11:05 -0800)]
vk/0.210.0: Fix how we handle access flags in barriers
The initial implementation in the 0.210.0 API update was misguieded as to
what the access flags meant. This should be more correct.
Jason Ekstrand [Thu, 3 Dec 2015 01:47:37 +0000 (17:47 -0800)]
vk/0.210.0: Update the VkFormat enum
Jason Ekstrand [Thu, 3 Dec 2015 01:18:41 +0000 (17:18 -0800)]
vk/0.210.0: Rework vkQueueSubmit
Jason Ekstrand [Thu, 3 Dec 2015 01:07:26 +0000 (17:07 -0800)]
vk/0.210.0: Remove depth clip and add depth clamp
Jason Ekstrand [Thu, 3 Dec 2015 00:58:54 +0000 (16:58 -0800)]
vk/0.210.0: Rework device features and limits
Jason Ekstrand [Thu, 3 Dec 2015 00:20:40 +0000 (16:20 -0800)]
vk/0.210.0: Rework QueueFamilyProperties
Jason Ekstrand [Thu, 3 Dec 2015 00:14:58 +0000 (16:14 -0800)]
vk/0.210.0: Rework result and structure type enums
By and large, this is just moving enum values around. However, it also
removed VK_UNSUPPORTED which we were returning a number of places. Those
places now return VK_ERROR_INCOMPATABLE_DRIVER.
Jason Ekstrand [Thu, 3 Dec 2015 00:08:13 +0000 (16:08 -0800)]
vk/0.210.0: Remove the VkShaderStage enum
This made for an unfortunately large amount of work since we were using it
fairly heavily internally. However, gl_shader_stage does basically the
same things, so it's not too bad.
Jason Ekstrand [Wed, 2 Dec 2015 22:35:07 +0000 (14:35 -0800)]
vk/0.210.0: Remove VkShader
Jason Ekstrand [Wed, 2 Dec 2015 18:39:15 +0000 (10:39 -0800)]
vk/0.210.0: Rework memory property flags
Jason Ekstrand [Wed, 2 Dec 2015 18:30:56 +0000 (10:30 -0800)]
vk/0.210.0: Remove some unused enum values
Jason Ekstrand [Wed, 2 Dec 2015 18:29:50 +0000 (10:29 -0800)]
vk/0.210.0: Update VkPipelineStageFlagBits
Jason Ekstrand [Wed, 2 Dec 2015 18:27:29 +0000 (10:27 -0800)]
vk/0.210.0: Trivial function argument name change
Jason Ekstrand [Wed, 2 Dec 2015 11:48:58 +0000 (03:48 -0800)]
vk/0.210.0: We now allocate command buffers; not create them
Jason Ekstrand [Wed, 2 Dec 2015 11:34:43 +0000 (03:34 -0800)]
vk/0.210.0: Rename a parameter to GetImageSparseMemoryRequirements
Jason Ekstrand [Wed, 2 Dec 2015 11:33:20 +0000 (03:33 -0800)]
vk/0.210.0: Delete three no longer existant entrypoints
Jason Ekstrand [Wed, 2 Dec 2015 11:28:27 +0000 (03:28 -0800)]
vk/0.210.0: Rework allocation to use the new pAllocator's
Jason Ekstrand [Tue, 1 Dec 2015 22:09:17 +0000 (14:09 -0800)]
vk/0.210.0: Use VkSampleCountFlagBits for sample counts
Jason Ekstrand [Tue, 1 Dec 2015 21:58:25 +0000 (13:58 -0800)]
vk/0.210.0: Rework VkInstanceCreateInfo
Jason Ekstrand [Tue, 1 Dec 2015 21:50:09 +0000 (13:50 -0800)]
vk/0.210.0: More function argument renaming
Jason Ekstrand [Tue, 1 Dec 2015 21:39:28 +0000 (13:39 -0800)]
vk/0.210.0: Replace MemoryInput/OutputFlags with AccessFlags
Jason Ekstrand [Tue, 1 Dec 2015 21:09:22 +0000 (13:09 -0800)]
vk/0.210.0: Rework render pass description structures
Jason Ekstrand [Tue, 1 Dec 2015 20:52:56 +0000 (12:52 -0800)]
vk/0.210.0: More structure field renaming
Jason Ekstrand [Tue, 1 Dec 2015 20:19:11 +0000 (12:19 -0800)]
vk/0.210.0: Get rid of VkImageAspect
Jason Ekstrand [Tue, 1 Dec 2015 20:05:21 +0000 (12:05 -0800)]
vk/0.210.0: Rework descriptor sets
Jason Ekstrand [Tue, 1 Dec 2015 05:18:12 +0000 (21:18 -0800)]
vk/0.210.0: Rename parameters to memory binding/mapping functions
Jason Ekstrand [Tue, 1 Dec 2015 05:10:14 +0000 (21:10 -0800)]
vk/0.210.0: Update to the new instance/device create structs
Jason Ekstrand [Tue, 1 Dec 2015 04:59:58 +0000 (20:59 -0800)]
vk/0.210.0: More trivial struct/enum changes
Jason Ekstrand [Tue, 1 Dec 2015 02:15:41 +0000 (18:15 -0800)]
vk/0.210.0: Trivial flag enum updates
Jason Ekstrand [Tue, 1 Dec 2015 02:12:55 +0000 (18:12 -0800)]
vk/0.210.0: Rename ChannelFlags to ColorComponentFlags
Jason Ekstrand [Tue, 1 Dec 2015 02:05:00 +0000 (18:05 -0800)]
vk/0.210.0: s/raster/rasterization/
Jason Ekstrand [Tue, 1 Dec 2015 01:54:51 +0000 (17:54 -0800)]
vk/0.210.0: Don't allow chaining of description structs
Jason Ekstrand [Tue, 1 Dec 2015 01:46:38 +0000 (17:46 -0800)]
vk/0.210.0: More fun with flags fields
Jason Ekstrand [Tue, 1 Dec 2015 01:30:38 +0000 (17:30 -0800)]
vk/0.210.0: Make pCode a uint32_t pointer
Jason Ekstrand [Tue, 1 Dec 2015 01:26:32 +0000 (17:26 -0800)]
vk/0.210.0: Rename origin fields of VkViewport
Jason Ekstrand [Tue, 1 Dec 2015 01:20:49 +0000 (17:20 -0800)]
vk/0.210.0: Move alphaToOne and alphaToCoverate to multisample state
Jason Ekstrand [Tue, 1 Dec 2015 01:13:19 +0000 (17:13 -0800)]
vk/0.210.0: Add flags fields to various pipeline create structs
Jason Ekstrand [Tue, 1 Dec 2015 01:00:30 +0000 (17:00 -0800)]
vk/0.210.0: Change field names in vertex input structs
Jason Ekstrand [Tue, 1 Dec 2015 00:42:12 +0000 (16:42 -0800)]
vk/0.210.0: Misc. no-op structure changes
The only non-trivial change is to sparse resources that we don't handle
anyway.
Jason Ekstrand [Tue, 1 Dec 2015 00:28:36 +0000 (16:28 -0800)]
vk/0.210.0: Rename property pCount parameters
Jason Ekstrand [Tue, 1 Dec 2015 00:21:38 +0000 (16:21 -0800)]
vk/0.210.0: Rework sampler filtering and mode enums
Jason Ekstrand [Tue, 1 Dec 2015 00:10:02 +0000 (16:10 -0800)]
vk/0.210.0: Misc. function argument renames
Jason Ekstrand [Mon, 30 Nov 2015 22:52:38 +0000 (14:52 -0800)]
vk/0.210.0: Rework copy/clear/blit API
Jason Ekstrand [Mon, 30 Nov 2015 22:19:41 +0000 (14:19 -0800)]
vk/0.210.0: Rework dynamic states
Jason Ekstrand [Mon, 30 Nov 2015 21:49:28 +0000 (13:49 -0800)]
vk/0.210.0: Rework color blending enums
Jason Ekstrand [Mon, 30 Nov 2015 21:29:49 +0000 (13:29 -0800)]
gen7/8/cmd_buffer: Inline vk_to_gen_swizzle
It's currently unused on IVB so we get compiler warnings.
Jason Ekstrand [Mon, 30 Nov 2015 21:28:09 +0000 (13:28 -0800)]
vk/0.210.0: Rework a few raster/input enums
Jason Ekstrand [Mon, 30 Nov 2015 21:06:12 +0000 (13:06 -0800)]
vk/0.210.0: Rework texture view component mapping
Jason Ekstrand [Mon, 30 Nov 2015 20:21:19 +0000 (12:21 -0800)]
vk/0.210.0: Switch to the new VKAPI function decorations
While we're at it, we do a bunch of the VkResult -> void updates
Jason Ekstrand [Mon, 30 Nov 2015 19:48:08 +0000 (11:48 -0800)]
vk/0.210.0: Rename CmdBuffer to CommandBuffer
Jason Ekstrand [Mon, 30 Nov 2015 19:12:44 +0000 (11:12 -0800)]
vk/0.210.0: A pile of minor enum updates
Jason Ekstrand [Mon, 30 Nov 2015 18:31:44 +0000 (10:31 -0800)]
vk/0.210.0: Switch to the new-style handle declarations
Jason Ekstrand [Mon, 30 Nov 2015 18:24:33 +0000 (10:24 -0800)]
vk: Add connonical 0.170.2 and 0.210.0 headers
This is in preparation for the API update
Kristian Høgsberg Kristensen [Thu, 26 Nov 2015 18:11:52 +0000 (10:11 -0800)]
vk: Fix 3DSTATE_WM_DEPTH_STENCIL for gen8
This packet is a different size on gen8 and we hit an assertion when we
try to merge a gen9 size dword array from the pipeline with the gen8
sized array we create from dynamic state.
Use a static assert in the merge macro and fix this issue by using different
wm_depth_stencil arrays on gen8 and gen9.
Kristian Høgsberg Kristensen [Thu, 26 Nov 2015 06:27:01 +0000 (22:27 -0800)]
vk: Add SKL support
Signed-off-by: Kristian Høgsberg Kristensen <kristian.h.kristensen@intel.com>
Kristian Høgsberg Kristensen [Tue, 15 Sep 2015 19:37:21 +0000 (12:37 -0700)]
vk: Make entrypoint generator output gen9 entry points
Signed-off-by: Kristian Høgsberg Kristensen <kristian.h.kristensen@intel.com>
Kristian Høgsberg Kristensen [Tue, 15 Sep 2015 19:26:24 +0000 (12:26 -0700)]
vk: Add GEN9 pack header
Kristian Høgsberg Kristensen [Wed, 25 Nov 2015 22:13:53 +0000 (14:13 -0800)]
vk: Move all gen8 files to gen8 lib
Jason Ekstrand [Mon, 23 Nov 2015 22:03:47 +0000 (14:03 -0800)]
Merge remote-tracking branch 'mesa-public/master' into vulkan
This pulls in nir cloning and some much-needed upstream refactors.
Jason Ekstrand [Thu, 12 Nov 2015 02:41:37 +0000 (18:41 -0800)]
i965: Use NIR for lowering texture swizzle
Now that nir_lower_tex can do texture swizzle lowering, we can use that
instead of repeating more-or-less the same code in both backends. This
both allows us to share code and means that things like the tg4
work-arounds are somewhat simpler because they don't have to take the
swizzle into account.
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Thu, 12 Nov 2015 02:30:31 +0000 (18:30 -0800)]
nir/lower_tex: Add support for lowering texture swizzle
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Thu, 12 Nov 2015 02:30:09 +0000 (18:30 -0800)]
nir: Add a tex_instr_is_query helper
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Thu, 12 Nov 2015 16:40:17 +0000 (08:40 -0800)]
nir: Add a ssa_def_rewrite_uses_after helper
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Thu, 12 Nov 2015 05:13:26 +0000 (21:13 -0800)]
nir: Use instr/if_rewrite in nir_ssa_def_rewrite_uses
nir_ssa_def_rewrite_uses is one of the older helpers in NIR and predated
both of those. Now it can be substantially simplified.
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Thu, 12 Nov 2015 18:38:12 +0000 (10:38 -0800)]
nir/validate: Validated dests after sources
Previously, if someone accidentally made an instruction that refers to its
own SSA destination, the validator wouldn't catch it. The reason for this
is that it validated the destination too early and, by the time it got to
the source, the destination SSA value was already added to the set of seen
SSA values so it would assume that it came from some previous instruction.
By moving destination validation to be after source validation, the SSA
value is not in the list of seen values and the validator will catch
self-referential instructions.
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Wed, 11 Nov 2015 19:01:59 +0000 (11:01 -0800)]
i965: Use nir_lower_tex for texture coordinate lowering
Previously, we had a rescale_texcoords helper in the FS backend for
handling rescaling of texture coordinates. Now that we can do variants in
NIR, we can use nir_lower_tex to do the rescaling for us. This allows us
to delete the i965-specific code and gives us proper TEXTURE_RECTANGLE and
GL_CLAMP handling in vertex and geometry shaders.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Jason Ekstrand [Wed, 11 Nov 2015 23:46:55 +0000 (15:46 -0800)]
i965/fs: Stomp the texture return type to UINT32 for resinfo messages
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Jason Ekstrand [Wed, 11 Nov 2015 20:01:20 +0000 (12:01 -0800)]
nir/lower_tex: Set the dest_type for txs instructions
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Jason Ekstrand [Wed, 11 Nov 2015 18:46:09 +0000 (10:46 -0800)]
nir/lower_tex: Report progress
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Jason Ekstrand [Wed, 11 Nov 2015 18:04:43 +0000 (10:04 -0800)]
i965: Move postprocess_nir to codegen time
This allows us to insert NIR passes between initial NIR compilation and
optimization (link time) and actual backend code-gen. In particular, it
will allow us to do shader variants in NIR and share some of that shader
variant code between backends.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Jason Ekstrand [Wed, 11 Nov 2015 17:40:51 +0000 (09:40 -0800)]
i965/nir: Split shader optimization and lowering into three stages
At the moment, brw_create_nir just calls the three stages in sequence so
there's not much difference. Soon, however, we will want to start doing
variants in NIR at which point the postprocessing step will have to move
from shader create time to codegen time.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Jason Ekstrand [Mon, 23 Nov 2015 18:53:01 +0000 (10:53 -0800)]
i965: Use ull immediates in brw_inst_bits
This fixes a regression introduced in
b1a83b5d1 that caused basically all
shaders to fail to compile on 32-bit platforms.
Reported-by: Mark Janes <mark.a.janes@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ilia Mirkin [Mon, 23 Nov 2015 17:31:59 +0000 (12:31 -0500)]
docs: add missed freedreno features to relnotes
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "11.1" <mesa-stable@lists.freedesktop.org>
Ilia Mirkin [Mon, 23 Nov 2015 17:04:50 +0000 (12:04 -0500)]
docs: update relnotes with new freedreno/a4xx support
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Jose Fonseca [Mon, 23 Nov 2015 16:45:28 +0000 (16:45 +0000)]
svga: Add ASTC formats to format table.
Fixes build. Otherwise untested.
Trivial.
Ilia Mirkin [Sun, 22 Nov 2015 22:46:13 +0000 (17:46 -0500)]
freedreno/ir3: add support for a few gs5 ops
Tested on a4xx. This is part of the builtins added by ARB_gpu_shader5
and GLSL ES 3.10.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Sun, 22 Nov 2015 22:37:47 +0000 (17:37 -0500)]
ttn: fix UMSB conversion
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Sun, 22 Nov 2015 21:47:25 +0000 (16:47 -0500)]
freedreno/a4xx: add ARB_texture_query_lod support
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Sun, 22 Nov 2015 21:41:16 +0000 (16:41 -0500)]
ttn: add LODQ support
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Sun, 22 Nov 2015 19:06:26 +0000 (14:06 -0500)]
freedreno/a4xx: re-emit program on dirty framebuffer
The program emit depends on certain fb details. Make sure those get
updated when the fb changes.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>