Commit Graph
1624 Commits
Author SHA1 Message Date
Andreas Reich 5e488d4fea Added player test with compute shader reading & writing an uninitialized buffer 2021-02-03 23:44:42 +01:00
bors[bot]andAndreas Reich 780637eec0 Merge #1187
1187: validate for strip_index_format being used with non-strip topology r=kvark a=Wumpf

**Description**
Hit lack of this validation when porting a project to newest wgpu-rs: Accidentally set the strip_index_format for a TriangleList pipeline which left me with a Vulkan validation error (since wgpu decides to set `restart_index` in presence of a `strip_index_format`)

```
VALIDATION [VUID-VkPipelineInputAssemblyStateCreateInfo-topology-00428 (-1077750125)] : Validation Error: [ VUID-VkPipelineInputAssemblyStateCreateInfo-topology-00428 ] Object 0: handle = 0x1ebbfa512a8, type = VK_OBJECT_TYPE_DEVICE; | MessageID 
= 0xbfc2d693 | vkCreateGraphicsPipelines() pCreateInfo[0]: topology is VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST and primitiveRestartEnable is VK_TRUE. It is invalid. The Vulkan spec states: If topology is VK_PRIMITIVE_TOPOLOGY_POINT_LIST, VK_PRIMITIVE_TOPOLOGY_LINE_LIST, VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST, VK_PRIMITIVE_TOPOLOGY_LINE_LIST_WITH_ADJACENCY, VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST_WITH_ADJACENCY or VK_PRIMITIVE_TOPOLOGY_PATCH_LIST, primitiveRestartEnable must be VK_FALSE (https://vulkan.lunarg.com/doc/view/1.2.141.0/windows/1.2-extensions/vkspec.html#VUID-VkPipelineInputAssemblyStateCreateInfo-topology-00428)
    object info: (type: DEVICE, hndl: 2112044208808)
```

(arguably wgpu could be more clever about this and just not set `restart_index`, but that leaves the user code in a non-sense state)

**Testing**
One-off test triggering the error through wgpu-rs.

Co-authored-by: Andreas Reich <r_andreas2@web.de>
2021-02-01 21:12:19 +00:00
Andreas Reich 9dd88cb338 validate for strip_index_format being used with non-strip topology 2021-02-01 21:46:24 +01:00
bors[bot]andDaniel Tang aea2af588c Merge #1183
1183: Improve docs language r=cwfitzgerald a=danielzgtg

**Description**

- s/fact culling/face culling/g
- s/are draw /are drawn /g

**Testing**

- There are no code change.
- The documentation should be less confusing


Co-authored-by: Daniel Tang <danielzgtg.opensource@gmail.com>
2021-02-01 08:13:20 +00:00
bors[bot]andConnor Fitzgerald e15eefc718 Merge #1185
1185: [master] Fix Readme Date r=trivial a=cwfitzgerald

Port #1184 to master

Co-authored-by: Connor Fitzgerald <connorwadefitzgerald@gmail.com>
2021-02-01 07:46:17 +00:00
Connor Fitzgerald c43bfaf8e0 Fix readme date 2021-02-01 02:45:31 -05:00
Daniel Tang 07a2c9a712 Improve docs language 2021-02-01 02:19:32 -05:00
bors[bot]andDzmitry Malyshau 7c7501cab7 Merge #1182
1182: Update all versions r=kvark a=kvark

**Connections**
Picks up https://github.com/gfx-rs/gfx/pull/3620 and a bunch of other fixes in gfx-rs and naga.

**Description**
Updates the dependencies ("gfx-9" naga train) as well as self version to 0.7, to match the v0.7 branch.

**Testing**
Should just work :)

Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com>
2021-02-01 05:37:39 +00:00
Dzmitry Malyshau 8ed543f26a Update all versions 2021-02-01 00:35:40 -05:00
bors[bot]andAndreas Reich da8d17d8d3 Merge #1159
1159: Zero initialize buffers r=kvark a=Wumpf

**Connections**
First half of  #563, focusing solely on buffers and ignoring same issue for textures

**Description**
Tracks for each buffer which parts are initialized (i.e. set to zero). Identified three interaction points for this:
* buffer mapping: Zeroing out ad-hoc and marking as initialized
* queue write to buffer: Marks target buffer regions as initialized (i.e. optimizes away buffer init down the line)
* use in binding or copy operation in a command buffer:
   * fine grained tracking of regions that may require init (or will be initialized implicitly) on each command buffer
   * set in motion on queue submit, init is exclusively with `fill_buffer`

Todo list for Ready-to-Review

- [x] memory barriers for `fill_buffer` calls
- [x] better data structure for `memory_init_tracker`
- [x] coarse filtering on command-buffer buffer init requirements (the list should almost always be empty whereas now it pushes any buffer use)
- [x] improve naming of things
- [x] at least pretend this is adequately tested

Todo list beyond this PR

* make data structures usable for textures
  * and.. well.. implement all this for textures!
* explore reusing barrier tracker for memory init tracking?


**Testing**

* Some basic testing by doing some changes to wgpu-rs samples and watching them in in the debugger.
* Added a ron test file for the player (love those!) to poke the feature a little bit
* MemoryInitTracker comes with simple unit tests

Overall this is a bit shallow but as so often in this area accurate testing is hard because the outcomes are quite indirect



Co-authored-by: Andreas Reich <r_andreas2@web.de>
2021-02-01 03:17:36 +00:00
Andreas Reich 822424e519 Fix buffer init in bundle.rs with mem::size_of::<wgt::DrawIndirectArgs> 2021-01-31 22:01:52 +01:00
Andreas Reich 4f5acba5fb Fixed too eager lock usage, fixed O(N²) buffer range collapse 2021-01-31 21:56:31 +01:00
Andreas Reich a8460b438c MemoryInitTracker drain is now O(N) 2021-01-31 21:56:31 +01:00
Andreas Reich 018ad05f56 memory init tracker check (previous is_initialized) clamps range now
and is O(log n)!
2021-01-31 21:56:31 +01:00
Andreas Reich 31d292b169 added clear helper method to memory_init_tracker, renamed drain 2021-01-31 21:56:31 +01:00
Andreas Reich da86b3f401 Cleanup, better use of some rust std lib functions 2021-01-31 21:56:31 +01:00
Andreas Reich cc183c885f Added player test for buffer-zero-init 2021-01-31 21:56:31 +01:00
Andreas Reich 216dcf949b MemoryInitTracker has now a custom impl that doesn't depend on range-alloc 2021-01-31 21:56:31 +01:00
Andreas Reich 32b4e32ac6 MemoryInitTracker tests and interface adjustments 2021-01-31 21:56:31 +01:00
Andreas Reich 492027fe6e command buffer buffer memory init tracking is only filled if necessary now 2021-01-31 21:56:22 +01:00
Andreas Reich 22fb9ab1eb buffer zero init on queue_submit emits correct barriers now
refactored this zero init into separate method
2021-01-31 21:55:49 +01:00
Andreas Reich 5d2afa52c1 Fix zero init for buffers mapped at cration that are not mappable 2021-01-31 21:55:49 +01:00
Andreas Reich 1efdc2b229 Queue_submit zeros out uninitialized buffer regions now
again but with fine grained scheme now
2021-01-31 21:55:49 +01:00
Andreas Reich 9595b39bb3 Fine grained tracking of buffer init requirements 2021-01-31 21:55:49 +01:00
Andreas Reich 5133b0da94 Introduced separate MemoryInitTracker 2021-01-31 21:55:49 +01:00
Andreas Reich 076aecd150 buffer map permeates zero init now in any case 2021-01-31 21:55:49 +01:00
Andreas Reich a51b4f9232 buffer zero initialization prior to use in submit 2021-01-31 21:55:49 +01:00
Andreas Reich 875cfb5961 buffers are now zero initialized for mappings
queue_write_buffer marks as initialized
2021-01-31 21:55:49 +01:00
bors[bot]andAndreas Reich 1668bb2451 Merge #1180
1180: Add dummy structs for indirect command buffers r=cwfitzgerald,kvark a=Wumpf

**Connections**
Review of #1159 which could make more use of this expressiveness patch of having the buffers spelled out.

**Description**
Add dummy structs for indirect command buffers and (bonus!) fix missing IndirectBufferOverrun error for IndirectDispatch.

**Testing**
squinting at it only 👁️👁️

Co-authored-by: Andreas Reich <r_andreas2@web.de>
2021-01-31 20:41:22 +00:00
Andreas Reich 6b51328b2e Add public structs for indirect command buffers
and fix missing IndirectBufferOverrun error for IndirectDispatch
2021-01-31 21:32:33 +01:00
bors[bot]andDzmitry Malyshau ac81f3e756 Merge #1179
1179: Fix depth bias documentation r=kvark a=kvark

**Connections**
Follow-up to #1178

**Description**
Clippy complained...

**Testing**
No need

Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com>
2021-01-29 05:27:29 +00:00
Dzmitry Malyshau 4099c2ca8e Fix depth bias documentation 2021-01-29 00:25:54 -05:00
bors[bot]andDzmitry Malyshau 513c08ecbb Merge #1178
1178: Add DepthBiasState, rename StencilStateFace r=kvark a=kvark

**Connections**
None

**Description**
Found the error in the `StencilStateFace` name and fixed it here.
Also identified that depth bias is a noticeable ergonomic offender, so moving it out into a struct, similar to what we have with stencil.

**Testing**
tested on wgpu-rs examples.

Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com>
2021-01-29 05:09:44 +00:00
Dzmitry Malyshau 14eb99da87 Add DepthBiasState, rename StencilStateFace 2021-01-29 00:07:46 -05:00
bors[bot]andDzmitry Malyshau 6c0e652f86 Merge #1177
1177: Update gfx with the new iterator API r=kvark a=kvark

**Connections**
Gets https://github.com/gfx-rs/gfx/pull/3615 and https://github.com/gfx-rs/gfx/pull/3619

**Description**
Avoids a few `collect()` calls when talking to gfx.

**Testing**
just local playtests

Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com>
2021-01-28 17:42:08 +00:00
Dzmitry Malyshau 4764ff21e9 Update gfx with the new iterator API 2021-01-28 12:40:24 -05:00
bors[bot]andDzmitry Malyshau 63b0d36074 Merge #1176
1176: Rewrite RenderPipelineDescriptor according to aspects r=grovesNL,cwfitzgerald a=kvark

**Connections**
Fixes #1166

**Description**
Totally re-imagines the structure tree for describing the rendering pipeline, based on the aspects instead of logical steps.

**Testing**
Just local testing. I don't think there is any concern that these changes might not work, they just need the clients to adapt.

Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com>
2021-01-28 06:21:57 +00:00
Dzmitry Malyshau cef0a30aa2 Add 0.7 changelog 2021-01-28 01:21:20 -05:00
Dzmitry Malyshau 0346b43982 Update the render pipeline API 2021-01-27 15:09:13 -05:00
bors[bot]andDzmitry Malyshau 60cd0f7e56 Merge #1173
1173: Handle zero-sized bindings as an error r=cwfitzgerald a=kvark

**Connections**
Fixes #1165

**Description**
Adds buffer IDs to the binding errors, also adds a variant for zero sized bindings.

**Testing**
Tested on a hand-written example.

Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com>
2021-01-26 05:19:26 +00:00
Dzmitry Malyshau c233a70afc Handle zero-sized bindings as an error 2021-01-25 22:33:44 -05:00
bors[bot]andDzmitry Malyshau c2c8d2203e Merge #1170
1170: Don't build shader interface without the validation flag r=kvark a=kvark

**Connections**
Fixes #1169

**Description**
We used to try building the interface of a Naga module before (!) the validation has happened, and even without a check for the validation flag.

**Testing**
Untested.

Co-authored-by: Dzmitry Malyshau <dmalyshau@mozilla.com>
2021-01-25 19:16:59 +00:00
Dzmitry Malyshau 6a684e5cdf Don't build shader interface without the validation flag 2021-01-25 14:15:27 -05:00
Jonathan Behrensandmonocodus[bot] a97ec9df91 Add SHADER_FLOAT64 feature (#1154)
* Add SHADER_FLOAT64 feature

* Also enable SHADER_FLOAT64 hal feature when creating the device

* Update wgpu-core/src/instance.rs

Co-authored-by: monocodus[bot] <49363530+monocodus[bot]@users.noreply.github.com>

* Add warning about 64-bit float performance

* Add (currently unsupported) VERTEX_ATTRIBUTE_64BIT feature

Co-authored-by: monocodus[bot] <49363530+monocodus[bot]@users.noreply.github.com>
2021-01-24 21:59:09 -05:00
bors[bot]andDzmitry Malyshau ab6af6ed4f Merge #1167
1167: Remove temporary allocation for gpu-descriptor r=kvark a=kvark

This is no longer needed as https://github.com/zakarumych/gpu-descriptor/pull/10 is merged

Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com>
2021-01-24 17:25:23 +00:00
Dzmitry Malyshau 6538ac5d26 Remove temporary allocation for gpu-descriptor 2021-01-24 12:24:37 -05:00
bors[bot]andDzmitry Malyshau 0e68554c52 Merge #1164
1164: Update naga to gfx-7 r=kvark a=kvark

**Connections**
Includes https://github.com/gfx-rs/gfx/pull/3612

**Description**
Carries a bunch of important Naga improvements.

**Testing**
Just local playtests

Co-authored-by: Dzmitry Malyshau <kvarkus@gmail.com>
2021-01-24 03:18:39 +00:00
Dzmitry Malyshau 952ccbcf85 Update naga to gfx-7 2021-01-23 22:17:09 -05:00
bors[bot]andMatt Keeter b23ece55e7 Merge #1163
1163: Bump gfx to latest master r=kvark a=mkeeter

**Connections**
- This bumps `gfx` to [#3610](https://github.com/gfx-rs/gfx/pull/3610), and also includes [#3608](https://github.com/gfx-rs/gfx/pull/3608) and [#3609](https://github.com/gfx-rs/gfx/pull/3608)
- [Here](https://github.com/gfx-rs/wgpu-native/pull/66) is a draft PR to `wgpu-native`
- `wgpu-rs` requires a one-line fix to the `texture_view_drop` call (which now takes a boolean); I can PR this next.

**Description**
This fixes [pathological shader complexity in SPIRV-Cross](https://github.com/KhronosGroup/SPIRV-Cross/pull/1594), as well as a few other `gfx` PRs.

**Testing**
I updated `wgpu-native` to use this branch, then updated my [toy raytracer](https://github.com/mkeeter/rayray) to use the resulting `dylibs` and confirmed that it no longer takes forever to compile the pathological shader.

In addition, I updated `wgpu-rs` and went through the examples; nothing seems out of place.

Co-authored-by: Matt Keeter <matt.j.keeter@gmail.com>
2021-01-23 16:57:10 +00:00
Matt Keeter 38b184428d Bump gfx to latest master 2021-01-22 08:58:51 -05:00