1741762543
Add a bunch of similar math/unary operations
2016-02-07 18:51:27 +01:00
b1ce9de562
OpLine handling needs to be reworked
2016-02-07 18:51:26 +01:00
8e633c2e46
Fix some indenting
2016-02-07 18:51:25 +01:00
93dac6d4d0
Disassemble image sample operations better
2016-02-07 18:51:24 +01:00
1dc1f071ff
Add support for a couple of misc opcodes
2016-02-07 18:51:23 +01:00
254321fcd1
Quick hacks to show SPIR-V disassembly with hlsl highlighting
2016-02-07 18:51:23 +01:00
1d15a4b2ca
Bump up maximum image attachments on framebuffer by 1 (8 colour + depth)
2016-02-07 18:51:22 +01:00
b4a9e58a87
Don't add extra memory heap if not fiddling with memory types
2016-02-07 18:51:21 +01:00
54a9c8bfd5
Use atomic operations to modify resource record refcount
...
* Scary that this has been broken for so long!
2016-02-07 18:51:20 +01:00
c94e2e60b7
Don't add cache as parent if it's set to NULL
2016-02-07 18:51:19 +01:00
c7269afff2
Add missing lock around m_ImageLayouts access
2016-02-07 18:51:18 +01:00
e3b306d861
Fix temp memory size calculation
2016-02-07 18:51:17 +01:00
1826c166b5
Avoid allocating/deleting array during capture when it's not needed
2016-02-07 18:51:16 +01:00
a98a5fe923
Make sure to unwrap parameters to vkInvalidateMappedMemoryRanges
2016-02-07 18:51:15 +01:00
3f869e55c1
Handle Queue not being created (usually if device is destroyed quickly)
2016-02-07 18:51:15 +01:00
d3c52bbe90
Handle immutable samplers correctly on writing, they are wrapped already
2016-02-07 18:51:14 +01:00
3eca6b5783
Use InstanceIndex and VertexIndex instead of InstanceID and VertexID
...
* These are the only supported indices in vulkan, and need careful
rebasing to obtain 0-based indices from them.
2016-02-07 18:51:13 +01:00
ee7e050756
Set specialization constants properly when creating pipelines
2016-02-07 18:51:12 +01:00
61dae8648f
Make disassembly resiliant to missing children
...
* For now this is lack of handling of spec constant IDs, but remember
this could happen any time in future due to unknown SPIR-V extensions.
2016-02-07 18:51:11 +01:00
fa3043e5c1
Set specialization constant size
2016-02-07 18:51:10 +01:00
c686dd2220
Make sure to only assign string without any trailing chars
2016-02-07 18:51:09 +01:00
51cffd3bb7
Serialise specialization info correctly
2016-02-07 18:51:08 +01:00
1f69ebdfe5
Skip push constant constant blocks when adding resource usage
2016-02-07 18:51:08 +01:00
0b47a7f9ad
Don't use pipeline caches on replay
2016-02-07 18:51:07 +01:00
79ea895e78
Check for fences being live (a bit pointless, but consistent)
2016-02-07 18:51:06 +01:00
6dd4381669
Handle sub-elements of VkGraphicsPipelineCreateInfo validly being NULL
2016-02-07 18:51:05 +01:00
666c9c078b
Don't accept size==0 on vkMapMemory, check for VK_WHOLE_SIZE
2016-02-07 18:51:04 +01:00
73ad2e7b1d
Remove references to $(ProgramFiles)\Windows Kits in search paths
2016-02-07 18:51:03 +01:00
3f0ef3a37a
Remove INCLUDE_D3D_11_1 and INCLUDE_DXGI_1_2
...
* There's no longer any reason for them, as dependencies are packed in
and you don't need the Win 8 SDK installed anymore.
2016-02-07 18:51:02 +01:00
74029dc9d9
Commit lightly modified D3D headers, to allow building without win SDK
...
* Headers taken from windows 10 SDK 10.0.10586.0
* A few includes stripped out of D3D10 and shader headers that aren't
needed. Also winapifamily.h changed from #include <> to "".
2016-02-07 18:51:01 +01:00
35ab1b3dfb
Add missing element being serialised
2016-02-07 18:51:00 +01:00
0039b011b9
Compile fix
2016-02-07 18:50:59 +01:00
570af71c95
Update debug callback to ignore currently false-positive/spammy msgs
...
* Need to re-check this later
2016-02-07 18:50:58 +01:00
cd9cfc5bfa
Only set write pointer for type of descriptor we're writing
2016-02-07 18:50:57 +01:00
9a3905272d
Update to the final frozen 1.0 header
2016-02-07 18:50:57 +01:00
79c863c21d
Make sure to add ALL leftovers after main loop in FakeProfileMarkers
2016-02-07 18:50:56 +01:00
f2d31c6f76
Initialise depthFB/depthRP in case they aren't assigned (no depth buf)
2016-02-07 18:50:55 +01:00
7b7d02c55c
Fix a couple of cases where depth/stencil aspects could still be wrong
2016-02-07 18:50:54 +01:00
7466b53e4b
Enumerate device extensions on CreateDevice for debug logging
...
* Also works around a temporary loader bug
2016-02-07 18:50:53 +01:00
b86389d3ca
Standardise on uint32_t for width/height storage, cast to int32_t
2016-02-07 18:50:52 +01:00
0bdd186550
VS2015 warning fix (shadowed variable)
2016-02-07 18:50:51 +01:00
5b375de3f1
Make sure instance gets wrapped/init'd even if vkCreateInstance fails
2016-02-07 18:50:50 +01:00
29761041d7
Don't track depth and stencil aspects separately, treat them as combined
2016-02-07 18:50:49 +01:00
378d3e9a47
When transitioning image subresources, set the range on each iteration
2016-02-07 18:50:49 +01:00
c81e407046
Set IMAGE_ASPECT_STENCIL correctly in barrier subresource ranges
2016-02-07 18:50:48 +01:00
60c4542bf9
Update glslang to latest
...
* Based on renderdoc branch with local modifications from upstream:
https://github.com/baldurk/glslang/tree/renderdoc
* hash 94b6c8a2c15e934aeec1f942124fd57b754ec252
2016-02-07 18:50:47 +01:00
ce1e11bd37
Gap needed for line break
2016-02-07 18:50:46 +01:00
2e0ffa7813
Batch update copyright years everywhere
2016-02-07 18:50:45 +01:00
0779ef2105
Update image layout transition assertions, but don't uncomment them yet
2016-02-07 18:50:42 +01:00
8179d29f64
Add some access masks for layouts to keep validation layers happy
2016-02-07 18:50:41 +01:00
baldurk