Commit Graph

420 Commits

Author SHA1 Message Date
Hans-Kristian Arntzen
012377f811 Refactor block IO emission to use bitset. 2018-03-12 15:59:34 +01:00
Hans-Kristian Arntzen
9fbd8b789e Update tests for latest SPIRV-Tools and glslang. 2018-03-12 15:11:55 +01:00
Hans-Kristian Arntzen
e8e58844d4 Rewrite everything to use Bitset rather than uint64_t. 2018-03-12 13:24:14 +01:00
Sidney Just
e66fd6c2a0 Added generation of the textureXYZGrad() legacy GLSL instruction 2018-03-12 00:59:06 +10:00
Hans-Kristian Arntzen
9a52713d77 Run format_all.sh. 2018-03-09 15:26:36 +01:00
Hans-Kristian Arntzen
1de74fdabd Explicitly emit break/continues in selection branches. 2018-03-09 13:59:39 +01:00
Hans-Kristian Arntzen
e0efa737ca Expand the implementation of inherit_expression_dependencies. 2018-03-09 13:21:38 +01:00
Hans-Kristian Arntzen
28cccc3dbb Emit complex continue blocks "properly". 2018-03-08 17:59:21 +01:00
Hans-Kristian Arntzen
8d557d4103 Handle cases where merge selects as also loop merge or continue blocks. 2018-03-08 14:01:10 +01:00
Hans-Kristian Arntzen
922420e346 Disallow arrays and structs from becoming loop variables.
Fixes awkward code-gen issue.
2018-03-07 14:54:11 +01:00
Hans-Kristian Arntzen
2e5d06d1ce
Merge pull request #485 from KhronosGroup/fix-484
Add true FP16 support.
2018-03-07 11:59:12 +01:00
Hans-Kristian Arntzen
47d94ff8d9 Add FP16 to HLSL.
Cannot be used in buffer types, similar to mediump in GLSL.
half is useless, because it's 32-bit in FXC.
2018-03-07 10:21:25 +01:00
Hans-Kristian Arntzen
15a941cd3d Add builtin GLSL functions to blacklist. 2018-03-06 17:37:47 +01:00
Hans-Kristian Arntzen
d9da2db442 Some compat fixes for MSL and Half. 2018-03-06 17:09:18 +01:00
Hans-Kristian Arntzen
24dfe496c2 Run format_all.sh. 2018-03-06 17:09:18 +01:00
Hans-Kristian Arntzen
05348a66ca Add test shader for FP16 support. 2018-03-06 17:09:18 +01:00
Hans-Kristian Arntzen
bc0f698df3 Fix packing alignments for float16. 2018-03-06 17:09:18 +01:00
Hans-Kristian Arntzen
91f85d3412 Begin adding float16_t support to GLSL. 2018-03-06 17:09:18 +01:00
Hans-Kristian Arntzen
294259e2f1 Fix type aliasing on MSL.
Be careful about who gets to be the alias master, and don't alias types
when we have packed types in play.
2018-03-05 16:27:04 +01:00
Hans-Kristian Arntzen
3c1b147272 Support Invariant for BuiltInPosition. 2018-03-01 12:31:39 +01:00
Hans-Kristian Arntzen
e69b1aeed8 Fix implicit conversion bug. 2018-02-26 09:29:18 +01:00
Hans-Kristian Arntzen
fda36f832b Fix function overload when SPIR-V overloads on pointer type. 2018-02-25 10:58:22 +01:00
Hans-Kristian Arntzen
dd603eab58 Support spec constant array size in blocks.
Won't really be correct if the spec constant is changed outside
SPIRV-Cross, but nothing we can do about that, really.
2018-02-23 15:11:45 +01:00
Hans-Kristian Arntzen
a04bdcc7f7 Handle overloaded functions which share the same OpName.
Awkward, but legal SPIR-V.
2018-02-23 14:15:51 +01:00
Hans-Kristian Arntzen
047ad7df0f Support special float constants (NaN/Inf). 2018-02-23 13:06:20 +01:00
Hans-Kristian Arntzen
fb3f92a3ff Overhaul clip/cull distance support in GLSL. 2018-02-22 14:36:50 +01:00
Hans-Kristian Arntzen
47b37423a2 Run format_all.sh. 2018-02-21 13:46:16 +01:00
Hans-Kristian Arntzen
1a2e4de7a5 Add test for texelFetch without sampler. 2018-02-21 13:45:59 +01:00
Hans-Kristian Arntzen
54a065bb5f Run format_all.sh. 2018-02-15 13:32:49 +01:00
Hans-Kristian Arntzen
3fa6cc8f2c Implement FRem. 2018-02-15 13:31:29 +01:00
Hans-Kristian Arntzen
636cc30088 Fix case where hoisted temporaries were used before being declared. 2018-02-15 10:52:56 +01:00
Bill Hollings
2964e328e6 CompilerMSL support gl_SampleMask and convert it to scalar uint from array. 2018-02-13 14:44:40 -05:00
Bill Hollings
b453348370 Merge branch 'master' of https://github.com/billhollings/SPIRV-Cross 2018-02-11 16:54:25 -05:00
Bill Hollings
607b0d6d42 CompilerMSL support smaller offsets for 3-row row-major matrices.
Support MSL typedefs to declare 3-row row-major matrices as 3-column matrices.
Allow those matrices to be decorated as packed.
Support transposing those matrices when used.
Modify how member alignments are calculated.
2018-02-11 16:52:57 -05:00
Hans-Kristian Arntzen
a3ae861844 Fix depth image usage in MSL for separate image/samplers. 2018-02-10 10:55:10 +01:00
Hans-Kristian Arntzen
18a594a76b Implement subpass input support in HLSL. 2018-02-10 10:54:42 +01:00
Hans-Kristian Arntzen
00ccd590ee Return arrays in HLSL/MSL by writing to an output variable instead. 2018-02-08 12:22:08 +01:00
Hans-Kristian Arntzen
9fa91f7e1c Support returning arrays from functions in GLSL/MSL.
Not possible in HLSL apparently, need workaround ...
2018-02-08 12:22:08 +01:00
msiglreith
d096f5cafe hlsl: Support custom root constant layout 2018-02-07 15:21:52 +01:00
Hans-Kristian Arntzen
4b58f65af7
Merge pull request #438 from zeux/msvc-warning-fix
Fix MSVC 2017 warning
2018-02-07 13:01:25 +01:00
Arseny Kapoulkine
24cf3080cd Fix MSVC 2015 warning
Fixes the following warning:
> C4146: unary minus operator applied to unsigned type, result still unsigned
2018-02-06 21:43:24 -08:00
twinaphex
59a12c7e71 Uniquely name these static functions 2018-02-03 23:23:30 +01:00
aliaspider
5fc48b61fa add matrix to the list of illegal names. 2018-02-02 12:56:15 +01:00
Hans-Kristian Arntzen
5d9df6a31c Do not declare constant composites inline in HLSL.
Move arrays and structs out to their own global static constants.

Also, replace illegal names in HLSL as well.
2018-02-02 10:12:26 +01:00
Hans-Kristian Arntzen
8841f15283 Fix case where loop header branches into continue block. 2018-02-01 10:00:16 +01:00
Bill Hollings
1c94715350 Update copyright dates to 2018 in main files. 2018-01-31 17:08:43 -05:00
Hans-Kristian Arntzen
6714a9fa23 Need to sort declared temporaries to ensure stable output. 2018-01-29 10:24:15 +01:00
Bill Hollings
e43f244399 Merge branch 'master' of https://github.com/KhronosGroup/SPIRV-Cross 2018-01-24 17:34:50 -05:00
Bill Hollings
fe3683eefa CompilerMSL declare threadgroup variables accessed in called functions. 2018-01-24 15:38:17 -05:00
Hans-Kristian Arntzen
b3f6e3de8e Fix CFG::update_common_dominator.
The algorithm was too conservative causing lots of unnecessary
temporaries to be created.
2018-01-24 20:32:11 +01:00
Hans-Kristian Arntzen
09f550f718 Handle exponential explosion of code-gen during first phase of compile.
Certain patterns with OpVectorShuffle (and probably others) will cascade
to so large, that they can cause OOM. After we have observed
force_recompile, don't spend unnecessary memory emitting code which will
never be used.
2018-01-24 18:12:41 +01:00
Hans-Kristian Arntzen
af0a887997 Add test for false loop init.
Clean up how for loop variables are declared.
2018-01-23 21:15:09 +01:00
Hans-Kristian Arntzen
4a7a37256e Check if a loop variable candidate is actually used.
Phi nodes used in continue blocks are not always loop variables.
Fix by checking if path from dominator to loop header has some variable
use.
2018-01-23 20:27:43 +01:00
Hans-Kristian Arntzen
56acf9728a
Merge pull request #406 from KhronosGroup/fix-405
Fix HLSL regression with struct declaration.
2018-01-23 16:54:13 +01:00
Hans-Kristian Arntzen
06041985d0 Fix HLSL regression with struct declaration.
It actually worked surprisingly. Fix it properly.
2018-01-23 16:36:20 +01:00
Hans-Kristian Arntzen
d0ce948df4 Do not merge swizzles for anything other than vectors. 2018-01-22 09:52:57 +01:00
Hans-Kristian Arntzen
7d223b8987 Fix CFG for forwarded temporaries.
Forwarded temporaries would never declare a temporary.
Figure out all result types ahead of time so we can deal with those
temporaries as well.
2018-01-18 12:11:33 +01:00
Hans-Kristian Arntzen
377c6db480
Merge pull request #398 from KhronosGroup/fix-396
Do not use inline for-loop initializers with different types.
2018-01-17 10:10:01 +01:00
Hans-Kristian Arntzen
34cbe91d12
Merge pull request #395 from KhronosGroup/cfg-analysis-opt-remove
Remove cfg_analysis option.
2018-01-16 13:05:42 +01:00
Hans-Kristian Arntzen
b902d5400c Do not use inline for-loop initializers with different types. 2018-01-16 10:27:58 +01:00
Hans-Kristian Arntzen
bfe6f50b8e Remove cfg_analysis option.
Not using this is broken, and won't work anymore.
2018-01-15 13:21:01 +01:00
Hans-Kristian Arntzen
0f4adaa09d Handle OpUnreachable in code-gen.
Forgot to add it to emit_block_chain. It should just be a noop.
2018-01-15 09:35:09 +01:00
Hans-Kristian Arntzen
17f9026248
Merge pull request #392 from billhollings/master
Use initializer list for composite initializations
2018-01-15 09:12:13 +01:00
Bill Hollings
ba1e415a9c Use initializer list for composite initializations if backend.use_initializer_list is on. 2018-01-12 17:19:24 -05:00
Hans-Kristian Arntzen
d4e470babd Analyze the CFG for temporaries as well.
Normally, temporary declaration must dominate any use of it,
so we generally did not need to analyze the CFG for these variables,
but there is an edge case where you have an inliner doing:

do {
	create_temporary;
	break;
} while(0);

use_temporary;

The inside of the loop dominates the outer scope, but we cannot emit
code like this in GLSL, so make sure we hoist these temporaries outside
the "loop".
2018-01-12 10:56:11 +01:00
Hans-Kristian Arntzen
44a4eb7562 Formatting. 2018-01-09 12:51:21 +01:00
Hans-Kristian Arntzen
9c3d4e7c60 Rewrite barrier handling in HLSL. 2018-01-09 12:41:13 +01:00
Hans-Kristian Arntzen
7bb8874b43 Optimize away redundant barriers. 2018-01-09 12:17:38 +01:00
Hans-Kristian Arntzen
9c72aa00c9 Overhaul barrier handling in GLSL. 2018-01-09 12:07:07 +01:00
Hans-Kristian Arntzen
1a5a755ce5 Flatten structs based on instance name rather than block name.
Block name is too brittle, optimizes will merge block types, making
flattening impossible with unique names.
2018-01-09 10:36:04 +01:00
Hans-Kristian Arntzen
f708b497a4 Opt in to gl_in/gl_out handling rather than other way around. 2018-01-09 09:16:33 +01:00
Bill Hollings
27d4af75a0 Revert to not forcing gl_in/gl_out block for MSL, and add MSL gl_ClipDistance tests. 2018-01-08 16:18:34 -05:00
Hans-Kristian Arntzen
513ba86fc8
Merge pull request #373 from billhollings/master
CompilerMSL enhancements PLUS refactoring of MSL tests to avoid optimization crashes
2018-01-08 09:04:03 +01:00
Bill Hollings
5ee6b46087 Fixes from review of PR #373.
Code fixes from review.
Refactor MSL tests back to using the SPIRV-Tools
and glslang loaded by checkout_glslang_spirv_tools.sh.
2018-01-05 23:22:36 -05:00
Bill Hollings
d8d2da9d8d CompilerMSL allow swizzle of packed_float3 vectors by unpacking to float3.
Pass packed indicator back through OpAccessChain and OpLoad.
Unpack packed vector before applying swizzle.
Add packed swizzle test.
2018-01-05 17:46:56 -05:00
Bill Hollings
8890578d2a CompilerMSL support conversion of non-square row-major matrices. 2018-01-04 16:33:45 -05:00
Hans-Kristian Arntzen
9bdfd7025e Fix tessellation control shaders from HLSL.
Need to take into account that gl_Position[i] needs to be rewritten as
gl_out[i].gl_Position instead.
2018-01-04 16:22:44 +01:00
Hans-Kristian Arntzen
8412b10550
Merge pull request #380 from KhronosGroup/fix-368
Add support for gl_ViewportIndex.
2018-01-04 14:09:36 +01:00
Hans-Kristian Arntzen
4590c63b40 Fix swizzling of integers.
Add scalar swizzling to constant composites as well.
2018-01-04 13:16:56 +01:00
Hans-Kristian Arntzen
f825bd90ef Support ViewportIndex builtin. 2018-01-04 12:41:25 +01:00
Bill Hollings
1845f31397 Add option to control use of temp vars for OpVectorShuffle. 2017-12-31 18:55:04 -05:00
Hans-Kristian Arntzen
ffad50b3c2 Remove duplicate swizzles in remap_swizzle.
Makes HLSL image-load-store look much nicer.
2017-12-12 13:01:10 +01:00
Hans-Kristian Arntzen
1c7980a778 Swizzle scalars when splatting in HLSL. 2017-12-12 12:52:45 +01:00
Hans-Kristian Arntzen
950732512e Fix compatibility with FXC. 2017-12-12 12:35:22 +01:00
Hans-Kristian Arntzen
2e68675ef7 Support mediump in desktop Vulkan GLSL. 2017-12-06 10:25:58 +01:00
Hans-Kristian Arntzen
b737d2bc07 Lift variable accesses in continue blocks out to loop header. 2017-12-05 17:40:23 +01:00
Hans-Kristian Arntzen
4a9d3cebd0
Merge pull request #355 from KhronosGroup/fix-345
Improve handling of block name declaration in GLSL.
2017-12-02 13:01:32 +01:00
Hans-Kristian Arntzen
3c52771aee Make sure image integer coords are int, not uint.
HLSL can emit uint here.
2017-12-01 15:02:50 +01:00
Hans-Kristian Arntzen
2c90ea3acc Improve handling of block name declaration in GLSL.
HLSL UAVs are a bit annoying because they can share block types,
so reflection becomes rather awkward. Sometimes we will need to make
some nasty fallbacks, so add a reflection interface which lets you query
post-shader compile which names was actually declared in the shader.
2017-12-01 14:30:10 +01:00
Hans-Kristian Arntzen
7238e57933 Enforce stable order for dominated phi variables. 2017-11-23 09:59:25 +01:00
Hans-Kristian Arntzen
3fac289dea Do not allow base expressions for arrays either. 2017-11-23 09:50:11 +01:00
Hans-Kristian Arntzen
473787ec0b Check forwarding in OpImageRead. 2017-11-23 09:50:11 +01:00
Hans-Kristian Arntzen
a8df0802db Clear member name cache in emit_flattened_io_block. 2017-11-23 09:50:11 +01:00
Hans-Kristian Arntzen
0fd028147f Hoist all phi-local variables to entry block.
We don't have a mechanism to move temporaries to their appropriate
scope, and Phi behavior is weird enough that it will be a heroic effort
to not do this rather ugly codegen :(
2017-11-23 09:50:11 +01:00
Hans-Kristian Arntzen
bcdff2d2e1 Fixups for PR #338 review. 2017-11-22 20:51:26 +01:00
Lou Kramer
6671f52334 Add support for new extensions. 2017-11-22 19:05:38 +01:00
Hans-Kristian Arntzen
b629ca1c33 Fix complicated Phi case.
A continue block might have used a temporary which existed only in the
loop body.
2017-11-21 09:27:49 +01:00
Hans-Kristian Arntzen
851e584338 Do not allow base expressions when extracting struct members. 2017-11-20 21:32:13 +01:00
Hans-Kristian Arntzen
ce18d4ce74 Run format_all.sh. 2017-11-17 13:38:29 +01:00
Hans-Kristian Arntzen
dd604fec9a Fix SSO for HLSL vertex shaders. 2017-11-17 13:37:28 +01:00