e4cb3b5081
wined3d: Recognize the SM4 or opcode.
2015-02-11 23:05:17 +09:00
3225fea30c
wined3d: Recognize the SM4 ne opcode.
2015-02-11 23:05:13 +09:00
8506da62e0
wined3d: Implement WINED3DSIH_SQRT in the GLSL shader backend.
2015-02-11 23:05:08 +09:00
74b4516bfa
wined3d: Only consider POSITION0 PS inputs as vpos.
2015-02-09 22:05:41 +09:00
96b48191c0
wined3d: Only copy POSITION0 output semantics into gl_Position.
2015-02-09 22:05:35 +09:00
d360a51f3c
wined3d: Use the core version of the shader object functions.
2015-01-23 12:02:37 +01:00
2912e2938e
wined3d: Handle the SM4 SV_POSITION pixel shader input semantic in the GLSL shader backend.
...
This is (mostly) equivalent to the vPos register in SM3.
2015-01-19 14:31:38 +01:00
2e5abc156a
wined3d: Implement WINED3DSIH_SAMPLE in the GLSL shader backend.
2015-01-19 14:31:32 +01:00
60d44af34e
wined3d: Merge shader_glsl_load_vsamplers() and shader_glsl_load_psamplers().
2015-01-15 17:24:45 +01:00
c6232e1d11
wined3d: Create GL sampler objects for wined3d sampler objects.
2015-01-15 17:24:43 +01:00
5c3d52468c
wined3d: Take the resource data type into account in shader_glsl_gen_sample_code().
2014-12-03 11:14:27 +01:00
3aa9fe6bef
wined3d: Record the data type of shader resources.
2014-12-03 11:14:25 +01:00
f5cef43738
wined3d: Store resource types instead of sampler types in struct wined3d_shader_reg_maps.
...
The difference doesn't really matter for SM1-3 since resources and samplers
are always tied together, but in SM4 they're separate.
2014-12-03 11:14:21 +01:00
c9b10beb36
wined3d: Vertex fog uses the absolute eye position z.
2014-11-27 13:27:13 +01:00
8c18ebf66a
wined3d: Recognize the SM4 uge opcode.
2014-11-17 19:30:37 +09:00
8681999eee
wined3d: Explicitly convert the condition to bool shader_glsl_if().
...
In SM4 shaders the condition is likely to be an integer instead of a bool
uniform like in SM3.
2014-11-17 19:30:35 +09:00
e789fadd56
wined3d: Don't replicate shader limits values for each shader.
2014-10-16 21:23:26 +09:00
33627f02aa
wined3d: Store bool uniforms location in the GLSL shader structures.
2014-08-28 15:30:26 +02:00
b005ad6f90
wined3d: Use rendertarget views for color output instead of surfaces.
2014-08-21 12:24:42 +02:00
80bca9bc9c
wined3d: Add constant buffer support to the GLSL shader backend.
2014-08-04 14:23:29 +02:00
fa312a32b2
wined3d: Add format information for WINED3DFMT_R8_UNORM.
2014-06-13 10:11:59 +02:00
7b8aa8606c
wined3d: Recognize the SM4 ishl opcode.
2014-05-20 09:44:25 +02:00
3555eab8b9
wined3d: Recognize the SM4 dp2 opcode.
2014-05-20 09:44:21 +02:00
c81d3b7336
wined3d: Recognize the SM4 discard opcode.
2014-05-20 09:44:17 +02:00
0eebe0515f
wined3d: Fix a couple of string declarations.
2014-05-02 13:45:10 +02:00
14e3f71c75
wined3d: Bump ps_1x_max_value.
...
Shader model 4 hardware should generally have real floats for everything, and
FLT_MAX matches the value from the Windows drivers for that kind of hardware.
OpenGL requires at least 2^10 for colors, and 2^32 for other floats.
2014-05-02 13:45:04 +02:00
23eed88fde
wined3d: Implement support for per-stage texture stage constants.
...
This patch is loosely based on an earlier patch by Christian Costa.
2014-04-23 15:00:43 +02:00
91c5cf33ad
wined3d: Avoid wined3d_ftoa buffer overflow.
2014-01-14 12:31:37 +01:00
d412eb568e
wined3d: Introduce a common GLSL handler for scalar instructions.
2014-01-09 15:50:30 +01:00
27dc41d0dc
wined3d: Fix the source swizzle for WINED3DSIH_LOG and WINED3DSIH_LOGP.
2013-12-12 12:08:09 +01:00
c8852c3ee3
wined3d: Fix the source swizzle for WINED3DSIH_EXPP.
2013-12-11 15:15:01 +01:00
6e9acca1f4
wined3d: Fix the source swizzle for WINED3DSIH_EXP.
2013-12-11 15:14:58 +01:00
c228ffbaf0
wined3d: Unify shader binding points.
2013-10-10 10:44:53 +02:00
079b6d4c11
wined3d: Unify the shader state IDs.
2013-10-09 11:11:07 +02:00
112617f00b
wined3d: Move tex_unit_map and friends into the context.
2013-09-16 13:05:58 +02:00
d8c43aabf7
wined3d: Move the decoded stream info into the context.
2013-09-12 17:48:14 +02:00
b4d8b52fa7
wined3d: Pass position_transformed and gl_info to find_ps_compile_args.
2013-09-11 17:09:59 +02:00
dfeee9077c
wined3d: Pass the swizzle map to find_vs_compile_args.
2013-09-11 17:09:56 +02:00
137590be31
wined3d: Return 1 instead of 0 for max_vertex_blend_matrices in glsl_vertex_pipe_vp_get_caps().
...
It's not necessarily clear to me that one is better than the other here, but
this is what we returned in the old GL fixed function implementation of the
D3D fixed function vertex pipe, and apparently EverQuest depends on this.
This fixes a regression introduced by commit 2014141a25
2013-09-06 12:00:09 +02:00
b73464a413
wined3d: Use the ftoa helper function in the ARB shader backend.
...
This is the ARB equivalent to e0494afa00
2013-09-05 17:55:48 +02:00
99092c2700
wined3d: Avoid a couple of useless glGetUniformLocationARB() calls.
2013-09-04 11:36:58 +02:00
ff6ce4c669
wined3d: Use clamp() instead of max() for calculating the clamped cosine in the ffp lighting calculations.
...
On typical hardware, the clamp() would be an instruction modifier, while the
max() would need an extra instruction.
2013-09-04 11:36:50 +02:00
0d2d847914
wined3d: Don't clamp texture lookups in the GLSL fixed function fragment pipe.
2013-09-03 16:58:17 +02:00
b85c2a74bb
wined3d: Replace the "stateBlock" wined3d_device field with a wined3d_state structure.
2013-08-27 11:50:53 +02:00
8b8e081d5f
wined3d: Keep track of contained indices in struct constant_heap itself.
2013-08-23 10:41:32 +02:00
74faebd68d
wined3d: Store the current GLSL program per-context.
2013-08-23 10:41:28 +02:00
69e1100c0c
wined3d: Handle NP2 fixup updates in shader_load_constants().
2013-08-21 12:13:56 +02:00
62859daf0b
wined3d: Avoid looking up shaders for shader stages that didn't change.
2013-08-21 12:13:56 +02:00
ad147d638c
wined3d: Just update a range of constants in walk_constant_heap().
...
The call overhead from separate calls typically ends up being more expensive
than just updating a few more constants.
2013-08-16 11:55:24 +02:00
37d1e8b79c
wined3d: Try to avoid redundant constant updates.
...
This gives a minor performance improvement. For example, in 3DMark03 Game Test
1 this gives about a 4% improvement, in the Counter Strike: Source stress test
it's about 1%. (NVIDIA GF9600M, Intel T9550, 1024x768).
2013-08-16 11:55:24 +02:00
Henri Verbeet