wine/vkd3d
0
0
Fork 0
mirror of https://gitlab.winehq.org/wine/vkd3d.git synced 2024-11-21 16:46:41 -08:00
Code Issues Packages Projects Releases Wiki Activity

vkd3d-shader/hlsl: Implement atan and atan2.

Browse Source 
Also narrows some more todos on the tests.
This commit is contained in:
Petrichor Park 2024-02-11 11:06:24 -06:00 committed by Alexandre Julliard
parent 69294c290b
commit 02a3667822
Notes: Alexandre Julliard 2024-02-21 23:33:43 +01:00 
Approved-by: Giovanni Mascellani (@giomasce)
Approved-by: Zebediah Figura (@zfigura)
Approved-by: Henri Verbeet (@hverbeet)
Approved-by: Alexandre Julliard (@julliard)
Merge-Request: https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/364
2 changed files with 117 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

95
libs/vkd3d-shader/hlsl.y
View File

@ -2798,6 +2798,99 @@ static bool intrinsic_asin(struct hlsl_ctx *ctx,
return write_acos_or_asin(ctx, params, loc, true); return write_acos_or_asin(ctx, params, loc, true);
} }
static bool write_atan_or_atan2(struct hlsl_ctx *ctx,
const struct parse_initializer *params,
const struct vkd3d_shader_location *loc, bool atan2_mode)
{
struct hlsl_ir_function_decl *func;
struct hlsl_type *type;
struct vkd3d_string_buffer *buf;
int ret;
static const char atan2_name[] = "atan2";
static const char atan_name[] = "atan";
static const char atan2_header_template[] =
"%s atan2(%s y, %s x)\n"
"{\n"
" %s in_y, in_x;\n"
" in_y = y;\n"
" in_x = x;\n";
static const char atan_header_template[] =
"%s atan(%s y)\n"
"{\n"
" %s in_y, in_x;\n"
" in_y = y;\n"
" in_x = 1.0;\n";
static const char body_template[] =
" %s recip, input, x2, poly_approx, flipped;"
" recip = 1.0 / max(abs(in_y), abs(in_x));\n"
" input = recip * min(abs(in_y), abs(in_x));\n"
" x2 = input * input;\n"
" poly_approx = ((((0.020835\n"
" * x2 - 0.085133)\n"
" * x2 + 0.180141)\n"
" * x2 - 0.330299)\n"
" * x2 + 0.999866)\n"
" * input;\n"
" flipped = poly_approx * -2.0 + 1.570796;\n"
" poly_approx += abs(in_x) < abs(in_y) ? flipped : 0.0;\n"
" poly_approx += in_x < 0.0 ? -3.1415927 : 0.0;\n"
" return (min(in_x, in_y) < 0.0 && max(in_x, in_y) >= 0.0)\n"
" ? -poly_approx\n"
" : poly_approx;\n"
"}";
if (!(type = elementwise_intrinsic_get_common_type(ctx, params, loc)))
return false;
type = hlsl_get_numeric_type(ctx, type->class, HLSL_TYPE_FLOAT, type->dimx, type->dimy);
if (!(buf = hlsl_get_string_buffer(ctx)))
return false;
if (atan2_mode)
ret = vkd3d_string_buffer_printf(buf, atan2_header_template,
type->name, type->name, type->name, type->name);
else
ret = vkd3d_string_buffer_printf(buf, atan_header_template,
type->name, type->name, type->name);
if (ret < 0)
{
vkd3d_string_buffer_cleanup(buf);
return false;
}
ret = vkd3d_string_buffer_printf(buf, body_template, type->name);
if (ret < 0)
{
vkd3d_string_buffer_cleanup(buf);
return false;
}
func = hlsl_compile_internal_function(ctx,
atan2_mode ? atan2_name : atan_name, buf->buffer);
vkd3d_string_buffer_cleanup(buf);
if (!func)
return false;
return add_user_call(ctx, func, params, loc);
}
static bool intrinsic_atan(struct hlsl_ctx *ctx,
const struct parse_initializer *params, const struct vkd3d_shader_location *loc)
{
return write_atan_or_atan2(ctx, params, loc, false);
}
static bool intrinsic_atan2(struct hlsl_ctx *ctx,
const struct parse_initializer *params, const struct vkd3d_shader_location *loc)
{
return write_atan_or_atan2(ctx, params, loc, true);
}
/* Find the type corresponding to the given source type, with the same /* Find the type corresponding to the given source type, with the same
* dimensions but a different base type. */ * dimensions but a different base type. */
static struct hlsl_type *convert_numeric_type(const struct hlsl_ctx *ctx, static struct hlsl_type *convert_numeric_type(const struct hlsl_ctx *ctx,
@ -4031,6 +4124,8 @@ intrinsic_functions[] =
{"asfloat", 1, true, intrinsic_asfloat}, {"asfloat", 1, true, intrinsic_asfloat},
{"asin", 1, true, intrinsic_asin}, {"asin", 1, true, intrinsic_asin},
{"asuint", -1, true, intrinsic_asuint}, {"asuint", -1, true, intrinsic_asuint},
{"atan", 1, true, intrinsic_atan},
{"atan2", 2, true, intrinsic_atan2},
{"ceil", 1, true, intrinsic_ceil}, {"ceil", 1, true, intrinsic_ceil},
{"clamp", 3, true, intrinsic_clamp}, {"clamp", 3, true, intrinsic_clamp},
{"clip", 1, true, intrinsic_clip}, {"clip", 1, true, intrinsic_clip},

44
tests/hlsl/inverse-trig.shader_test
View File

@ -92,7 +92,7 @@ todo(sm<4) draw quad
probe all rgba (31416.0, 0.0, 0.0, 0.0) probe all rgba (31416.0, 0.0, 0.0, 0.0)
[pixel shader todo] [pixel shader todo(sm<4)]
uniform float4 a; uniform float4 a;
float4 main() : sv_target float4 main() : sv_target
@ -102,26 +102,26 @@ float4 main() : sv_target
[test] [test]
uniform 0 float4 -1.0 0.0 0.0 0.0 uniform 0 float4 -1.0 0.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (-0.785409629, 0.0, 0.0, 0.0) 512 probe all rgba (-0.785409629, 0.0, 0.0, 0.0) 512
uniform 0 float4 -0.5 0.0 0.0 0.0 uniform 0 float4 -0.5 0.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (-0.4636476, 0.0, 0.0, 0.0) 256 probe all rgba (-0.4636476, 0.0, 0.0, 0.0) 256
uniform 0 float4 0.0 0.0 0.0 0.0 uniform 0 float4 0.0 0.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (0.0, 0.0, 0.0, 0.0) 256 probe all rgba (0.0, 0.0, 0.0, 0.0) 256
uniform 0 float4 0.5 0.0 0.0 0.0 uniform 0 float4 0.5 0.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (0.4636476, 0.0, 0.0, 0.0) 256 probe all rgba (0.4636476, 0.0, 0.0, 0.0) 256
uniform 0 float4 1.0 0.0 0.0 0.0 uniform 0 float4 1.0 0.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (0.785409629, 0.0, 0.0, 0.0) 512 probe all rgba (0.785409629, 0.0, 0.0, 0.0) 512
[pixel shader todo] [pixel shader todo(sm<4)]
uniform float4 a; uniform float4 a;
float4 main() : sv_target float4 main() : sv_target
@ -133,64 +133,64 @@ float4 main() : sv_target
[test] [test]
% Non-degenerate cases % Non-degenerate cases
uniform 0 float4 1.0 1.0 0.0 0.0 uniform 0 float4 1.0 1.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (0.785385, 0.0, 0.0, 0.0) 512 probe all rgba (0.785385, 0.0, 0.0, 0.0) 512
uniform 0 float4 5.0 -5.0 0.0 0.0 uniform 0 float4 5.0 -5.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (2.356194, 0.0, 0.0, 0.0) 256 probe all rgba (2.356194, 0.0, 0.0, 0.0) 256
uniform 0 float4 -3.0 -3.0 0.0 0.0 uniform 0 float4 -3.0 -3.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (-2.356194, 0.0, 0.0, 0.0) 256 probe all rgba (-2.356194, 0.0, 0.0, 0.0) 256
uniform 0 float4 1.0 0.0 0.0 0.0 uniform 0 float4 1.0 0.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (1.570796, 0.0, 0.0, 0.0) 256 probe all rgba (1.570796, 0.0, 0.0, 0.0) 256
uniform 0 float4 -1.0 0.0 0.0 0.0 uniform 0 float4 -1.0 0.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (-1.570796, 0.0, 0.0, 0.0) 256 probe all rgba (-1.570796, 0.0, 0.0, 0.0) 256
uniform 0 float4 0.0 1.0 0.0 0.0 uniform 0 float4 0.0 1.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (0.0, 0.0, 0.0, 0.0) 256 probe all rgba (0.0, 0.0, 0.0, 0.0) 256
uniform 0 float4 0.0 -1.0 0.0 0.0 uniform 0 float4 0.0 -1.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (3.1415927, 0.0, 0.0, 0.0) 256 probe all rgba (3.1415927, 0.0, 0.0, 0.0) 256
% Degenerate cases % Degenerate cases
uniform 0 float4 0.00001 0.00002 0.0 0.0 uniform 0 float4 0.00001 0.00002 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (0.463647, 0.0, 0.0, 0.0) 256 probe all rgba (0.463647, 0.0, 0.0, 0.0) 256
uniform 0 float4 0.00001 -0.00002 0.0 0.0 uniform 0 float4 0.00001 -0.00002 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (2.677945, 0.0, 0.0, 0.0) 256 probe all rgba (2.677945, 0.0, 0.0, 0.0) 256
uniform 0 float4 -0.00001 100000.0 0.0 0.0 uniform 0 float4 -0.00001 100000.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (-0.000000000099986595, 0.0, 0.0, 0.0) 2048 probe all rgba (-0.000000000099986595, 0.0, 0.0, 0.0) 2048
uniform 0 float4 10000000.0 0.00000001 0.0 0.0 uniform 0 float4 10000000.0 0.00000001 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (1.570796, 0.0, 0.0, 0.0) 256 probe all rgba (1.570796, 0.0, 0.0, 0.0) 256
% Negative zero behavior should be to treat it the % Negative zero behavior should be to treat it the
% same as normal zero. % same as normal zero.
uniform 0 float4 1000000000.0 0.0 0.0 0.0 uniform 0 float4 1000000000.0 0.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (1.570796, 0.0, 0.0, 0.0) 256 probe all rgba (1.570796, 0.0, 0.0, 0.0) 256
uniform 0 float4 1000000000.0 -0.0 0.0 0.0 uniform 0 float4 1000000000.0 -0.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (1.570796, 0.0, 0.0, 0.0) 256 probe all rgba (1.570796, 0.0, 0.0, 0.0) 256
uniform 0 float4 0.0 -1.0 0.0 0.0 uniform 0 float4 0.0 -1.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (3.1415927, 0.0, 0.0, 0.0) 256 probe all rgba (3.1415927, 0.0, 0.0, 0.0) 256
uniform 0 float4 -0.0 -1.0 0.0 0.0 uniform 0 float4 -0.0 -1.0 0.0 0.0
todo(sm<6) draw quad todo(sm<4) draw quad
probe all rgba (3.1415927, 0.0, 0.0, 0.0) 256 probe all rgba (3.1415927, 0.0, 0.0, 0.0) 256