1 /**************************************************************************
3 * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
6 * Permission is hereby granted, free of charge, to any person obtaining a
7 * copy of this software and associated documentation files (the
8 * "Software"), to deal in the Software without restriction, including
9 * without limitation the rights to use, copy, modify, merge, publish,
10 * distribute, sub license, and/or sell copies of the Software, and to
11 * permit persons to whom the Software is furnished to do so, subject to
12 * the following conditions:
14 * The above copyright notice and this permission notice (including the
15 * next paragraph) shall be included in all copies or substantial portions
18 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
19 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
20 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
21 * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
22 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
23 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
24 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
26 **************************************************************************/
29 #include "pipe/p_util.h"
30 #include "pipe/p_shader_tokens.h"
31 #include "tgsi/util/tgsi_parse.h"
32 #include "tgsi/util/tgsi_util.h"
33 #include "tgsi/exec/tgsi_exec.h"
35 #include "draw_vs_aos.h"
36 #include "draw_vertex.h"
38 #include "rtasm/rtasm_x86sse.h"
41 #define X87_CW_EXCEPTION_INV_OP (1<<0)
42 #define X87_CW_EXCEPTION_DENORM_OP (1<<1)
43 #define X87_CW_EXCEPTION_ZERO_DIVIDE (1<<2)
44 #define X87_CW_EXCEPTION_OVERFLOW (1<<3)
45 #define X87_CW_EXCEPTION_UNDERFLOW (1<<4)
46 #define X87_CW_EXCEPTION_PRECISION (1<<5)
47 #define X87_CW_PRECISION_SINGLE (0<<8)
48 #define X87_CW_PRECISION_RESERVED (1<<8)
49 #define X87_CW_PRECISION_DOUBLE (2<<8)
50 #define X87_CW_PRECISION_DOUBLE_EXT (3<<8)
51 #define X87_CW_PRECISION_MASK (3<<8)
52 #define X87_CW_ROUND_NEAREST (0<<10)
53 #define X87_CW_ROUND_DOWN (1<<10)
54 #define X87_CW_ROUND_UP (2<<10)
55 #define X87_CW_ROUND_ZERO (3<<10)
56 #define X87_CW_ROUND_MASK (3<<10)
57 #define X87_CW_INFINITY (1<<12)
60 PIPE_CDECL
void aos_do_lit( struct aos_machine
*machine
,
76 const float epsilon
= 1.0F
/ 256.0F
;
77 float exponent
= CLAMP(in
[3], -(128.0F
- epsilon
), (128.0F
- epsilon
));
80 result
[2] = powf(in
[1], exponent
);
94 static PIPE_CDECL
void do_lit_lut( struct aos_machine
*machine
,
110 if (machine
->lit_info
[count
].shine_tab
->exponent
!= in
[3]) {
111 machine
->lit_info
[count
].func
= aos_do_lit
;
117 const float *tab
= machine
->lit_info
[count
].shine_tab
->values
;
118 float f
= in
[1] * 256;
120 float frac
= f
- (float)k
;
124 result
[2] = tab
[k
] + frac
*(tab
[k
+1]-tab
[k
]);
131 const float epsilon
= 1.0F
/ 256.0F
;
132 float exponent
= CLAMP(in
[3], -(128.0F
- epsilon
), (128.0F
- epsilon
));
135 result
[2] = powf(in
[1], exponent
);
149 static void do_populate_lut( struct shine_tab
*tab
,
150 float unclamped_exponent
)
152 const float epsilon
= 1.0F
/ 256.0F
;
153 float exponent
= CLAMP(unclamped_exponent
, -(128.0F
- epsilon
), (128.0F
- epsilon
));
156 tab
->exponent
= unclamped_exponent
; /* for later comparison */
160 for (i
= 1; i
< 258; i
++) {
161 tab
->values
[i
] = 1.0;
165 for (i
= 1; i
< 258; i
++) {
166 tab
->values
[i
] = powf((float)i
* epsilon
, exponent
);
174 static void PIPE_CDECL
populate_lut( struct aos_machine
*machine
,
181 /* Search for an existing table for this value. Note that without
182 * static analysis we don't really know if in[3] will be constant,
183 * but it usually is...
185 for (tab
= 0; tab
< 4; tab
++) {
186 if (machine
->shine_tab
[tab
].exponent
== in
[3]) {
191 for (tab
= 0, i
= 1; i
< 4; i
++) {
192 if (machine
->shine_tab
[i
].last_used
< machine
->shine_tab
[tab
].last_used
)
196 if (machine
->shine_tab
[tab
].last_used
== machine
->now
) {
197 /* No unused tables (this is not a ffvertex program...). Just
198 * call pow each time:
200 machine
->lit_info
[count
].func
= aos_do_lit
;
201 machine
->lit_info
[count
].func( machine
, result
, in
, count
);
205 do_populate_lut( &machine
->shine_tab
[tab
], in
[3] );
209 machine
->shine_tab
[tab
].last_used
= machine
->now
;
210 machine
->lit_info
[count
].shine_tab
= &machine
->shine_tab
[tab
];
211 machine
->lit_info
[count
].func
= do_lit_lut
;
212 machine
->lit_info
[count
].func( machine
, result
, in
, count
);
216 void draw_vs_aos_machine_constants( struct aos_machine
*machine
,
217 const float (*constants
)[4] )
219 machine
->constants
= constants
;
223 for (i
= 0; i
< MAX_LIT_INFO
; i
++) {
224 machine
->lit_info
[i
].func
= populate_lut
;
231 void draw_vs_aos_machine_viewport( struct aos_machine
*machine
,
232 const struct pipe_viewport_state
*viewport
)
234 memcpy(machine
->scale
, viewport
->scale
, 4 * sizeof(float));
235 memcpy(machine
->translate
, viewport
->translate
, 4 * sizeof(float));
240 void draw_vs_aos_machine_destroy( struct aos_machine
*machine
)
245 struct aos_machine
*draw_vs_aos_machine( void )
247 struct aos_machine
*machine
;
249 float inv
= 1.0f
/255.0f
;
252 machine
= align_malloc(sizeof(struct aos_machine
), 16);
256 memset(machine
, 0, sizeof(*machine
));
258 ASSIGN_4V(machine
->internal
[IMM_SWZ
], 1.0f
, -1.0f
, 0.0f
, 1.0f
);
259 *(unsigned *)&machine
->internal
[IMM_SWZ
][3] = 0xffffffff;
261 ASSIGN_4V(machine
->internal
[IMM_ONES
], 1.0f
, 1.0f
, 1.0f
, 1.0f
);
262 ASSIGN_4V(machine
->internal
[IMM_NEGS
], -1.0f
, -1.0f
, -1.0f
, -1.0f
);
263 ASSIGN_4V(machine
->internal
[IMM_IDENTITY
], 0.0f
, 0.0f
, 0.0f
, 1.0f
);
264 ASSIGN_4V(machine
->internal
[IMM_INV_255
], inv
, inv
, inv
, inv
);
265 ASSIGN_4V(machine
->internal
[IMM_255
], f255
, f255
, f255
, f255
);
266 ASSIGN_4V(machine
->internal
[IMM_RSQ
], -.5f
, 1.5f
, 0.0f
, 0.0f
);
269 machine
->fpu_rnd_nearest
= (X87_CW_EXCEPTION_INV_OP
|
270 X87_CW_EXCEPTION_DENORM_OP
|
271 X87_CW_EXCEPTION_ZERO_DIVIDE
|
272 X87_CW_EXCEPTION_OVERFLOW
|
273 X87_CW_EXCEPTION_UNDERFLOW
|
274 X87_CW_EXCEPTION_PRECISION
|
276 X87_CW_ROUND_NEAREST
|
277 X87_CW_PRECISION_DOUBLE_EXT
);
279 assert(machine
->fpu_rnd_nearest
== 0x37f);
281 machine
->fpu_rnd_neg_inf
= (X87_CW_EXCEPTION_INV_OP
|
282 X87_CW_EXCEPTION_DENORM_OP
|
283 X87_CW_EXCEPTION_ZERO_DIVIDE
|
284 X87_CW_EXCEPTION_OVERFLOW
|
285 X87_CW_EXCEPTION_UNDERFLOW
|
286 X87_CW_EXCEPTION_PRECISION
|
289 X87_CW_PRECISION_DOUBLE_EXT
);
291 for (i
= 0; i
< MAX_SHINE_TAB
; i
++)
292 do_populate_lut( &machine
->shine_tab
[i
], 1.0f
);