ShuriZma
/
xemu
mirror of https://github.com/xemu-project/xemu.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #50 from JayFoxRox/fix-cmp 

Fix CMP vertex attrib type
This commit is contained in:
espes 2015-09-02 17:33:17 +10:00
parent b42d8c5344 1ab319cacf
commit bb981ab2bb
2 changed files with 14 additions and 242 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
hw/xbox/nv2a.c
View File

@ -32,7 +32,6 @@
#include "hw/xbox/g-lru-cache.h"
#include "hw/xbox/swizzle.h"
#include "hw/xbox/u_format_r11g11b10f.h"
#include "hw/xbox/nv2a_shaders.h"
#include "hw/xbox/nv2a_debug.h"
@ -2051,12 +2050,20 @@ static void pgraph_bind_vertex_attributes(NV2AState *d,
uint8_t *out = attribute->converted_buffer + j * out_stride;
switch (attribute->format) {
case NV097_SET_VERTEX_DATA_ARRAY_FORMAT_TYPE_CMP:
r11g11b10f_to_float3(ldl_le_p((uint32_t*)in),
(float*)out);
case NV097_SET_VERTEX_DATA_ARRAY_FORMAT_TYPE_CMP: {
uint32_t p = ldl_le_p((uint32_t*)in);
float *xyz = (float*)out;
xyz[0] = ((int32_t)(((p >> 0) & 0x7FF) << 21) >> 21)
/ 1023.0f;
xyz[1] = ((int32_t)(((p >> 11) & 0x7FF) << 21) >> 21)
/ 1023.0f;
xyz[2] = ((int32_t)(((p >> 22) & 0x3FF) << 22) >> 22)
/ 511.0f;
break;
}
default:
assert(false);
break;
}
}
@ -2067,10 +2074,9 @@ static void pgraph_bind_vertex_attributes(NV2AState *d,
num_elements * out_stride,
attribute->converted_buffer,
GL_DYNAMIC_DRAW);
attribute->converted_elements = num_elements;
}
attribute->converted_elements = num_elements;
glVertexAttribPointer(i,
attribute->converted_count,
@ -4947,14 +4953,12 @@ static void pgraph_method(NV2AState *d,
vertex_attribute->needs_conversion = false;
break;
case NV097_SET_VERTEX_DATA_ARRAY_FORMAT_TYPE_CMP:
/* "3 signed, normalized components packed in 32-bits. (11,11,10)"
* TODO: This could use ARB_vertex_type_10f_11f_11f_rev where
* available. */
/* 3 signed, normalized components packed in 32-bits. (11,11,10) */
vertex_attribute->size = 4;
vertex_attribute->gl_type = GL_FLOAT;
vertex_attribute->gl_normalize = GL_FALSE;
vertex_attribute->needs_conversion = true;
vertex_attribute->converted_size = 4;
vertex_attribute->converted_size = sizeof(float);
vertex_attribute->converted_count = 3 * vertex_attribute->count;
break;
default:

232
hw/xbox/u_format_r11g11b10f.h
View File

@ -1,232 +0,0 @@
/*
* Copyright (C) 2011 Marek Olšák <maraeo@gmail.com>
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
* DEALINGS IN THE SOFTWARE.
*/
/* Based on code from The OpenGL Programming Guide / 7th Edition, Appendix J.
* Available here: http://www.opengl-redbook.com/appendices/
* The algorithm in the book contains a bug though, which is fixed in the code
* below.
*/
#define UF11(e, m) ((e << 6) | (m))
#define UF11_EXPONENT_BIAS 15
#define UF11_EXPONENT_BITS 0x1F
#define UF11_EXPONENT_SHIFT 6
#define UF11_MANTISSA_BITS 0x3F
#define UF11_MANTISSA_SHIFT (23 - UF11_EXPONENT_SHIFT)
#define UF11_MAX_EXPONENT (UF11_EXPONENT_BITS << UF11_EXPONENT_SHIFT)
#define UF10(e, m) ((e << 5) | (m))
#define UF10_EXPONENT_BIAS 15
#define UF10_EXPONENT_BITS 0x1F
#define UF10_EXPONENT_SHIFT 5
#define UF10_MANTISSA_BITS 0x1F
#define UF10_MANTISSA_SHIFT (23 - UF10_EXPONENT_SHIFT)
#define UF10_MAX_EXPONENT (UF10_EXPONENT_BITS << UF10_EXPONENT_SHIFT)
#define F32_INFINITY 0x7f800000
static inline unsigned f32_to_uf11(float val)
{
union {
float f;
uint32_t ui;
} f32 = {val};
uint16_t uf11 = 0;
/* Decode little-endian 32-bit floating-point value */
int sign = (f32.ui >> 16) & 0x8000;
/* Map exponent to the range [-127,128] */
int exponent = ((f32.ui >> 23) & 0xff) - 127;
int mantissa = f32.ui & 0x007fffff;
if (exponent == 128) { /* Infinity or NaN */
/* From the GL_EXT_packed_float spec:
*
* "Additionally: negative infinity is converted to zero; positive
* infinity is converted to positive infinity; and both positive and
* negative NaN are converted to positive NaN."
*/
uf11 = UF11_MAX_EXPONENT;
if (mantissa) {
uf11 |= 1; /* NaN */
} else {
if (sign)
uf11 = 0; /* 0.0 */
}
} else if (sign) {
return 0;
} else if (val > 65024.0f) {
/* From the GL_EXT_packed_float spec:
*
* "Likewise, finite positive values greater than 65024 (the maximum
* finite representable unsigned 11-bit floating-point value) are
* converted to 65024."
*/
uf11 = UF11(30, 63);
}
else if (exponent > -15) { /* Representable value */
exponent += UF11_EXPONENT_BIAS;
mantissa >>= UF11_MANTISSA_SHIFT;
uf11 = exponent << UF11_EXPONENT_SHIFT | mantissa;
}
return uf11;
}
static inline float uf11_to_f32(uint16_t val)
{
union {
float f;
uint32_t ui;
} f32;
int exponent = (val & 0x07c0) >> UF11_EXPONENT_SHIFT;
int mantissa = (val & 0x003f);
f32.f = 0.0;
if (exponent == 0) {
if (mantissa != 0) {
const float scale = 1.0 / (1 << 20);
f32.f = scale * mantissa;
}
}
else if (exponent == 31) {
f32.ui = F32_INFINITY | mantissa;
}
else {
float scale, decimal;
exponent -= 15;
if (exponent < 0) {
scale = 1.0f / (1 << -exponent);
}
else {
scale = (float) (1 << exponent);
}
decimal = 1.0f + (float) mantissa / 64;
f32.f = scale * decimal;
}
return f32.f;
}
static inline unsigned f32_to_uf10(float val)
{
union {
float f;
uint32_t ui;
} f32 = {val};
uint16_t uf10 = 0;
/* Decode little-endian 32-bit floating-point value */
int sign = (f32.ui >> 16) & 0x8000;
/* Map exponent to the range [-127,128] */
int exponent = ((f32.ui >> 23) & 0xff) - 127;
int mantissa = f32.ui & 0x007fffff;
if (exponent == 128) {
/* From the GL_EXT_packed_float spec:
*
* "Additionally: negative infinity is converted to zero; positive
* infinity is converted to positive infinity; and both positive and
* negative NaN are converted to positive NaN."
*/
uf10 = UF10_MAX_EXPONENT;
if (mantissa) {
uf10 |= 1; /* NaN */
} else {
if (sign)
uf10 = 0; /* 0.0 */
}
} else if (sign) {
return 0;
} else if (val > 64512.0f) { /* Overflow - flush to Infinity */
/* From the GL_EXT_packed_float spec:
*
* "Likewise, finite positive values greater than 64512 (the maximum
* finite representable unsigned 10-bit floating-point value) are
* converted to 64512."
*/
uf10 = UF10(30, 31);
}
else if (exponent > -15) { /* Representable value */
exponent += UF10_EXPONENT_BIAS;
mantissa >>= UF10_MANTISSA_SHIFT;
uf10 = exponent << UF10_EXPONENT_SHIFT | mantissa;
}
return uf10;
}
static inline float uf10_to_f32(uint16_t val)
{
union {
float f;
uint32_t ui;
} f32;
int exponent = (val & 0x03e0) >> UF10_EXPONENT_SHIFT;
int mantissa = (val & 0x001f);
f32.f = 0.0;
if (exponent == 0) {
if (mantissa != 0) {
const float scale = 1.0 / (1 << 20);
f32.f = scale * mantissa;
}
}
else if (exponent == 31) {
f32.ui = F32_INFINITY | mantissa;
}
else {
float scale, decimal;
exponent -= 15;
if (exponent < 0) {
scale = 1.0f / (1 << -exponent);
}
else {
scale = (float) (1 << exponent);
}
decimal = 1.0f + (float) mantissa / 32;
f32.f = scale * decimal;
}
return f32.f;
}
static inline unsigned float3_to_r11g11b10f(const float rgb[3])
{
return ( f32_to_uf11(rgb[0]) & 0x7ff) |
((f32_to_uf11(rgb[1]) & 0x7ff) << 11) |
((f32_to_uf10(rgb[2]) & 0x3ff) << 22);
}
static inline void r11g11b10f_to_float3(unsigned rgb, float retval[3])
{
retval[0] = uf11_to_f32( rgb & 0x7ff);
retval[1] = uf11_to_f32((rgb >> 11) & 0x7ff);
retval[2] = uf10_to_f32((rgb >> 22) & 0x3ff);
}