ObjFW  Check-in [b5c3a36731]

Overview
Comment:OFMatrix4x4: Minor cleanups
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA3-256: b5c3a36731dadb45e2ef02743b9b4786d2505e5ce10925e7ea521bbb7e3fd9ed
User & Date: js on 2023-11-02 23:00:22
Other Links: manifest | tags
Context
2023-11-03
00:16
OFMatrix4x4: SSE4.1 for -[transformVectors:count:] check-in: 7f304f573b user: js tags: trunk
2023-11-02
23:00
OFMatrix4x4: Minor cleanups check-in: b5c3a36731 user: js tags: trunk
22:48
OFMatrix4x4: Move __asm__ out of loop check-in: 0d671245d4 user: js tags: trunk
Changes
Hide Diffs Unified Diffs Ignore Whitespace Patch

Modified src/OFMatrix4x4.m from [a12ee064b0] to [740fcdd16f].

︙︙
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42

	{ 0, 0, 0, 1 }
};

@implementation OFMatrix4x4
#if (defined(OF_AMD64) || defined(OF_X86)) && defined(__GNUC__)
# ifndef __clang__
#  pragma GCC push_options
#  pragma GCC target("3dnow")
# endif
static void
multiplyWithMatrix_enhanced3DNow(OFMatrix4x4 *self, SEL _cmd,
    OFMatrix4x4 *matrix)
{
	float *left = &matrix->_values[0][0], *right = &self->_values[0][0];
	float result[4][4], *resultPtr = &result[0][0];








|








28
29
30
31
32
33
34
35
36
37
38
39
40
41
42

	{ 0, 0, 0, 1 }
};

@implementation OFMatrix4x4
#if (defined(OF_AMD64) || defined(OF_X86)) && defined(__GNUC__)
# ifndef __clang__
#  pragma GCC push_options
#  pragma GCC target("3dnow,3dnowa")
# endif
static void
multiplyWithMatrix_enhanced3DNow(OFMatrix4x4 *self, SEL _cmd,
    OFMatrix4x4 *matrix)
{
	float *left = &matrix->_values[0][0], *right = &self->_values[0][0];
	float result[4][4], *resultPtr = &result[0][0];
︙︙
72
73
74
75
76
77
78




































































79
80
81
82
83
84
85

	    "femms"
	    : "+r"(resultPtr), "+r"(left), "+r"(right)
	    :: "cx", "mm0", "mm1", "memory"
	);

	memcpy(self->_values, result, sizeof(result));
}




































































static void
multiplyWithMatrix_3DNow(OFMatrix4x4 *self, SEL _cmd, OFMatrix4x4 *matrix)
{
	float *left = &matrix->_values[0][0], *right = &self->_values[0][0];
	float result[4][4], *resultPtr = &result[0][0];

	__asm__ __volatile__ (








>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>








72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153

	    "femms"
	    : "+r"(resultPtr), "+r"(left), "+r"(right)
	    :: "cx", "mm0", "mm1", "memory"
	);

	memcpy(self->_values, result, sizeof(result));
}

static void
transformVectors_enhanced3DNow(OFMatrix4x4 *self, SEL _cmd, OFVector4D *vectors,
    size_t count)
{
	__asm__ __volatile__ (
	    "test	%0, %0\n\t"
	    "jz		0f\n"
	    "\n\t"
	    "0:\n\t"
	    "movq	(%1), %%mm0\n\t"
	    "movq	8(%1), %%mm1\n"
	    "\n\t"
	    "movq	%%mm0, %%mm2\n\t"
	    "movq	%%mm1, %%mm3\n\t"
	    "pfmul	(%2), %%mm2\n\t"
	    "pfmul	8(%2), %%mm3\n\t"
	    "pfadd	%%mm3, %%mm2\n\t"
	    "pswapd	%%mm2, %%mm3\n\t"
	    "pfadd	%%mm3, %%mm2\n"
	    "\n\t"
	    "movq	%%mm0, %%mm3\n\t"
	    "movq	%%mm1, %%mm4\n\t"
	    "pfmul	16(%2), %%mm3\n\t"
	    "pfmul	24(%2), %%mm4\n\t"
	    "pfadd	%%mm4, %%mm3\n\t"
	    "pswapd	%%mm3, %%mm4\n\t"
	    "pfadd	%%mm4, %%mm3\n"
	    "\n\t"
	    "punpckldq	%%mm3, %%mm2\n\t"
	    "movq	%%mm2, (%1)\n"
	    "\n\t"
	    "movq	%%mm0, %%mm2\n\t"
	    "movq	%%mm1, %%mm3\n\t"
	    "pfmul	32(%2), %%mm2\n\t"
	    "pfmul	40(%2), %%mm3\n\t"
	    "pfadd	%%mm3, %%mm2\n\t"
	    "pswapd	%%mm2, %%mm3\n\t"
	    "pfadd	%%mm3, %%mm2\n"
	    "\n\t"
	    "pfmul	48(%2), %%mm0\n\t"
	    "pfmul	56(%2), %%mm1\n\t"
	    "pfadd	%%mm1, %%mm0\n\t"
	    "pswapd	%%mm0, %%mm1\n\t"
	    "pfadd	%%mm1, %%mm0\n"
	    "\n\t"
	    "punpckldq	%%mm0, %%mm2\n\t"
	    "movq	%%mm2, 8(%1)\n"
	    "\n\t"
	    "add	$16, %1\n\t"
	    "dec	%0\n\t"
	    "jnz	0b\n"
	    "\n\t"
	    "0:\n\t"
	    "femms"
	    : "+r"(count), "+r"(vectors)
	    : "r"(&self->_values)
	    : "mm0", "mm1", "mm2", "mm3", "mm4", "memory"
	);
}
# ifndef __clang__
#  pragma GCC pop_options
# endif

# ifndef __clang__
#  pragma GCC push_options
#  pragma GCC target("3dnow")
# endif
static void
multiplyWithMatrix_3DNow(OFMatrix4x4 *self, SEL _cmd, OFMatrix4x4 *matrix)
{
	float *left = &matrix->_values[0][0], *right = &self->_values[0][0];
	float result[4][4], *resultPtr = &result[0][0];

	__asm__ __volatile__ (
︙︙
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190

191
192
193
194
195
196
197

	    : "+r"(resultPtr), "+r"(left), "+r"(right)
	    :: "cx", "mm0", "mm1", "memory"
	);

	memcpy(self->_values, result, sizeof(result));
}

static void
transformVectors_enhanced3DNow(OFMatrix4x4 *self, SEL _cmd, OFVector4D *vectors,
    size_t count)
{
	__asm__ __volatile__ (
	    "0:\n\t"
	    "test	%0, %0\n\t"
	    "jz		0f\n"
	    "\n\t"
	    "movq	(%1), %%mm0\n\t"
	    "movq	8(%1), %%mm1\n"
	    "\n\t"
	    "movq	%%mm0, %%mm2\n\t"
	    "movq	%%mm1, %%mm3\n\t"
	    "pfmul	(%2), %%mm2\n\t"
	    "pfmul	8(%2), %%mm3\n\t"
	    "pfadd	%%mm3, %%mm2\n\t"
	    "pswapd	%%mm2, %%mm3\n\t"
	    "pfadd	%%mm3, %%mm2\n"
	    "\n\t"
	    "movq	%%mm0, %%mm3\n\t"
	    "movq	%%mm1, %%mm4\n\t"
	    "pfmul	16(%2), %%mm3\n\t"
	    "pfmul	24(%2), %%mm4\n\t"
	    "pfadd	%%mm4, %%mm3\n\t"
	    "pswapd	%%mm3, %%mm4\n\t"
	    "pfadd	%%mm4, %%mm3\n"
	    "\n\t"
	    "punpckldq	%%mm3, %%mm2\n\t"
	    "movq	%%mm2, (%1)\n"
	    "\n\t"
	    "movq	%%mm0, %%mm2\n\t"
	    "movq	%%mm1, %%mm3\n\t"
	    "pfmul	32(%2), %%mm2\n\t"
	    "pfmul	40(%2), %%mm3\n\t"
	    "pfadd	%%mm3, %%mm2\n\t"
	    "pswapd	%%mm2, %%mm3\n\t"
	    "pfadd	%%mm3, %%mm2\n"
	    "\n\t"
	    "pfmul	48(%2), %%mm0\n\t"
	    "pfmul	56(%2), %%mm1\n\t"
	    "pfadd	%%mm1, %%mm0\n\t"
	    "pswapd	%%mm0, %%mm1\n\t"
	    "pfadd	%%mm1, %%mm0\n"
	    "\n\t"
	    "punpckldq	%%mm0, %%mm2\n\t"
	    "movq	%%mm2, 8(%1)\n"
	    "\n\t"
	    "add	$16, %1\n\t"
	    "dec	%0\n\t"
	    "jmp	0b\n"
	    "\n\t"
	    "0:\n\t"
	    "femms"
	    : "+r"(count), "+r"(vectors)
	    : "r"(&self->_values)
	    : "mm0", "mm1", "mm2", "mm3", "mm4", "memory"
	);
}

static void
transformVectors_3DNow(OFMatrix4x4 *self, SEL _cmd, OFVector4D *vectors,
    size_t count)
{
	__asm__ __volatile__ (
	    "0:\n\t"
	    "test	%0, %0\n\t"
	    "jz		0f\n"
	    "\n\t"

	    "movq	(%1), %%mm0\n\t"
	    "movq	8(%1), %%mm1\n"
	    "\n\t"
	    "movq	%%mm0, %%mm2\n\t"
	    "movq	%%mm1, %%mm3\n\t"
	    "pfmul	(%2), %%mm2\n\t"
	    "pfmul	8(%2), %%mm3\n\t"









<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<




<



>








183
184
185
186
187
188
189
190




























































191
192
193
194

195
196
197
198
199
200
201
202
203
204
205

	    : "+r"(resultPtr), "+r"(left), "+r"(right)
	    :: "cx", "mm0", "mm1", "memory"
	);

	memcpy(self->_values, result, sizeof(result));
}

static void




























































transformVectors_3DNow(OFMatrix4x4 *self, SEL _cmd, OFVector4D *vectors,
    size_t count)
{
	__asm__ __volatile__ (

	    "test	%0, %0\n\t"
	    "jz		0f\n"
	    "\n\t"
	    "0:\n\t"
	    "movq	(%1), %%mm0\n\t"
	    "movq	8(%1), %%mm1\n"
	    "\n\t"
	    "movq	%%mm0, %%mm2\n\t"
	    "movq	%%mm1, %%mm3\n\t"
	    "pfmul	(%2), %%mm2\n\t"
	    "pfmul	8(%2), %%mm3\n\t"
︙︙
229
230
231
232
233
234
235
236

237
238
239
240
241
242
243

	    "pfadd	%%mm1, %%mm0\n"
	    "\n\t"
	    "punpckldq	%%mm0, %%mm2\n\t"
	    "movq	%%mm2, 8(%1)\n"
	    "\n\t"
	    "add	$16, %1\n\t"
	    "dec	%0\n\t"
	    "jmp	0b\n"

	    "0:\n\t"
	    "femms"
	    : "+r"(count), "+r"(vectors)
	    : "r"(&self->_values)
	    : "mm0", "mm1", "mm2", "mm3", "mm4", "memory"
	);
}








|
>








237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252

	    "pfadd	%%mm1, %%mm0\n"
	    "\n\t"
	    "punpckldq	%%mm0, %%mm2\n\t"
	    "movq	%%mm2, 8(%1)\n"
	    "\n\t"
	    "add	$16, %1\n\t"
	    "dec	%0\n\t"
	    "jnz	0b\n"
	    "\n\t"
	    "0:\n\t"
	    "femms"
	    : "+r"(count), "+r"(vectors)
	    : "r"(&self->_values)
	    : "mm0", "mm1", "mm2", "mm3", "mm4", "memory"
	);
}
︙︙
253
254
255
256
257
258
259

260
261
262
263
264
265
266
267
268
269

270
271
272
273
274
275
276

		return;

# define REPLACE(selector, func)					\
	typeEncoding = method_getTypeEncoding(				\
	    class_getInstanceMethod(self, selector));			\
	class_replaceMethod(self, selector, (IMP)func, typeEncoding);


	if ([OFSystemInfo supportsEnhanced3DNow]) {
		REPLACE(@selector(multiplyWithMatrix:),
		    multiplyWithMatrix_enhanced3DNow)
		REPLACE(@selector(transformVectors:count:),
		    transformVectors_enhanced3DNow)
	} else if ([OFSystemInfo supports3DNow]) {
		REPLACE(@selector(multiplyWithMatrix:),
		    multiplyWithMatrix_3DNow)
		REPLACE(@selector(transformVectors:count:),
		    transformVectors_3DNow)

	}

# undef REPLACE
}
#endif

+ (OFMatrix4x4 *)identityMatrix








>
|
|
|
|
|
|
|
|
|
|
>








262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287

		return;

# define REPLACE(selector, func)					\
	typeEncoding = method_getTypeEncoding(				\
	    class_getInstanceMethod(self, selector));			\
	class_replaceMethod(self, selector, (IMP)func, typeEncoding);

	if ([OFSystemInfo supports3DNow]) {
		if ([OFSystemInfo supportsEnhanced3DNow]) {
			REPLACE(@selector(multiplyWithMatrix:),
			    multiplyWithMatrix_enhanced3DNow)
			REPLACE(@selector(transformVectors:count:),
			    transformVectors_enhanced3DNow)
		} else {
			REPLACE(@selector(multiplyWithMatrix:),
			    multiplyWithMatrix_3DNow)
			REPLACE(@selector(transformVectors:count:),
			    transformVectors_3DNow)
		}
	}

# undef REPLACE
}
#endif

+ (OFMatrix4x4 *)identityMatrix
︙︙