2 * Copyright © 2014 Timothy Arceri
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
18 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
24 * Timothy Arceri <t_arceri@yahoo.com.au>
28 #include "main/sse_minmax.h"
29 #include <smmintrin.h>
33 _mesa_uint_array_min_max(const unsigned *ui_indices
, unsigned *min_index
,
34 unsigned *max_index
, const unsigned count
)
37 unsigned min_ui
= ~0U;
39 unsigned aligned_count
= count
;
41 /* handle the first few values without SSE until the pointer is aligned */
42 while (((uintptr_t)ui_indices
& 15) && aligned_count
) {
43 if (*ui_indices
> max_ui
)
45 if (*ui_indices
< min_ui
)
52 /* TODO: The actual threshold for SSE begin useful may be higher than 8.
53 * Some careful microbenchmarks and measurement are required to
54 * find the actual tipping point.
56 if (aligned_count
>= 8) {
57 unsigned max_arr
[4] __attribute__ ((aligned (16)));
58 unsigned min_arr
[4] __attribute__ ((aligned (16)));
60 __m128i max_ui4
= _mm_setzero_si128();
61 __m128i min_ui4
= _mm_set1_epi32(~0U);
63 __m128i
*ui_indices_ptr
;
65 vec_count
= aligned_count
& ~0x3;
66 ui_indices_ptr
= (__m128i
*)ui_indices
;
67 for (i
= 0; i
< vec_count
/ 4; i
++) {
68 ui_indices4
= _mm_load_si128(&ui_indices_ptr
[i
]);
69 max_ui4
= _mm_max_epu32(ui_indices4
, max_ui4
);
70 min_ui4
= _mm_min_epu32(ui_indices4
, min_ui4
);
73 _mm_store_si128((__m128i
*)max_arr
, max_ui4
);
74 _mm_store_si128((__m128i
*)min_arr
, min_ui4
);
76 for (i
= 0; i
< 4; i
++) {
77 if (max_arr
[i
] > max_ui
)
79 if (min_arr
[i
] < min_ui
)
85 for (; i
< aligned_count
; i
++) {
86 if (ui_indices
[i
] > max_ui
)
87 max_ui
= ui_indices
[i
];
88 if (ui_indices
[i
] < min_ui
)
89 min_ui
= ui_indices
[i
];