1 /*
2 * Copyright © 2015 Connor Abbott
3 *
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:
10 *
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
13 * Software.
14 *
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
21 * IN THE SOFTWARE.
22 */
23
24 #include "util/half_float.h"
25 #include "elk_fs.h"
26 #include "elk_cfg.h"
27 #include "elk_fs_builder.h"
28
29 using namespace elk;
30
31 bool
lower_pack()32 elk_fs_visitor::lower_pack()
33 {
34 bool progress = false;
35
36 foreach_block_and_inst_safe(block, elk_fs_inst, inst, cfg) {
37 if (inst->opcode != ELK_FS_OPCODE_PACK &&
38 inst->opcode != ELK_FS_OPCODE_PACK_HALF_2x16_SPLIT)
39 continue;
40
41 assert(inst->dst.file == VGRF);
42 assert(inst->saturate == false);
43 elk_fs_reg dst = inst->dst;
44
45 const fs_builder ibld(this, block, inst);
46 /* The lowering generates 2 instructions for what was previously 1. This
47 * can trick the IR to believe we're doing partial writes, but the
48 * register is actually fully written. Mark it as undef to help the IR
49 * reduce the liveness of the register.
50 */
51 if (!inst->is_partial_write())
52 ibld.emit_undef_for_dst(inst);
53
54 switch (inst->opcode) {
55 case ELK_FS_OPCODE_PACK:
56 for (unsigned i = 0; i < inst->sources; i++)
57 ibld.MOV(subscript(dst, inst->src[i].type, i), inst->src[i]);
58 break;
59 case ELK_FS_OPCODE_PACK_HALF_2x16_SPLIT:
60 assert(dst.type == ELK_REGISTER_TYPE_UD);
61
62 for (unsigned i = 0; i < inst->sources; i++) {
63 if (inst->src[i].file == IMM) {
64 const uint32_t half = _mesa_float_to_half(inst->src[i].f);
65 ibld.MOV(subscript(dst, ELK_REGISTER_TYPE_UW, i),
66 elk_imm_uw(half));
67 } else if (i == 1) {
68 /* Pre-Skylake requires DWord aligned destinations */
69 elk_fs_reg tmp = ibld.vgrf(ELK_REGISTER_TYPE_UD);
70 ibld.F32TO16(subscript(tmp, ELK_REGISTER_TYPE_HF, 0),
71 inst->src[i]);
72 ibld.MOV(subscript(dst, ELK_REGISTER_TYPE_UW, 1),
73 subscript(tmp, ELK_REGISTER_TYPE_UW, 0));
74 } else {
75 ibld.F32TO16(subscript(dst, ELK_REGISTER_TYPE_HF, i),
76 inst->src[i]);
77 }
78 }
79 break;
80 default:
81 unreachable("skipped above");
82 }
83
84 inst->remove(block);
85 progress = true;
86 }
87
88 if (progress)
89 invalidate_analysis(DEPENDENCY_INSTRUCTIONS);
90
91 return progress;
92 }
93