| author | |
| committer | |
| log | 7adf15682ca263c7d4e0e5306517f22272310294 |
| tree | 7e9846e9389d883fa4035f9abdfd5e331968e635 |
| parent | d6467dcf71c40d4f09993f8c449b33adeca55ce5 |
13 files changed, 720 insertions(+), 80 deletions(-)
lib/std/target.zig+1| ... | ... | @@ -430,6 +430,7 @@ pub const Target = struct { |
| 430 | 430 | }; |
| 431 | 431 | |
| 432 | 432 | pub const aarch64 = @import("target/aarch64.zig"); |
| 433 | pub const arc = @import("target/arc.zig"); | |
| 433 | 434 | pub const amdgpu = @import("target/amdgpu.zig"); |
| 434 | 435 | pub const arm = @import("target/arm.zig"); |
| 435 | 436 | pub const avr = @import("target/avr.zig"); |
lib/std/target/aarch64.zig+107-19| ... | ... | @@ -65,6 +65,7 @@ pub const Feature = enum { |
| 65 | 65 | fuse_csel, |
| 66 | 66 | fuse_literals, |
| 67 | 67 | harden_sls_blr, |
| 68 | harden_sls_nocomdat, | |
| 68 | 69 | harden_sls_retbr, |
| 69 | 70 | hcx, |
| 70 | 71 | i8mm, |
| ... | ... | @@ -81,6 +82,7 @@ pub const Feature = enum { |
| 81 | 82 | neoverse_n2, |
| 82 | 83 | neoverse_v1, |
| 83 | 84 | no_neg_immediates, |
| 85 | no_zcz_fp, | |
| 84 | 86 | nv, |
| 85 | 87 | outline_atomics, |
| 86 | 88 | pan, |
| ... | ... | @@ -120,6 +122,7 @@ pub const Feature = enum { |
| 120 | 122 | reserve_x6, |
| 121 | 123 | reserve_x7, |
| 122 | 124 | reserve_x9, |
| 125 | rme, | |
| 123 | 126 | sb, |
| 124 | 127 | sel2, |
| 125 | 128 | sha2, |
| ... | ... | @@ -128,6 +131,9 @@ pub const Feature = enum { |
| 128 | 131 | slow_paired_128, |
| 129 | 132 | slow_strqro_store, |
| 130 | 133 | sm4, |
| 134 | sme, | |
| 135 | sme_f64, | |
| 136 | sme_i64, | |
| 131 | 137 | spe, |
| 132 | 138 | spe_eef, |
| 133 | 139 | specrestrict, |
| ... | ... | @@ -148,7 +154,6 @@ pub const Feature = enum { |
| 148 | 154 | tracev8_4, |
| 149 | 155 | trbe, |
| 150 | 156 | uaops, |
| 151 | use_aa, | |
| 152 | 157 | use_experimental_zeroing_pseudos, |
| 153 | 158 | use_postra_scheduler, |
| 154 | 159 | use_reciprocal_square_root, |
| ... | ... | @@ -166,7 +171,6 @@ pub const Feature = enum { |
| 166 | 171 | xs, |
| 167 | 172 | zcm, |
| 168 | 173 | zcz, |
| 169 | zcz_fp, | |
| 170 | 174 | zcz_fp_workaround, |
| 171 | 175 | zcz_gp, |
| 172 | 176 | }; |
| ... | ... | @@ -594,6 +598,11 @@ pub const all_features = blk: { |
| 594 | 598 | .description = "Harden against straight line speculation across BLR instructions", |
| 595 | 599 | .dependencies = featureSet(&[_]Feature{}), |
| 596 | 600 | }; |
| 601 | result[@enumToInt(Feature.harden_sls_nocomdat)] = .{ | |
| 602 | .llvm_name = "harden-sls-nocomdat", | |
| 603 | .description = "Generate thunk code for SLS mitigation in the normal text section", | |
| 604 | .dependencies = featureSet(&[_]Feature{}), | |
| 605 | }; | |
| 597 | 606 | result[@enumToInt(Feature.harden_sls_retbr)] = .{ |
| 598 | 607 | .llvm_name = "harden-sls-retbr", |
| 599 | 608 | .description = "Harden against straight line speculation across RET and BR instructions", |
| ... | ... | @@ -660,8 +669,10 @@ pub const all_features = blk: { |
| 660 | 669 | .crypto, |
| 661 | 670 | .dotprod, |
| 662 | 671 | .fullfp16, |
| 672 | .fuse_aes, | |
| 663 | 673 | .rcpc, |
| 664 | 674 | .ssbs, |
| 675 | .use_postra_scheduler, | |
| 665 | 676 | .v8_2a, |
| 666 | 677 | }), |
| 667 | 678 | }; |
| ... | ... | @@ -672,9 +683,11 @@ pub const all_features = blk: { |
| 672 | 683 | .crypto, |
| 673 | 684 | .dotprod, |
| 674 | 685 | .fullfp16, |
| 686 | .fuse_aes, | |
| 675 | 687 | .rcpc, |
| 676 | 688 | .spe, |
| 677 | 689 | .ssbs, |
| 690 | .use_postra_scheduler, | |
| 678 | 691 | .v8_2a, |
| 679 | 692 | }), |
| 680 | 693 | }; |
| ... | ... | @@ -683,10 +696,13 @@ pub const all_features = blk: { |
| 683 | 696 | .description = "Neoverse N2 ARM processors", |
| 684 | 697 | .dependencies = featureSet(&[_]Feature{ |
| 685 | 698 | .bf16, |
| 699 | .crypto, | |
| 686 | 700 | .ete, |
| 701 | .fuse_aes, | |
| 687 | 702 | .i8mm, |
| 688 | 703 | .mte, |
| 689 | 704 | .sve2_bitperm, |
| 705 | .use_postra_scheduler, | |
| 690 | 706 | .v8_5a, |
| 691 | 707 | }), |
| 692 | 708 | }; |
| ... | ... | @@ -714,6 +730,11 @@ pub const all_features = blk: { |
| 714 | 730 | .description = "Convert immediates and instructions to their negated or complemented equivalent when the immediate does not fit in the encoding.", |
| 715 | 731 | .dependencies = featureSet(&[_]Feature{}), |
| 716 | 732 | }; |
| 733 | result[@enumToInt(Feature.no_zcz_fp)] = .{ | |
| 734 | .llvm_name = "no-zcz-fp", | |
| 735 | .description = "Has no zero-cycle zeroing instructions for FP registers", | |
| 736 | .dependencies = featureSet(&[_]Feature{}), | |
| 737 | }; | |
| 717 | 738 | result[@enumToInt(Feature.nv)] = .{ |
| 718 | 739 | .llvm_name = "nv", |
| 719 | 740 | .description = "Enable v8.4-A Nested Virtualization Enchancement", |
| ... | ... | @@ -913,6 +934,11 @@ pub const all_features = blk: { |
| 913 | 934 | .description = "Reserve X9, making it unavailable as a GPR", |
| 914 | 935 | .dependencies = featureSet(&[_]Feature{}), |
| 915 | 936 | }; |
| 937 | result[@enumToInt(Feature.rme)] = .{ | |
| 938 | .llvm_name = "rme", | |
| 939 | .description = "Enable Realm Management Extension", | |
| 940 | .dependencies = featureSet(&[_]Feature{}), | |
| 941 | }; | |
| 916 | 942 | result[@enumToInt(Feature.sb)] = .{ |
| 917 | 943 | .llvm_name = "sb", |
| 918 | 944 | .description = "Enable v8.5 Speculation Barrier", |
| ... | ... | @@ -959,6 +985,28 @@ pub const all_features = blk: { |
| 959 | 985 | .neon, |
| 960 | 986 | }), |
| 961 | 987 | }; |
| 988 | result[@enumToInt(Feature.sme)] = .{ | |
| 989 | .llvm_name = "sme", | |
| 990 | .description = "Enable Scalable Matrix Extension (SME)", | |
| 991 | .dependencies = featureSet(&[_]Feature{ | |
| 992 | .bf16, | |
| 993 | .sve2, | |
| 994 | }), | |
| 995 | }; | |
| 996 | result[@enumToInt(Feature.sme_f64)] = .{ | |
| 997 | .llvm_name = "sme-f64", | |
| 998 | .description = "Enable Scalable Matrix Extension (SME) F64F64 instructions", | |
| 999 | .dependencies = featureSet(&[_]Feature{ | |
| 1000 | .sme, | |
| 1001 | }), | |
| 1002 | }; | |
| 1003 | result[@enumToInt(Feature.sme_i64)] = .{ | |
| 1004 | .llvm_name = "sme-i64", | |
| 1005 | .description = "Enable Scalable Matrix Extension (SME) I16I64 instructions", | |
| 1006 | .dependencies = featureSet(&[_]Feature{ | |
| 1007 | .sme, | |
| 1008 | }), | |
| 1009 | }; | |
| 962 | 1010 | result[@enumToInt(Feature.spe)] = .{ |
| 963 | 1011 | .llvm_name = "spe", |
| 964 | 1012 | .description = "Enable Statistical Profiling extension", |
| ... | ... | @@ -1074,11 +1122,6 @@ pub const all_features = blk: { |
| 1074 | 1122 | .description = "Enable v8.2 UAO PState", |
| 1075 | 1123 | .dependencies = featureSet(&[_]Feature{}), |
| 1076 | 1124 | }; |
| 1077 | result[@enumToInt(Feature.use_aa)] = .{ | |
| 1078 | .llvm_name = "use-aa", | |
| 1079 | .description = "Use alias analysis during codegen", | |
| 1080 | .dependencies = featureSet(&[_]Feature{}), | |
| 1081 | }; | |
| 1082 | 1125 | result[@enumToInt(Feature.use_experimental_zeroing_pseudos)] = .{ |
| 1083 | 1126 | .llvm_name = "use-experimental-zeroing-pseudos", |
| 1084 | 1127 | .description = "Hint to the compiler that the MOVPRFX instruction is merged with destructive operations", |
| ... | ... | @@ -1251,15 +1294,9 @@ pub const all_features = blk: { |
| 1251 | 1294 | .llvm_name = "zcz", |
| 1252 | 1295 | .description = "Has zero-cycle zeroing instructions", |
| 1253 | 1296 | .dependencies = featureSet(&[_]Feature{ |
| 1254 | .zcz_fp, | |
| 1255 | 1297 | .zcz_gp, |
| 1256 | 1298 | }), |
| 1257 | 1299 | }; |
| 1258 | result[@enumToInt(Feature.zcz_fp)] = .{ | |
| 1259 | .llvm_name = "zcz-fp", | |
| 1260 | .description = "Has zero-cycle zeroing instructions for FP registers", | |
| 1261 | .dependencies = featureSet(&[_]Feature{}), | |
| 1262 | }; | |
| 1263 | 1300 | result[@enumToInt(Feature.zcz_fp_workaround)] = .{ |
| 1264 | 1301 | .llvm_name = "zcz-fp-workaround", |
| 1265 | 1302 | .description = "The zero-cycle floating-point zeroing instruction has a bug", |
| ... | ... | @@ -1403,7 +1440,62 @@ pub const cpu = struct { |
| 1403 | 1440 | .name = "apple_latest", |
| 1404 | 1441 | .llvm_name = "apple-latest", |
| 1405 | 1442 | .features = featureSet(&[_]Feature{ |
| 1406 | .apple_a13, | |
| 1443 | .aggressive_fma, | |
| 1444 | .alternate_sextload_cvt_f32_pattern, | |
| 1445 | .altnzcv, | |
| 1446 | .arith_bcc_fusion, | |
| 1447 | .arith_cbz_fusion, | |
| 1448 | .ccdp, | |
| 1449 | .crypto, | |
| 1450 | .disable_latency_sched_heuristic, | |
| 1451 | .fp16fml, | |
| 1452 | .fptoint, | |
| 1453 | .fuse_address, | |
| 1454 | .fuse_aes, | |
| 1455 | .fuse_arith_logic, | |
| 1456 | .fuse_crypto_eor, | |
| 1457 | .fuse_csel, | |
| 1458 | .fuse_literals, | |
| 1459 | .perfmon, | |
| 1460 | .predres, | |
| 1461 | .sb, | |
| 1462 | .sha3, | |
| 1463 | .specrestrict, | |
| 1464 | .ssbs, | |
| 1465 | .v8_4a, | |
| 1466 | .zcm, | |
| 1467 | .zcz, | |
| 1468 | }), | |
| 1469 | }; | |
| 1470 | pub const apple_m1 = CpuModel{ | |
| 1471 | .name = "apple_m1", | |
| 1472 | .llvm_name = "apple-m1", | |
| 1473 | .features = featureSet(&[_]Feature{ | |
| 1474 | .aggressive_fma, | |
| 1475 | .alternate_sextload_cvt_f32_pattern, | |
| 1476 | .altnzcv, | |
| 1477 | .arith_bcc_fusion, | |
| 1478 | .arith_cbz_fusion, | |
| 1479 | .ccdp, | |
| 1480 | .crypto, | |
| 1481 | .disable_latency_sched_heuristic, | |
| 1482 | .fp16fml, | |
| 1483 | .fptoint, | |
| 1484 | .fuse_address, | |
| 1485 | .fuse_aes, | |
| 1486 | .fuse_arith_logic, | |
| 1487 | .fuse_crypto_eor, | |
| 1488 | .fuse_csel, | |
| 1489 | .fuse_literals, | |
| 1490 | .perfmon, | |
| 1491 | .predres, | |
| 1492 | .sb, | |
| 1493 | .sha3, | |
| 1494 | .specrestrict, | |
| 1495 | .ssbs, | |
| 1496 | .v8_4a, | |
| 1497 | .zcm, | |
| 1498 | .zcz, | |
| 1407 | 1499 | }), |
| 1408 | 1500 | }; |
| 1409 | 1501 | pub const apple_s4 = CpuModel{ |
| ... | ... | @@ -1459,7 +1551,6 @@ pub const cpu = struct { |
| 1459 | 1551 | .custom_cheap_as_move, |
| 1460 | 1552 | .fuse_aes, |
| 1461 | 1553 | .perfmon, |
| 1462 | .use_aa, | |
| 1463 | 1554 | .use_postra_scheduler, |
| 1464 | 1555 | .v8a, |
| 1465 | 1556 | }), |
| ... | ... | @@ -1474,6 +1565,7 @@ pub const cpu = struct { |
| 1474 | 1565 | .fuse_aes, |
| 1475 | 1566 | .perfmon, |
| 1476 | 1567 | .rcpc, |
| 1568 | .use_postra_scheduler, | |
| 1477 | 1569 | .v8_2a, |
| 1478 | 1570 | }), |
| 1479 | 1571 | }; |
| ... | ... | @@ -1649,7 +1741,6 @@ pub const cpu = struct { |
| 1649 | 1741 | .use_postra_scheduler, |
| 1650 | 1742 | .use_reciprocal_square_root, |
| 1651 | 1743 | .v8a, |
| 1652 | .zcz_fp, | |
| 1653 | 1744 | }), |
| 1654 | 1745 | }; |
| 1655 | 1746 | pub const exynos_m2 = CpuModel{ |
| ... | ... | @@ -1666,7 +1757,6 @@ pub const cpu = struct { |
| 1666 | 1757 | .slow_paired_128, |
| 1667 | 1758 | .use_postra_scheduler, |
| 1668 | 1759 | .v8a, |
| 1669 | .zcz_fp, | |
| 1670 | 1760 | }), |
| 1671 | 1761 | }; |
| 1672 | 1762 | pub const exynos_m3 = CpuModel{ |
| ... | ... | @@ -1686,7 +1776,6 @@ pub const cpu = struct { |
| 1686 | 1776 | .predictable_select_expensive, |
| 1687 | 1777 | .use_postra_scheduler, |
| 1688 | 1778 | .v8a, |
| 1689 | .zcz_fp, | |
| 1690 | 1779 | }), |
| 1691 | 1780 | }; |
| 1692 | 1781 | pub const exynos_m4 = CpuModel{ |
| ... | ... | @@ -1824,7 +1913,6 @@ pub const cpu = struct { |
| 1824 | 1913 | .perfmon, |
| 1825 | 1914 | .predictable_select_expensive, |
| 1826 | 1915 | .strict_align, |
| 1827 | .use_aa, | |
| 1828 | 1916 | .use_postra_scheduler, |
| 1829 | 1917 | .v8_3a, |
| 1830 | 1918 | }), |
lib/std/target/amdgpu.zig+255-8| ... | ... | @@ -9,6 +9,7 @@ pub const Feature = enum { |
| 9 | 9 | a16, |
| 10 | 10 | add_no_carry_insts, |
| 11 | 11 | aperture_regs, |
| 12 | architected_flat_scratch, | |
| 12 | 13 | atomic_fadd_insts, |
| 13 | 14 | auto_waitcnt_before_barrier, |
| 14 | 15 | ci_insts, |
| ... | ... | @@ -20,11 +21,14 @@ pub const Feature = enum { |
| 20 | 21 | dot4_insts, |
| 21 | 22 | dot5_insts, |
| 22 | 23 | dot6_insts, |
| 24 | dot7_insts, | |
| 23 | 25 | dpp, |
| 24 | 26 | dpp8, |
| 27 | dpp_64bit, | |
| 25 | 28 | ds_src2_insts, |
| 26 | 29 | enable_ds128, |
| 27 | 30 | enable_prt_strict_null, |
| 31 | extended_image_insts, | |
| 28 | 32 | fast_denormal_f32, |
| 29 | 33 | fast_fmaf, |
| 30 | 34 | flat_address_space, |
| ... | ... | @@ -36,16 +40,19 @@ pub const Feature = enum { |
| 36 | 40 | fma_mix_insts, |
| 37 | 41 | fmaf, |
| 38 | 42 | fp64, |
| 43 | full_rate_64_ops, | |
| 39 | 44 | g16, |
| 40 | 45 | gcn3_encoding, |
| 41 | 46 | get_wave_id_inst, |
| 42 | 47 | gfx10, |
| 43 | 48 | gfx10_3_insts, |
| 49 | gfx10_a_encoding, | |
| 44 | 50 | gfx10_b_encoding, |
| 45 | 51 | gfx10_insts, |
| 46 | 52 | gfx7_gfx8_gfx9_insts, |
| 47 | 53 | gfx8_insts, |
| 48 | 54 | gfx9, |
| 55 | gfx90a_insts, | |
| 49 | 56 | gfx9_insts, |
| 50 | 57 | half_rate_64_ops, |
| 51 | 58 | image_gather4_d16_bug, |
| ... | ... | @@ -70,11 +77,18 @@ pub const Feature = enum { |
| 70 | 77 | mfma_inline_literal_bug, |
| 71 | 78 | mimg_r128, |
| 72 | 79 | movrel, |
| 80 | negative_scratch_offset_bug, | |
| 81 | negative_unaligned_scratch_offset_bug, | |
| 73 | 82 | no_data_dep_hazard, |
| 74 | 83 | no_sdst_cmpx, |
| 84 | nsa_clause_bug, | |
| 75 | 85 | nsa_encoding, |
| 86 | nsa_max_size_13, | |
| 87 | nsa_max_size_5, | |
| 76 | 88 | nsa_to_vmem_bug, |
| 77 | 89 | offset_3f_bug, |
| 90 | packed_fp32_ops, | |
| 91 | packed_tid, | |
| 78 | 92 | pk_fmac_f16_inst, |
| 79 | 93 | promote_alloca, |
| 80 | 94 | r128_a16, |
| ... | ... | @@ -92,11 +106,13 @@ pub const Feature = enum { |
| 92 | 106 | sdwa_sdst, |
| 93 | 107 | sea_islands, |
| 94 | 108 | sgpr_init_bug, |
| 109 | shader_cycles_register, | |
| 95 | 110 | si_scheduler, |
| 96 | 111 | smem_to_vector_write_hazard, |
| 97 | 112 | southern_islands, |
| 98 | 113 | sramecc, |
| 99 | 114 | sramecc_support, |
| 115 | tgsplit, | |
| 100 | 116 | trap_handler, |
| 101 | 117 | trig_reduced_range, |
| 102 | 118 | unaligned_access_mode, |
| ... | ... | @@ -149,6 +165,11 @@ pub const all_features = blk: { |
| 149 | 165 | .description = "Has Memory Aperture Base and Size Registers", |
| 150 | 166 | .dependencies = featureSet(&[_]Feature{}), |
| 151 | 167 | }; |
| 168 | result[@enumToInt(Feature.architected_flat_scratch)] = .{ | |
| 169 | .llvm_name = "architected-flat-scratch", | |
| 170 | .description = "Flat Scratch register is a readonly SPI initialized architected register", | |
| 171 | .dependencies = featureSet(&[_]Feature{}), | |
| 172 | }; | |
| 152 | 173 | result[@enumToInt(Feature.atomic_fadd_insts)] = .{ |
| 153 | 174 | .llvm_name = "atomic-fadd-insts", |
| 154 | 175 | .description = "Has buffer_atomic_add_f32, buffer_atomic_pk_add_f16, global_atomic_add_f32, global_atomic_pk_add_f16 instructions", |
| ... | ... | @@ -183,7 +204,7 @@ pub const all_features = blk: { |
| 183 | 204 | }; |
| 184 | 205 | result[@enumToInt(Feature.dot2_insts)] = .{ |
| 185 | 206 | .llvm_name = "dot2-insts", |
| 186 | .description = "Has v_dot2_f32_f16, v_dot2_i32_i16, v_dot2_u32_u16, v_dot4_u32_u8, v_dot8_u32_u4 instructions", | |
| 207 | .description = "Has v_dot2_i32_i16, v_dot2_u32_u16 instructions", | |
| 187 | 208 | .dependencies = featureSet(&[_]Feature{}), |
| 188 | 209 | }; |
| 189 | 210 | result[@enumToInt(Feature.dot3_insts)] = .{ |
| ... | ... | @@ -206,6 +227,11 @@ pub const all_features = blk: { |
| 206 | 227 | .description = "Has v_dot4c_i32_i8 instruction", |
| 207 | 228 | .dependencies = featureSet(&[_]Feature{}), |
| 208 | 229 | }; |
| 230 | result[@enumToInt(Feature.dot7_insts)] = .{ | |
| 231 | .llvm_name = "dot7-insts", | |
| 232 | .description = "Has v_dot2_f32_f16, v_dot4_u32_u8, v_dot8_u32_u4 instructions", | |
| 233 | .dependencies = featureSet(&[_]Feature{}), | |
| 234 | }; | |
| 209 | 235 | result[@enumToInt(Feature.dpp)] = .{ |
| 210 | 236 | .llvm_name = "dpp", |
| 211 | 237 | .description = "Support DPP (Data Parallel Primitives) extension", |
| ... | ... | @@ -216,6 +242,11 @@ pub const all_features = blk: { |
| 216 | 242 | .description = "Support DPP8 (Data Parallel Primitives) extension", |
| 217 | 243 | .dependencies = featureSet(&[_]Feature{}), |
| 218 | 244 | }; |
| 245 | result[@enumToInt(Feature.dpp_64bit)] = .{ | |
| 246 | .llvm_name = "dpp-64bit", | |
| 247 | .description = "Support DPP (Data Parallel Primitives) extension", | |
| 248 | .dependencies = featureSet(&[_]Feature{}), | |
| 249 | }; | |
| 219 | 250 | result[@enumToInt(Feature.ds_src2_insts)] = .{ |
| 220 | 251 | .llvm_name = "ds-src2-insts", |
| 221 | 252 | .description = "Has ds_*_src2 instructions", |
| ... | ... | @@ -231,6 +262,11 @@ pub const all_features = blk: { |
| 231 | 262 | .description = "Enable zeroing of result registers for sparse texture fetches", |
| 232 | 263 | .dependencies = featureSet(&[_]Feature{}), |
| 233 | 264 | }; |
| 265 | result[@enumToInt(Feature.extended_image_insts)] = .{ | |
| 266 | .llvm_name = "extended-image-insts", | |
| 267 | .description = "Support mips != 0, lod != 0, gather4, and get_lod", | |
| 268 | .dependencies = featureSet(&[_]Feature{}), | |
| 269 | }; | |
| 234 | 270 | result[@enumToInt(Feature.fast_denormal_f32)] = .{ |
| 235 | 271 | .llvm_name = "fast-denormal-f32", |
| 236 | 272 | .description = "Enabling denormals does not cause f32 instructions to run at f64 rates", |
| ... | ... | @@ -268,7 +304,7 @@ pub const all_features = blk: { |
| 268 | 304 | }; |
| 269 | 305 | result[@enumToInt(Feature.flat_segment_offset_bug)] = .{ |
| 270 | 306 | .llvm_name = "flat-segment-offset-bug", |
| 271 | .description = "GFX10 bug, inst_offset ignored in flat segment", | |
| 307 | .description = "GFX10 bug where inst_offset is ignored when flat instructions access global memory", | |
| 272 | 308 | .dependencies = featureSet(&[_]Feature{}), |
| 273 | 309 | }; |
| 274 | 310 | result[@enumToInt(Feature.fma_mix_insts)] = .{ |
| ... | ... | @@ -286,6 +322,11 @@ pub const all_features = blk: { |
| 286 | 322 | .description = "Enable double precision operations", |
| 287 | 323 | .dependencies = featureSet(&[_]Feature{}), |
| 288 | 324 | }; |
| 325 | result[@enumToInt(Feature.full_rate_64_ops)] = .{ | |
| 326 | .llvm_name = "full-rate-64-ops", | |
| 327 | .description = "Most fp64 instructions are full rate", | |
| 328 | .dependencies = featureSet(&[_]Feature{}), | |
| 329 | }; | |
| 289 | 330 | result[@enumToInt(Feature.g16)] = .{ |
| 290 | 331 | .llvm_name = "g16", |
| 291 | 332 | .description = "Support G16 for 16-bit gradient image operands", |
| ... | ... | @@ -312,6 +353,7 @@ pub const all_features = blk: { |
| 312 | 353 | .ci_insts, |
| 313 | 354 | .dpp, |
| 314 | 355 | .dpp8, |
| 356 | .extended_image_insts, | |
| 315 | 357 | .fast_denormal_f32, |
| 316 | 358 | .fast_fmaf, |
| 317 | 359 | .flat_address_space, |
| ... | ... | @@ -334,6 +376,7 @@ pub const all_features = blk: { |
| 334 | 376 | .pk_fmac_f16_inst, |
| 335 | 377 | .register_banking, |
| 336 | 378 | .s_memrealtime, |
| 379 | .s_memtime_inst, | |
| 337 | 380 | .sdwa, |
| 338 | 381 | .sdwa_omod, |
| 339 | 382 | .sdwa_scalar, |
| ... | ... | @@ -350,6 +393,11 @@ pub const all_features = blk: { |
| 350 | 393 | .description = "Additional instructions for GFX10.3", |
| 351 | 394 | .dependencies = featureSet(&[_]Feature{}), |
| 352 | 395 | }; |
| 396 | result[@enumToInt(Feature.gfx10_a_encoding)] = .{ | |
| 397 | .llvm_name = "gfx10_a-encoding", | |
| 398 | .description = "Has BVH ray tracing instructions", | |
| 399 | .dependencies = featureSet(&[_]Feature{}), | |
| 400 | }; | |
| 353 | 401 | result[@enumToInt(Feature.gfx10_b_encoding)] = .{ |
| 354 | 402 | .llvm_name = "gfx10_b-encoding", |
| 355 | 403 | .description = "Encoding format GFX10_B", |
| ... | ... | @@ -379,7 +427,6 @@ pub const all_features = blk: { |
| 379 | 427 | .aperture_regs, |
| 380 | 428 | .ci_insts, |
| 381 | 429 | .dpp, |
| 382 | .ds_src2_insts, | |
| 383 | 430 | .fast_denormal_f32, |
| 384 | 431 | .fast_fmaf, |
| 385 | 432 | .flat_address_space, |
| ... | ... | @@ -394,7 +441,7 @@ pub const all_features = blk: { |
| 394 | 441 | .int_clamp_insts, |
| 395 | 442 | .inv_2pi_inline_imm, |
| 396 | 443 | .localmemorysize65536, |
| 397 | .mad_mac_f32_insts, | |
| 444 | .negative_scratch_offset_bug, | |
| 398 | 445 | .r128_a16, |
| 399 | 446 | .s_memrealtime, |
| 400 | 447 | .s_memtime_inst, |
| ... | ... | @@ -413,6 +460,11 @@ pub const all_features = blk: { |
| 413 | 460 | .xnack_support, |
| 414 | 461 | }), |
| 415 | 462 | }; |
| 463 | result[@enumToInt(Feature.gfx90a_insts)] = .{ | |
| 464 | .llvm_name = "gfx90a-insts", | |
| 465 | .description = "Additional instructions for GFX90A+", | |
| 466 | .dependencies = featureSet(&[_]Feature{}), | |
| 467 | }; | |
| 416 | 468 | result[@enumToInt(Feature.gfx9_insts)] = .{ |
| 417 | 469 | .llvm_name = "gfx9-insts", |
| 418 | 470 | .description = "Additional instructions for GFX9+", |
| ... | ... | @@ -533,6 +585,16 @@ pub const all_features = blk: { |
| 533 | 585 | .description = "Has v_movrel*_b32 instructions", |
| 534 | 586 | .dependencies = featureSet(&[_]Feature{}), |
| 535 | 587 | }; |
| 588 | result[@enumToInt(Feature.negative_scratch_offset_bug)] = .{ | |
| 589 | .llvm_name = "negative-scratch-offset-bug", | |
| 590 | .description = "Negative immediate offsets in scratch instructions with an SGPR offset page fault on GFX9", | |
| 591 | .dependencies = featureSet(&[_]Feature{}), | |
| 592 | }; | |
| 593 | result[@enumToInt(Feature.negative_unaligned_scratch_offset_bug)] = .{ | |
| 594 | .llvm_name = "negative-unaligned-scratch-offset-bug", | |
| 595 | .description = "Scratch instructions with a VGPR offset and a negative immediate offset that is not a multiple of 4 read wrong memory on GFX10", | |
| 596 | .dependencies = featureSet(&[_]Feature{}), | |
| 597 | }; | |
| 536 | 598 | result[@enumToInt(Feature.no_data_dep_hazard)] = .{ |
| 537 | 599 | .llvm_name = "no-data-dep-hazard", |
| 538 | 600 | .description = "Does not need SW waitstates", |
| ... | ... | @@ -543,11 +605,26 @@ pub const all_features = blk: { |
| 543 | 605 | .description = "V_CMPX does not write VCC/SGPR in addition to EXEC", |
| 544 | 606 | .dependencies = featureSet(&[_]Feature{}), |
| 545 | 607 | }; |
| 608 | result[@enumToInt(Feature.nsa_clause_bug)] = .{ | |
| 609 | .llvm_name = "nsa-clause-bug", | |
| 610 | .description = "MIMG-NSA in a hard clause has unpredictable results on GFX10.1", | |
| 611 | .dependencies = featureSet(&[_]Feature{}), | |
| 612 | }; | |
| 546 | 613 | result[@enumToInt(Feature.nsa_encoding)] = .{ |
| 547 | 614 | .llvm_name = "nsa-encoding", |
| 548 | 615 | .description = "Support NSA encoding for image instructions", |
| 549 | 616 | .dependencies = featureSet(&[_]Feature{}), |
| 550 | 617 | }; |
| 618 | result[@enumToInt(Feature.nsa_max_size_13)] = .{ | |
| 619 | .llvm_name = "nsa-max-size-13", | |
| 620 | .description = "The maximum non-sequential address size in VGPRs.", | |
| 621 | .dependencies = featureSet(&[_]Feature{}), | |
| 622 | }; | |
| 623 | result[@enumToInt(Feature.nsa_max_size_5)] = .{ | |
| 624 | .llvm_name = "nsa-max-size-5", | |
| 625 | .description = "The maximum non-sequential address size in VGPRs.", | |
| 626 | .dependencies = featureSet(&[_]Feature{}), | |
| 627 | }; | |
| 551 | 628 | result[@enumToInt(Feature.nsa_to_vmem_bug)] = .{ |
| 552 | 629 | .llvm_name = "nsa-to-vmem-bug", |
| 553 | 630 | .description = "MIMG-NSA followed by VMEM fail if EXEC_LO or EXEC_HI equals zero", |
| ... | ... | @@ -558,6 +635,16 @@ pub const all_features = blk: { |
| 558 | 635 | .description = "Branch offset of 3f hardware bug", |
| 559 | 636 | .dependencies = featureSet(&[_]Feature{}), |
| 560 | 637 | }; |
| 638 | result[@enumToInt(Feature.packed_fp32_ops)] = .{ | |
| 639 | .llvm_name = "packed-fp32-ops", | |
| 640 | .description = "Support packed fp32 instructions", | |
| 641 | .dependencies = featureSet(&[_]Feature{}), | |
| 642 | }; | |
| 643 | result[@enumToInt(Feature.packed_tid)] = .{ | |
| 644 | .llvm_name = "packed-tid", | |
| 645 | .description = "Workitem IDs are packed into v0 at kernel launch", | |
| 646 | .dependencies = featureSet(&[_]Feature{}), | |
| 647 | }; | |
| 561 | 648 | result[@enumToInt(Feature.pk_fmac_f16_inst)] = .{ |
| 562 | 649 | .llvm_name = "pk-fmac-f16-inst", |
| 563 | 650 | .description = "Has v_pk_fmac_f16 instruction", |
| ... | ... | @@ -639,6 +726,7 @@ pub const all_features = blk: { |
| 639 | 726 | .dependencies = featureSet(&[_]Feature{ |
| 640 | 727 | .ci_insts, |
| 641 | 728 | .ds_src2_insts, |
| 729 | .extended_image_insts, | |
| 642 | 730 | .flat_address_space, |
| 643 | 731 | .fp64, |
| 644 | 732 | .gfx7_gfx8_gfx9_insts, |
| ... | ... | @@ -657,6 +745,11 @@ pub const all_features = blk: { |
| 657 | 745 | .description = "VI SGPR initialization bug requiring a fixed SGPR allocation size", |
| 658 | 746 | .dependencies = featureSet(&[_]Feature{}), |
| 659 | 747 | }; |
| 748 | result[@enumToInt(Feature.shader_cycles_register)] = .{ | |
| 749 | .llvm_name = "shader-cycles-register", | |
| 750 | .description = "Has SHADER_CYCLES hardware register", | |
| 751 | .dependencies = featureSet(&[_]Feature{}), | |
| 752 | }; | |
| 660 | 753 | result[@enumToInt(Feature.si_scheduler)] = .{ |
| 661 | 754 | .llvm_name = "si-scheduler", |
| 662 | 755 | .description = "Enable SI Machine Scheduler", |
| ... | ... | @@ -672,6 +765,7 @@ pub const all_features = blk: { |
| 672 | 765 | .description = "SOUTHERN_ISLANDS GPU generation", |
| 673 | 766 | .dependencies = featureSet(&[_]Feature{ |
| 674 | 767 | .ds_src2_insts, |
| 768 | .extended_image_insts, | |
| 675 | 769 | .fp64, |
| 676 | 770 | .ldsbankcount32, |
| 677 | 771 | .localmemorysize32768, |
| ... | ... | @@ -693,6 +787,11 @@ pub const all_features = blk: { |
| 693 | 787 | .description = "Hardware supports SRAMECC", |
| 694 | 788 | .dependencies = featureSet(&[_]Feature{}), |
| 695 | 789 | }; |
| 790 | result[@enumToInt(Feature.tgsplit)] = .{ | |
| 791 | .llvm_name = "tgsplit", | |
| 792 | .description = "Enable threadgroup split execution", | |
| 793 | .dependencies = featureSet(&[_]Feature{}), | |
| 794 | }; | |
| 696 | 795 | result[@enumToInt(Feature.trap_handler)] = .{ |
| 697 | 796 | .llvm_name = "trap-handler", |
| 698 | 797 | .description = "Trap handler support", |
| ... | ... | @@ -761,6 +860,7 @@ pub const all_features = blk: { |
| 761 | 860 | .ci_insts, |
| 762 | 861 | .dpp, |
| 763 | 862 | .ds_src2_insts, |
| 863 | .extended_image_insts, | |
| 764 | 864 | .fast_denormal_f32, |
| 765 | 865 | .flat_address_space, |
| 766 | 866 | .fp64, |
| ... | ... | @@ -892,10 +992,12 @@ pub const cpu = struct { |
| 892 | 992 | .lds_misaligned_bug, |
| 893 | 993 | .ldsbankcount32, |
| 894 | 994 | .mad_mac_f32_insts, |
| 995 | .negative_unaligned_scratch_offset_bug, | |
| 996 | .nsa_clause_bug, | |
| 895 | 997 | .nsa_encoding, |
| 998 | .nsa_max_size_5, | |
| 896 | 999 | .nsa_to_vmem_bug, |
| 897 | 1000 | .offset_3f_bug, |
| 898 | .s_memtime_inst, | |
| 899 | 1001 | .scalar_atomics, |
| 900 | 1002 | .scalar_flat_scratch_insts, |
| 901 | 1003 | .scalar_stores, |
| ... | ... | @@ -916,6 +1018,7 @@ pub const cpu = struct { |
| 916 | 1018 | .dot2_insts, |
| 917 | 1019 | .dot5_insts, |
| 918 | 1020 | .dot6_insts, |
| 1021 | .dot7_insts, | |
| 919 | 1022 | .ds_src2_insts, |
| 920 | 1023 | .flat_segment_offset_bug, |
| 921 | 1024 | .get_wave_id_inst, |
| ... | ... | @@ -925,10 +1028,12 @@ pub const cpu = struct { |
| 925 | 1028 | .lds_misaligned_bug, |
| 926 | 1029 | .ldsbankcount32, |
| 927 | 1030 | .mad_mac_f32_insts, |
| 1031 | .negative_unaligned_scratch_offset_bug, | |
| 1032 | .nsa_clause_bug, | |
| 928 | 1033 | .nsa_encoding, |
| 1034 | .nsa_max_size_5, | |
| 929 | 1035 | .nsa_to_vmem_bug, |
| 930 | 1036 | .offset_3f_bug, |
| 931 | .s_memtime_inst, | |
| 932 | 1037 | .scalar_atomics, |
| 933 | 1038 | .scalar_flat_scratch_insts, |
| 934 | 1039 | .scalar_stores, |
| ... | ... | @@ -949,6 +1054,7 @@ pub const cpu = struct { |
| 949 | 1054 | .dot2_insts, |
| 950 | 1055 | .dot5_insts, |
| 951 | 1056 | .dot6_insts, |
| 1057 | .dot7_insts, | |
| 952 | 1058 | .ds_src2_insts, |
| 953 | 1059 | .flat_segment_offset_bug, |
| 954 | 1060 | .get_wave_id_inst, |
| ... | ... | @@ -958,10 +1064,44 @@ pub const cpu = struct { |
| 958 | 1064 | .lds_misaligned_bug, |
| 959 | 1065 | .ldsbankcount32, |
| 960 | 1066 | .mad_mac_f32_insts, |
| 1067 | .negative_unaligned_scratch_offset_bug, | |
| 1068 | .nsa_clause_bug, | |
| 961 | 1069 | .nsa_encoding, |
| 1070 | .nsa_max_size_5, | |
| 1071 | .nsa_to_vmem_bug, | |
| 1072 | .offset_3f_bug, | |
| 1073 | .scalar_atomics, | |
| 1074 | .scalar_flat_scratch_insts, | |
| 1075 | .scalar_stores, | |
| 1076 | .smem_to_vector_write_hazard, | |
| 1077 | .vcmpx_exec_war_hazard, | |
| 1078 | .vcmpx_permlane_hazard, | |
| 1079 | .vmem_to_scalar_write_hazard, | |
| 1080 | .wavefrontsize32, | |
| 1081 | .xnack_support, | |
| 1082 | }), | |
| 1083 | }; | |
| 1084 | pub const gfx1013 = CpuModel{ | |
| 1085 | .name = "gfx1013", | |
| 1086 | .llvm_name = "gfx1013", | |
| 1087 | .features = featureSet(&[_]Feature{ | |
| 1088 | .dl_insts, | |
| 1089 | .ds_src2_insts, | |
| 1090 | .flat_segment_offset_bug, | |
| 1091 | .get_wave_id_inst, | |
| 1092 | .gfx10, | |
| 1093 | .gfx10_a_encoding, | |
| 1094 | .inst_fwd_prefetch_bug, | |
| 1095 | .lds_branch_vmem_war_hazard, | |
| 1096 | .lds_misaligned_bug, | |
| 1097 | .ldsbankcount32, | |
| 1098 | .mad_mac_f32_insts, | |
| 1099 | .negative_unaligned_scratch_offset_bug, | |
| 1100 | .nsa_clause_bug, | |
| 1101 | .nsa_encoding, | |
| 1102 | .nsa_max_size_5, | |
| 962 | 1103 | .nsa_to_vmem_bug, |
| 963 | 1104 | .offset_3f_bug, |
| 964 | .s_memtime_inst, | |
| 965 | 1105 | .scalar_atomics, |
| 966 | 1106 | .scalar_flat_scratch_insts, |
| 967 | 1107 | .scalar_stores, |
| ... | ... | @@ -982,11 +1122,15 @@ pub const cpu = struct { |
| 982 | 1122 | .dot2_insts, |
| 983 | 1123 | .dot5_insts, |
| 984 | 1124 | .dot6_insts, |
| 1125 | .dot7_insts, | |
| 985 | 1126 | .gfx10, |
| 986 | 1127 | .gfx10_3_insts, |
| 1128 | .gfx10_a_encoding, | |
| 987 | 1129 | .gfx10_b_encoding, |
| 988 | 1130 | .ldsbankcount32, |
| 989 | 1131 | .nsa_encoding, |
| 1132 | .nsa_max_size_13, | |
| 1133 | .shader_cycles_register, | |
| 990 | 1134 | .wavefrontsize32, |
| 991 | 1135 | }), |
| 992 | 1136 | }; |
| ... | ... | @@ -999,11 +1143,15 @@ pub const cpu = struct { |
| 999 | 1143 | .dot2_insts, |
| 1000 | 1144 | .dot5_insts, |
| 1001 | 1145 | .dot6_insts, |
| 1146 | .dot7_insts, | |
| 1002 | 1147 | .gfx10, |
| 1003 | 1148 | .gfx10_3_insts, |
| 1149 | .gfx10_a_encoding, | |
| 1004 | 1150 | .gfx10_b_encoding, |
| 1005 | 1151 | .ldsbankcount32, |
| 1006 | 1152 | .nsa_encoding, |
| 1153 | .nsa_max_size_13, | |
| 1154 | .shader_cycles_register, | |
| 1007 | 1155 | .wavefrontsize32, |
| 1008 | 1156 | }), |
| 1009 | 1157 | }; |
| ... | ... | @@ -1016,11 +1164,15 @@ pub const cpu = struct { |
| 1016 | 1164 | .dot2_insts, |
| 1017 | 1165 | .dot5_insts, |
| 1018 | 1166 | .dot6_insts, |
| 1167 | .dot7_insts, | |
| 1019 | 1168 | .gfx10, |
| 1020 | 1169 | .gfx10_3_insts, |
| 1170 | .gfx10_a_encoding, | |
| 1021 | 1171 | .gfx10_b_encoding, |
| 1022 | 1172 | .ldsbankcount32, |
| 1023 | 1173 | .nsa_encoding, |
| 1174 | .nsa_max_size_13, | |
| 1175 | .shader_cycles_register, | |
| 1024 | 1176 | .wavefrontsize32, |
| 1025 | 1177 | }), |
| 1026 | 1178 | }; |
| ... | ... | @@ -1033,11 +1185,57 @@ pub const cpu = struct { |
| 1033 | 1185 | .dot2_insts, |
| 1034 | 1186 | .dot5_insts, |
| 1035 | 1187 | .dot6_insts, |
| 1188 | .dot7_insts, | |
| 1189 | .gfx10, | |
| 1190 | .gfx10_3_insts, | |
| 1191 | .gfx10_a_encoding, | |
| 1192 | .gfx10_b_encoding, | |
| 1193 | .ldsbankcount32, | |
| 1194 | .nsa_encoding, | |
| 1195 | .nsa_max_size_13, | |
| 1196 | .shader_cycles_register, | |
| 1197 | .wavefrontsize32, | |
| 1198 | }), | |
| 1199 | }; | |
| 1200 | pub const gfx1034 = CpuModel{ | |
| 1201 | .name = "gfx1034", | |
| 1202 | .llvm_name = "gfx1034", | |
| 1203 | .features = featureSet(&[_]Feature{ | |
| 1204 | .dl_insts, | |
| 1205 | .dot1_insts, | |
| 1206 | .dot2_insts, | |
| 1207 | .dot5_insts, | |
| 1208 | .dot6_insts, | |
| 1209 | .dot7_insts, | |
| 1036 | 1210 | .gfx10, |
| 1037 | 1211 | .gfx10_3_insts, |
| 1212 | .gfx10_a_encoding, | |
| 1038 | 1213 | .gfx10_b_encoding, |
| 1039 | 1214 | .ldsbankcount32, |
| 1040 | 1215 | .nsa_encoding, |
| 1216 | .nsa_max_size_13, | |
| 1217 | .shader_cycles_register, | |
| 1218 | .wavefrontsize32, | |
| 1219 | }), | |
| 1220 | }; | |
| 1221 | pub const gfx1035 = CpuModel{ | |
| 1222 | .name = "gfx1035", | |
| 1223 | .llvm_name = "gfx1035", | |
| 1224 | .features = featureSet(&[_]Feature{ | |
| 1225 | .dl_insts, | |
| 1226 | .dot1_insts, | |
| 1227 | .dot2_insts, | |
| 1228 | .dot5_insts, | |
| 1229 | .dot6_insts, | |
| 1230 | .dot7_insts, | |
| 1231 | .gfx10, | |
| 1232 | .gfx10_3_insts, | |
| 1233 | .gfx10_a_encoding, | |
| 1234 | .gfx10_b_encoding, | |
| 1235 | .ldsbankcount32, | |
| 1236 | .nsa_encoding, | |
| 1237 | .nsa_max_size_13, | |
| 1238 | .shader_cycles_register, | |
| 1041 | 1239 | .wavefrontsize32, |
| 1042 | 1240 | }), |
| 1043 | 1241 | }; |
| ... | ... | @@ -1171,9 +1369,12 @@ pub const cpu = struct { |
| 1171 | 1369 | .name = "gfx900", |
| 1172 | 1370 | .llvm_name = "gfx900", |
| 1173 | 1371 | .features = featureSet(&[_]Feature{ |
| 1372 | .ds_src2_insts, | |
| 1373 | .extended_image_insts, | |
| 1174 | 1374 | .gfx9, |
| 1175 | 1375 | .image_gather4_d16_bug, |
| 1176 | 1376 | .ldsbankcount32, |
| 1377 | .mad_mac_f32_insts, | |
| 1177 | 1378 | .mad_mix_insts, |
| 1178 | 1379 | }), |
| 1179 | 1380 | }; |
| ... | ... | @@ -1181,9 +1382,12 @@ pub const cpu = struct { |
| 1181 | 1382 | .name = "gfx902", |
| 1182 | 1383 | .llvm_name = "gfx902", |
| 1183 | 1384 | .features = featureSet(&[_]Feature{ |
| 1385 | .ds_src2_insts, | |
| 1386 | .extended_image_insts, | |
| 1184 | 1387 | .gfx9, |
| 1185 | 1388 | .image_gather4_d16_bug, |
| 1186 | 1389 | .ldsbankcount32, |
| 1390 | .mad_mac_f32_insts, | |
| 1187 | 1391 | .mad_mix_insts, |
| 1188 | 1392 | }), |
| 1189 | 1393 | }; |
| ... | ... | @@ -1191,10 +1395,13 @@ pub const cpu = struct { |
| 1191 | 1395 | .name = "gfx904", |
| 1192 | 1396 | .llvm_name = "gfx904", |
| 1193 | 1397 | .features = featureSet(&[_]Feature{ |
| 1398 | .ds_src2_insts, | |
| 1399 | .extended_image_insts, | |
| 1194 | 1400 | .fma_mix_insts, |
| 1195 | 1401 | .gfx9, |
| 1196 | 1402 | .image_gather4_d16_bug, |
| 1197 | 1403 | .ldsbankcount32, |
| 1404 | .mad_mac_f32_insts, | |
| 1198 | 1405 | }), |
| 1199 | 1406 | }; |
| 1200 | 1407 | pub const gfx906 = CpuModel{ |
| ... | ... | @@ -1204,11 +1411,15 @@ pub const cpu = struct { |
| 1204 | 1411 | .dl_insts, |
| 1205 | 1412 | .dot1_insts, |
| 1206 | 1413 | .dot2_insts, |
| 1414 | .dot7_insts, | |
| 1415 | .ds_src2_insts, | |
| 1416 | .extended_image_insts, | |
| 1207 | 1417 | .fma_mix_insts, |
| 1208 | 1418 | .gfx9, |
| 1209 | 1419 | .half_rate_64_ops, |
| 1210 | 1420 | .image_gather4_d16_bug, |
| 1211 | 1421 | .ldsbankcount32, |
| 1422 | .mad_mac_f32_insts, | |
| 1212 | 1423 | .sramecc_support, |
| 1213 | 1424 | }), |
| 1214 | 1425 | }; |
| ... | ... | @@ -1224,11 +1435,15 @@ pub const cpu = struct { |
| 1224 | 1435 | .dot4_insts, |
| 1225 | 1436 | .dot5_insts, |
| 1226 | 1437 | .dot6_insts, |
| 1438 | .dot7_insts, | |
| 1439 | .ds_src2_insts, | |
| 1440 | .extended_image_insts, | |
| 1227 | 1441 | .fma_mix_insts, |
| 1228 | 1442 | .gfx9, |
| 1229 | 1443 | .half_rate_64_ops, |
| 1230 | 1444 | .image_gather4_d16_bug, |
| 1231 | 1445 | .ldsbankcount32, |
| 1446 | .mad_mac_f32_insts, | |
| 1232 | 1447 | .mai_insts, |
| 1233 | 1448 | .mfma_inline_literal_bug, |
| 1234 | 1449 | .pk_fmac_f16_inst, |
| ... | ... | @@ -1239,21 +1454,53 @@ pub const cpu = struct { |
| 1239 | 1454 | .name = "gfx909", |
| 1240 | 1455 | .llvm_name = "gfx909", |
| 1241 | 1456 | .features = featureSet(&[_]Feature{ |
| 1457 | .ds_src2_insts, | |
| 1458 | .extended_image_insts, | |
| 1242 | 1459 | .gfx9, |
| 1243 | 1460 | .image_gather4_d16_bug, |
| 1244 | 1461 | .ldsbankcount32, |
| 1462 | .mad_mac_f32_insts, | |
| 1245 | 1463 | .mad_mix_insts, |
| 1246 | 1464 | }), |
| 1247 | 1465 | }; |
| 1466 | pub const gfx90a = CpuModel{ | |
| 1467 | .name = "gfx90a", | |
| 1468 | .llvm_name = "gfx90a", | |
| 1469 | .features = featureSet(&[_]Feature{ | |
| 1470 | .atomic_fadd_insts, | |
| 1471 | .dl_insts, | |
| 1472 | .dot1_insts, | |
| 1473 | .dot2_insts, | |
| 1474 | .dot3_insts, | |
| 1475 | .dot4_insts, | |
| 1476 | .dot5_insts, | |
| 1477 | .dot6_insts, | |
| 1478 | .dot7_insts, | |
| 1479 | .dpp_64bit, | |
| 1480 | .fma_mix_insts, | |
| 1481 | .full_rate_64_ops, | |
| 1482 | .gfx9, | |
| 1483 | .gfx90a_insts, | |
| 1484 | .ldsbankcount32, | |
| 1485 | .mad_mac_f32_insts, | |
| 1486 | .mai_insts, | |
| 1487 | .packed_fp32_ops, | |
| 1488 | .packed_tid, | |
| 1489 | .pk_fmac_f16_inst, | |
| 1490 | .sramecc_support, | |
| 1491 | }), | |
| 1492 | }; | |
| 1248 | 1493 | pub const gfx90c = CpuModel{ |
| 1249 | 1494 | .name = "gfx90c", |
| 1250 | 1495 | .llvm_name = "gfx90c", |
| 1251 | 1496 | .features = featureSet(&[_]Feature{ |
| 1497 | .ds_src2_insts, | |
| 1498 | .extended_image_insts, | |
| 1252 | 1499 | .gfx9, |
| 1253 | 1500 | .image_gather4_d16_bug, |
| 1254 | 1501 | .ldsbankcount32, |
| 1502 | .mad_mac_f32_insts, | |
| 1255 | 1503 | .mad_mix_insts, |
| 1256 | .xnack, | |
| 1257 | 1504 | }), |
| 1258 | 1505 | }; |
| 1259 | 1506 | pub const hainan = CpuModel{ |
lib/std/target/arc.zig created+39| ... | ... | @@ -0,0 +1,39 @@ |
| 1 | //! This file is auto-generated by tools/update_cpu_features.zig. | |
| 2 | ||
| 3 | const std = @import("../std.zig"); | |
| 4 | const CpuFeature = std.Target.Cpu.Feature; | |
| 5 | const CpuModel = std.Target.Cpu.Model; | |
| 6 | ||
| 7 | pub const Feature = enum { | |
| 8 | norm, | |
| 9 | }; | |
| 10 | ||
| 11 | pub const featureSet = CpuFeature.feature_set_fns(Feature).featureSet; | |
| 12 | pub const featureSetHas = CpuFeature.feature_set_fns(Feature).featureSetHas; | |
| 13 | pub const featureSetHasAny = CpuFeature.feature_set_fns(Feature).featureSetHasAny; | |
| 14 | pub const featureSetHasAll = CpuFeature.feature_set_fns(Feature).featureSetHasAll; | |
| 15 | ||
| 16 | pub const all_features = blk: { | |
| 17 | const len = @typeInfo(Feature).Enum.fields.len; | |
| 18 | std.debug.assert(len <= CpuFeature.Set.needed_bit_count); | |
| 19 | var result: [len]CpuFeature = undefined; | |
| 20 | result[@enumToInt(Feature.norm)] = .{ | |
| 21 | .llvm_name = "norm", | |
| 22 | .description = "Enable support for norm instruction.", | |
| 23 | .dependencies = featureSet(&[_]Feature{}), | |
| 24 | }; | |
| 25 | const ti = @typeInfo(Feature); | |
| 26 | for (result) |*elem, i| { | |
| 27 | elem.index = i; | |
| 28 | elem.name = ti.Enum.fields[i].name; | |
| 29 | } | |
| 30 | break :blk result; | |
| 31 | }; | |
| 32 | ||
| 33 | pub const cpu = struct { | |
| 34 | pub const generic = CpuModel{ | |
| 35 | .name = "generic", | |
| 36 | .llvm_name = "generic", | |
| 37 | .features = featureSet(&[_]Feature{}), | |
| 38 | }; | |
| 39 | }; |
lib/std/target/arm.zig+12-8| ... | ... | @@ -51,6 +51,7 @@ pub const Feature = enum { |
| 51 | 51 | fuse_aes, |
| 52 | 52 | fuse_literals, |
| 53 | 53 | harden_sls_blr, |
| 54 | harden_sls_nocomdat, | |
| 54 | 55 | harden_sls_retbr, |
| 55 | 56 | has_v4t, |
| 56 | 57 | has_v5t, |
| ... | ... | @@ -487,6 +488,11 @@ pub const all_features = blk: { |
| 487 | 488 | .description = "Harden against straight line speculation across indirect calls", |
| 488 | 489 | .dependencies = featureSet(&[_]Feature{}), |
| 489 | 490 | }; |
| 491 | result[@enumToInt(Feature.harden_sls_nocomdat)] = .{ | |
| 492 | .llvm_name = "harden-sls-nocomdat", | |
| 493 | .description = "Generate thunk code for SLS mitigation in the normal text section", | |
| 494 | .dependencies = featureSet(&[_]Feature{}), | |
| 495 | }; | |
| 490 | 496 | result[@enumToInt(Feature.harden_sls_retbr)] = .{ |
| 491 | 497 | .llvm_name = "harden-sls-retbr", |
| 492 | 498 | .description = "Harden against straight line speculation across RETurn and BranchRegister instructions", |
| ... | ... | @@ -1245,7 +1251,7 @@ pub const all_features = blk: { |
| 1245 | 1251 | }; |
| 1246 | 1252 | result[@enumToInt(Feature.v8_7a)] = .{ |
| 1247 | 1253 | .llvm_name = "armv8.7-a", |
| 1248 | .description = "ARMv86a architecture", | |
| 1254 | .description = "ARMv87a architecture", | |
| 1249 | 1255 | .dependencies = featureSet(&[_]Feature{ |
| 1250 | 1256 | .aclass, |
| 1251 | 1257 | .crc, |
| ... | ... | @@ -1517,13 +1523,6 @@ pub const cpu = struct { |
| 1517 | 1523 | .vfp2, |
| 1518 | 1524 | }), |
| 1519 | 1525 | }; |
| 1520 | pub const arm1176j_s = CpuModel{ | |
| 1521 | .name = "arm1176j_s", | |
| 1522 | .llvm_name = "arm1176j-s", | |
| 1523 | .features = featureSet(&[_]Feature{ | |
| 1524 | .v6kz, | |
| 1525 | }), | |
| 1526 | }; | |
| 1527 | 1526 | pub const arm1176jz_s = CpuModel{ |
| 1528 | 1527 | .name = "arm1176jz_s", |
| 1529 | 1528 | .llvm_name = "arm1176jz-s", |
| ... | ... | @@ -1889,6 +1888,7 @@ pub const cpu = struct { |
| 1889 | 1888 | .name = "cortex_m0", |
| 1890 | 1889 | .llvm_name = "cortex-m0", |
| 1891 | 1890 | .features = featureSet(&[_]Feature{ |
| 1891 | .no_branch_predictor, | |
| 1892 | 1892 | .v6m, |
| 1893 | 1893 | }), |
| 1894 | 1894 | }; |
| ... | ... | @@ -1896,6 +1896,7 @@ pub const cpu = struct { |
| 1896 | 1896 | .name = "cortex_m0plus", |
| 1897 | 1897 | .llvm_name = "cortex-m0plus", |
| 1898 | 1898 | .features = featureSet(&[_]Feature{ |
| 1899 | .no_branch_predictor, | |
| 1899 | 1900 | .v6m, |
| 1900 | 1901 | }), |
| 1901 | 1902 | }; |
| ... | ... | @@ -1903,6 +1904,7 @@ pub const cpu = struct { |
| 1903 | 1904 | .name = "cortex_m1", |
| 1904 | 1905 | .llvm_name = "cortex-m1", |
| 1905 | 1906 | .features = featureSet(&[_]Feature{ |
| 1907 | .no_branch_predictor, | |
| 1906 | 1908 | .v6m, |
| 1907 | 1909 | }), |
| 1908 | 1910 | }; |
| ... | ... | @@ -1910,6 +1912,7 @@ pub const cpu = struct { |
| 1910 | 1912 | .name = "cortex_m23", |
| 1911 | 1913 | .llvm_name = "cortex-m23", |
| 1912 | 1914 | .features = featureSet(&[_]Feature{ |
| 1915 | .no_branch_predictor, | |
| 1913 | 1916 | .no_movt, |
| 1914 | 1917 | .v8m, |
| 1915 | 1918 | }), |
| ... | ... | @@ -2225,6 +2228,7 @@ pub const cpu = struct { |
| 2225 | 2228 | .name = "sc000", |
| 2226 | 2229 | .llvm_name = "sc000", |
| 2227 | 2230 | .features = featureSet(&[_]Feature{ |
| 2231 | .no_branch_predictor, | |
| 2228 | 2232 | .v6m, |
| 2229 | 2233 | }), |
| 2230 | 2234 | }; |
lib/std/target/hexagon.zig+35| ... | ... | @@ -16,6 +16,7 @@ pub const Feature = enum { |
| 16 | 16 | hvxv65, |
| 17 | 17 | hvxv66, |
| 18 | 18 | hvxv67, |
| 19 | hvxv68, | |
| 19 | 20 | long_calls, |
| 20 | 21 | mem_noshuf, |
| 21 | 22 | memops, |
| ... | ... | @@ -35,6 +36,7 @@ pub const Feature = enum { |
| 35 | 36 | v65, |
| 36 | 37 | v66, |
| 37 | 38 | v67, |
| 39 | v68, | |
| 38 | 40 | zreg, |
| 39 | 41 | }; |
| 40 | 42 | |
| ... | ... | @@ -117,6 +119,13 @@ pub const all_features = blk: { |
| 117 | 119 | .hvxv66, |
| 118 | 120 | }), |
| 119 | 121 | }; |
| 122 | result[@enumToInt(Feature.hvxv68)] = .{ | |
| 123 | .llvm_name = "hvxv68", | |
| 124 | .description = "Hexagon HVX instructions", | |
| 125 | .dependencies = featureSet(&[_]Feature{ | |
| 126 | .hvxv67, | |
| 127 | }), | |
| 128 | }; | |
| 120 | 129 | result[@enumToInt(Feature.long_calls)] = .{ |
| 121 | 130 | .llvm_name = "long-calls", |
| 122 | 131 | .description = "Use constant-extended calls", |
| ... | ... | @@ -216,6 +225,11 @@ pub const all_features = blk: { |
| 216 | 225 | .description = "Enable Hexagon V67 architecture", |
| 217 | 226 | .dependencies = featureSet(&[_]Feature{}), |
| 218 | 227 | }; |
| 228 | result[@enumToInt(Feature.v68)] = .{ | |
| 229 | .llvm_name = "v68", | |
| 230 | .description = "Enable Hexagon V68 architecture", | |
| 231 | .dependencies = featureSet(&[_]Feature{}), | |
| 232 | }; | |
| 219 | 233 | result[@enumToInt(Feature.zreg)] = .{ |
| 220 | 234 | .llvm_name = "zreg", |
| 221 | 235 | .description = "Hexagon ZReg extension instructions", |
| ... | ... | @@ -385,4 +399,25 @@ pub const cpu = struct { |
| 385 | 399 | .v67, |
| 386 | 400 | }), |
| 387 | 401 | }; |
| 402 | pub const hexagonv68 = CpuModel{ | |
| 403 | .name = "hexagonv68", | |
| 404 | .llvm_name = "hexagonv68", | |
| 405 | .features = featureSet(&[_]Feature{ | |
| 406 | .compound, | |
| 407 | .duplex, | |
| 408 | .mem_noshuf, | |
| 409 | .memops, | |
| 410 | .nvj, | |
| 411 | .nvs, | |
| 412 | .small_data, | |
| 413 | .v5, | |
| 414 | .v55, | |
| 415 | .v60, | |
| 416 | .v62, | |
| 417 | .v65, | |
| 418 | .v66, | |
| 419 | .v67, | |
| 420 | .v68, | |
| 421 | }), | |
| 422 | }; | |
| 388 | 423 | }; |
lib/std/target/nvptx.zig+26| ... | ... | @@ -17,6 +17,8 @@ pub const Feature = enum { |
| 17 | 17 | ptx64, |
| 18 | 18 | ptx65, |
| 19 | 19 | ptx70, |
| 20 | ptx71, | |
| 21 | ptx72, | |
| 20 | 22 | sm_20, |
| 21 | 23 | sm_21, |
| 22 | 24 | sm_30, |
| ... | ... | @@ -33,6 +35,7 @@ pub const Feature = enum { |
| 33 | 35 | sm_72, |
| 34 | 36 | sm_75, |
| 35 | 37 | sm_80, |
| 38 | sm_86, | |
| 36 | 39 | }; |
| 37 | 40 | |
| 38 | 41 | pub const featureSet = CpuFeature.feature_set_fns(Feature).featureSet; |
| ... | ... | @@ -104,6 +107,16 @@ pub const all_features = blk: { |
| 104 | 107 | .description = "Use PTX version 7.0", |
| 105 | 108 | .dependencies = featureSet(&[_]Feature{}), |
| 106 | 109 | }; |
| 110 | result[@enumToInt(Feature.ptx71)] = .{ | |
| 111 | .llvm_name = "ptx71", | |
| 112 | .description = "Use PTX version 7.1", | |
| 113 | .dependencies = featureSet(&[_]Feature{}), | |
| 114 | }; | |
| 115 | result[@enumToInt(Feature.ptx72)] = .{ | |
| 116 | .llvm_name = "ptx72", | |
| 117 | .description = "Use PTX version 7.2", | |
| 118 | .dependencies = featureSet(&[_]Feature{}), | |
| 119 | }; | |
| 107 | 120 | result[@enumToInt(Feature.sm_20)] = .{ |
| 108 | 121 | .llvm_name = "sm_20", |
| 109 | 122 | .description = "Target SM 2.0", |
| ... | ... | @@ -184,6 +197,11 @@ pub const all_features = blk: { |
| 184 | 197 | .description = "Target SM 8.0", |
| 185 | 198 | .dependencies = featureSet(&[_]Feature{}), |
| 186 | 199 | }; |
| 200 | result[@enumToInt(Feature.sm_86)] = .{ | |
| 201 | .llvm_name = "sm_86", | |
| 202 | .description = "Target SM 8.6", | |
| 203 | .dependencies = featureSet(&[_]Feature{}), | |
| 204 | }; | |
| 187 | 205 | const ti = @typeInfo(Feature); |
| 188 | 206 | for (result) |*elem, i| { |
| 189 | 207 | elem.index = i; |
| ... | ... | @@ -317,4 +335,12 @@ pub const cpu = struct { |
| 317 | 335 | .sm_80, |
| 318 | 336 | }), |
| 319 | 337 | }; |
| 338 | pub const sm_86 = CpuModel{ | |
| 339 | .name = "sm_86", | |
| 340 | .llvm_name = "sm_86", | |
| 341 | .features = featureSet(&[_]Feature{ | |
| 342 | .ptx71, | |
| 343 | .sm_86, | |
| 344 | }), | |
| 345 | }; | |
| 320 | 346 | }; |
lib/std/target/powerpc.zig+34-1| ... | ... | @@ -37,6 +37,7 @@ pub const Feature = enum { |
| 37 | 37 | htm, |
| 38 | 38 | icbt, |
| 39 | 39 | invariant_function_descriptors, |
| 40 | isa_v207_instructions, | |
| 40 | 41 | isa_v30_instructions, |
| 41 | 42 | isa_v31_instructions, |
| 42 | 43 | isel, |
| ... | ... | @@ -62,7 +63,10 @@ pub const Feature = enum { |
| 62 | 63 | ppc_prera_sched, |
| 63 | 64 | predictable_select_expensive, |
| 64 | 65 | prefix_instrs, |
| 66 | privileged, | |
| 67 | quadword_atomics, | |
| 65 | 68 | recipprec, |
| 69 | rop_protect, | |
| 66 | 70 | secure_plt, |
| 67 | 71 | slow_popcntd, |
| 68 | 72 | spe, |
| ... | ... | @@ -277,10 +281,17 @@ pub const all_features = blk: { |
| 277 | 281 | .description = "Assume function descriptors are invariant", |
| 278 | 282 | .dependencies = featureSet(&[_]Feature{}), |
| 279 | 283 | }; |
| 284 | result[@enumToInt(Feature.isa_v207_instructions)] = .{ | |
| 285 | .llvm_name = "isa-v207-instructions", | |
| 286 | .description = "Enable instructions in ISA 2.07.", | |
| 287 | .dependencies = featureSet(&[_]Feature{}), | |
| 288 | }; | |
| 280 | 289 | result[@enumToInt(Feature.isa_v30_instructions)] = .{ |
| 281 | 290 | .llvm_name = "isa-v30-instructions", |
| 282 | 291 | .description = "Enable instructions in ISA 3.0.", |
| 283 | .dependencies = featureSet(&[_]Feature{}), | |
| 292 | .dependencies = featureSet(&[_]Feature{ | |
| 293 | .isa_v207_instructions, | |
| 294 | }), | |
| 284 | 295 | }; |
| 285 | 296 | result[@enumToInt(Feature.isa_v31_instructions)] = .{ |
| 286 | 297 | .llvm_name = "isa-v31-instructions", |
| ... | ... | @@ -433,11 +444,26 @@ pub const all_features = blk: { |
| 433 | 444 | .power9_altivec, |
| 434 | 445 | }), |
| 435 | 446 | }; |
| 447 | result[@enumToInt(Feature.privileged)] = .{ | |
| 448 | .llvm_name = "privileged", | |
| 449 | .description = "Add privileged instructions", | |
| 450 | .dependencies = featureSet(&[_]Feature{}), | |
| 451 | }; | |
| 452 | result[@enumToInt(Feature.quadword_atomics)] = .{ | |
| 453 | .llvm_name = "quadword-atomics", | |
| 454 | .description = "Enable lqarx and stqcx.", | |
| 455 | .dependencies = featureSet(&[_]Feature{}), | |
| 456 | }; | |
| 436 | 457 | result[@enumToInt(Feature.recipprec)] = .{ |
| 437 | 458 | .llvm_name = "recipprec", |
| 438 | 459 | .description = "Assume higher precision reciprocal estimates", |
| 439 | 460 | .dependencies = featureSet(&[_]Feature{}), |
| 440 | 461 | }; |
| 462 | result[@enumToInt(Feature.rop_protect)] = .{ | |
| 463 | .llvm_name = "rop-protect", | |
| 464 | .description = "Add ROP protect", | |
| 465 | .dependencies = featureSet(&[_]Feature{}), | |
| 466 | }; | |
| 441 | 467 | result[@enumToInt(Feature.secure_plt)] = .{ |
| 442 | 468 | .llvm_name = "secure-plt", |
| 443 | 469 | .description = "Enable secure plt mode", |
| ... | ... | @@ -696,6 +722,7 @@ pub const cpu = struct { |
| 696 | 722 | .ppc_postra_sched, |
| 697 | 723 | .ppc_prera_sched, |
| 698 | 724 | .predictable_select_expensive, |
| 725 | .quadword_atomics, | |
| 699 | 726 | .recipprec, |
| 700 | 727 | .stfiwx, |
| 701 | 728 | .two_const_nr, |
| ... | ... | @@ -790,6 +817,7 @@ pub const cpu = struct { |
| 790 | 817 | .fuse_addis_load, |
| 791 | 818 | .htm, |
| 792 | 819 | .icbt, |
| 820 | .isa_v207_instructions, | |
| 793 | 821 | .isel, |
| 794 | 822 | .ldbrx, |
| 795 | 823 | .lfiwax, |
| ... | ... | @@ -798,6 +826,7 @@ pub const cpu = struct { |
| 798 | 826 | .popcntd, |
| 799 | 827 | .power8_vector, |
| 800 | 828 | .predictable_select_expensive, |
| 829 | .quadword_atomics, | |
| 801 | 830 | .recipprec, |
| 802 | 831 | .stfiwx, |
| 803 | 832 | .two_const_nr, |
| ... | ... | @@ -837,6 +866,7 @@ pub const cpu = struct { |
| 837 | 866 | .ppc_postra_sched, |
| 838 | 867 | .ppc_prera_sched, |
| 839 | 868 | .predictable_select_expensive, |
| 869 | .quadword_atomics, | |
| 840 | 870 | .recipprec, |
| 841 | 871 | .stfiwx, |
| 842 | 872 | .two_const_nr, |
| ... | ... | @@ -989,6 +1019,7 @@ pub const cpu = struct { |
| 989 | 1019 | .fuse_addis_load, |
| 990 | 1020 | .htm, |
| 991 | 1021 | .icbt, |
| 1022 | .isa_v207_instructions, | |
| 992 | 1023 | .isel, |
| 993 | 1024 | .ldbrx, |
| 994 | 1025 | .lfiwax, |
| ... | ... | @@ -997,6 +1028,7 @@ pub const cpu = struct { |
| 997 | 1028 | .popcntd, |
| 998 | 1029 | .power8_vector, |
| 999 | 1030 | .predictable_select_expensive, |
| 1031 | .quadword_atomics, | |
| 1000 | 1032 | .recipprec, |
| 1001 | 1033 | .stfiwx, |
| 1002 | 1034 | .two_const_nr, |
| ... | ... | @@ -1033,6 +1065,7 @@ pub const cpu = struct { |
| 1033 | 1065 | .ppc_postra_sched, |
| 1034 | 1066 | .ppc_prera_sched, |
| 1035 | 1067 | .predictable_select_expensive, |
| 1068 | .quadword_atomics, | |
| 1036 | 1069 | .recipprec, |
| 1037 | 1070 | .stfiwx, |
| 1038 | 1071 | .two_const_nr, |
lib/std/target/riscv.zig+1-1| ... | ... | @@ -185,7 +185,7 @@ pub const all_features = blk: { |
| 185 | 185 | }; |
| 186 | 186 | result[@enumToInt(Feature.experimental_zvamo)] = .{ |
| 187 | 187 | .llvm_name = "experimental-zvamo", |
| 188 | .description = "'Zvamo'(Vector AMO Operations)", | |
| 188 | .description = "'Zvamo' (Vector AMO Operations)", | |
| 189 | 189 | .dependencies = featureSet(&[_]Feature{ |
| 190 | 190 | .experimental_v, |
| 191 | 191 | }), |
lib/std/target/systemz.zig+76| ... | ... | @@ -5,6 +5,7 @@ const CpuFeature = std.Target.Cpu.Feature; |
| 5 | 5 | const CpuModel = std.Target.Cpu.Model; |
| 6 | 6 | |
| 7 | 7 | pub const Feature = enum { |
| 8 | bear_enhancement, | |
| 8 | 9 | deflate_conversion, |
| 9 | 10 | dfp_packed_conversion, |
| 10 | 11 | dfp_zoned_conversion, |
| ... | ... | @@ -31,8 +32,11 @@ pub const Feature = enum { |
| 31 | 32 | miscellaneous_extensions, |
| 32 | 33 | miscellaneous_extensions_2, |
| 33 | 34 | miscellaneous_extensions_3, |
| 35 | nnp_assist, | |
| 34 | 36 | population_count, |
| 37 | processor_activity_instrumentation, | |
| 35 | 38 | processor_assist, |
| 39 | reset_dat_protection, | |
| 36 | 40 | reset_reference_bits_multiple, |
| 37 | 41 | soft_float, |
| 38 | 42 | transactional_execution, |
| ... | ... | @@ -41,6 +45,7 @@ pub const Feature = enum { |
| 41 | 45 | vector_enhancements_2, |
| 42 | 46 | vector_packed_decimal, |
| 43 | 47 | vector_packed_decimal_enhancement, |
| 48 | vector_packed_decimal_enhancement_2, | |
| 44 | 49 | }; |
| 45 | 50 | |
| 46 | 51 | pub const featureSet = CpuFeature.feature_set_fns(Feature).featureSet; |
| ... | ... | @@ -52,6 +57,11 @@ pub const all_features = blk: { |
| 52 | 57 | const len = @typeInfo(Feature).Enum.fields.len; |
| 53 | 58 | std.debug.assert(len <= CpuFeature.Set.needed_bit_count); |
| 54 | 59 | var result: [len]CpuFeature = undefined; |
| 60 | result[@enumToInt(Feature.bear_enhancement)] = .{ | |
| 61 | .llvm_name = "bear-enhancement", | |
| 62 | .description = "Assume that the BEAR-enhancement facility is installed", | |
| 63 | .dependencies = featureSet(&[_]Feature{}), | |
| 64 | }; | |
| 55 | 65 | result[@enumToInt(Feature.deflate_conversion)] = .{ |
| 56 | 66 | .llvm_name = "deflate-conversion", |
| 57 | 67 | .description = "Assume that the deflate-conversion facility is installed", |
| ... | ... | @@ -182,16 +192,31 @@ pub const all_features = blk: { |
| 182 | 192 | .description = "Assume that the miscellaneous-extensions facility 3 is installed", |
| 183 | 193 | .dependencies = featureSet(&[_]Feature{}), |
| 184 | 194 | }; |
| 195 | result[@enumToInt(Feature.nnp_assist)] = .{ | |
| 196 | .llvm_name = "nnp-assist", | |
| 197 | .description = "Assume that the NNP-assist facility is installed", | |
| 198 | .dependencies = featureSet(&[_]Feature{}), | |
| 199 | }; | |
| 185 | 200 | result[@enumToInt(Feature.population_count)] = .{ |
| 186 | 201 | .llvm_name = "population-count", |
| 187 | 202 | .description = "Assume that the population-count facility is installed", |
| 188 | 203 | .dependencies = featureSet(&[_]Feature{}), |
| 189 | 204 | }; |
| 205 | result[@enumToInt(Feature.processor_activity_instrumentation)] = .{ | |
| 206 | .llvm_name = "processor-activity-instrumentation", | |
| 207 | .description = "Assume that the processor-activity-instrumentation facility is installed", | |
| 208 | .dependencies = featureSet(&[_]Feature{}), | |
| 209 | }; | |
| 190 | 210 | result[@enumToInt(Feature.processor_assist)] = .{ |
| 191 | 211 | .llvm_name = "processor-assist", |
| 192 | 212 | .description = "Assume that the processor-assist facility is installed", |
| 193 | 213 | .dependencies = featureSet(&[_]Feature{}), |
| 194 | 214 | }; |
| 215 | result[@enumToInt(Feature.reset_dat_protection)] = .{ | |
| 216 | .llvm_name = "reset-dat-protection", | |
| 217 | .description = "Assume that the reset-DAT-protection facility is installed", | |
| 218 | .dependencies = featureSet(&[_]Feature{}), | |
| 219 | }; | |
| 195 | 220 | result[@enumToInt(Feature.reset_reference_bits_multiple)] = .{ |
| 196 | 221 | .llvm_name = "reset-reference-bits-multiple", |
| 197 | 222 | .description = "Assume that the reset-reference-bits-multiple facility is installed", |
| ... | ... | @@ -232,6 +257,11 @@ pub const all_features = blk: { |
| 232 | 257 | .description = "Assume that the vector packed decimal enhancement facility is installed", |
| 233 | 258 | .dependencies = featureSet(&[_]Feature{}), |
| 234 | 259 | }; |
| 260 | result[@enumToInt(Feature.vector_packed_decimal_enhancement_2)] = .{ | |
| 261 | .llvm_name = "vector-packed-decimal-enhancement-2", | |
| 262 | .description = "Assume that the vector packed decimal enhancement facility 2 is installed", | |
| 263 | .dependencies = featureSet(&[_]Feature{}), | |
| 264 | }; | |
| 235 | 265 | const ti = @typeInfo(Feature); |
| 236 | 266 | for (result) |*elem, i| { |
| 237 | 267 | elem.index = i; |
| ... | ... | @@ -368,6 +398,52 @@ pub const cpu = struct { |
| 368 | 398 | .vector_packed_decimal_enhancement, |
| 369 | 399 | }), |
| 370 | 400 | }; |
| 401 | pub const arch14 = CpuModel{ | |
| 402 | .name = "arch14", | |
| 403 | .llvm_name = "arch14", | |
| 404 | .features = featureSet(&[_]Feature{ | |
| 405 | .bear_enhancement, | |
| 406 | .deflate_conversion, | |
| 407 | .dfp_packed_conversion, | |
| 408 | .dfp_zoned_conversion, | |
| 409 | .distinct_ops, | |
| 410 | .enhanced_dat_2, | |
| 411 | .enhanced_sort, | |
| 412 | .execution_hint, | |
| 413 | .fast_serialization, | |
| 414 | .fp_extension, | |
| 415 | .guarded_storage, | |
| 416 | .high_word, | |
| 417 | .insert_reference_bits_multiple, | |
| 418 | .interlocked_access1, | |
| 419 | .load_and_trap, | |
| 420 | .load_and_zero_rightmost_byte, | |
| 421 | .load_store_on_cond, | |
| 422 | .load_store_on_cond_2, | |
| 423 | .message_security_assist_extension3, | |
| 424 | .message_security_assist_extension4, | |
| 425 | .message_security_assist_extension5, | |
| 426 | .message_security_assist_extension7, | |
| 427 | .message_security_assist_extension8, | |
| 428 | .message_security_assist_extension9, | |
| 429 | .miscellaneous_extensions, | |
| 430 | .miscellaneous_extensions_2, | |
| 431 | .miscellaneous_extensions_3, | |
| 432 | .nnp_assist, | |
| 433 | .population_count, | |
| 434 | .processor_activity_instrumentation, | |
| 435 | .processor_assist, | |
| 436 | .reset_dat_protection, | |
| 437 | .reset_reference_bits_multiple, | |
| 438 | .transactional_execution, | |
| 439 | .vector, | |
| 440 | .vector_enhancements_1, | |
| 441 | .vector_enhancements_2, | |
| 442 | .vector_packed_decimal, | |
| 443 | .vector_packed_decimal_enhancement, | |
| 444 | .vector_packed_decimal_enhancement_2, | |
| 445 | }), | |
| 446 | }; | |
| 371 | 447 | pub const arch8 = CpuModel{ |
| 372 | 448 | .name = "arch8", |
| 373 | 449 | .llvm_name = "arch8", |
lib/std/target/wasm.zig-8| ... | ... | @@ -15,7 +15,6 @@ pub const Feature = enum { |
| 15 | 15 | sign_ext, |
| 16 | 16 | simd128, |
| 17 | 17 | tail_call, |
| 18 | unimplemented_simd128, | |
| 19 | 18 | }; |
| 20 | 19 | |
| 21 | 20 | pub const featureSet = CpuFeature.feature_set_fns(Feature).featureSet; |
| ... | ... | @@ -77,13 +76,6 @@ pub const all_features = blk: { |
| 77 | 76 | .description = "Enable tail call instructions", |
| 78 | 77 | .dependencies = featureSet(&[_]Feature{}), |
| 79 | 78 | }; |
| 80 | result[@enumToInt(Feature.unimplemented_simd128)] = .{ | |
| 81 | .llvm_name = "unimplemented-simd128", | |
| 82 | .description = "Enable 128-bit SIMD not yet implemented in engines", | |
| 83 | .dependencies = featureSet(&[_]Feature{ | |
| 84 | .simd128, | |
| 85 | }), | |
| 86 | }; | |
| 87 | 79 | const ti = @typeInfo(Feature); |
| 88 | 80 | for (result) |*elem, i| { |
| 89 | 81 | elem.index = i; |
lib/std/target/x86.zig+134-33| ... | ... | @@ -55,10 +55,12 @@ pub const Feature = enum { |
| 55 | 55 | fast_gather, |
| 56 | 56 | fast_hops, |
| 57 | 57 | fast_lzcnt, |
| 58 | fast_movbe, | |
| 58 | 59 | fast_scalar_fsqrt, |
| 59 | 60 | fast_scalar_shift_masks, |
| 60 | 61 | fast_shld_rotate, |
| 61 | fast_variable_shuffle, | |
| 62 | fast_variable_crosslane_shuffle, | |
| 63 | fast_variable_perlane_shuffle, | |
| 62 | 64 | fast_vector_fsqrt, |
| 63 | 65 | fast_vector_shift_masks, |
| 64 | 66 | fma, |
| ... | ... | @@ -458,6 +460,11 @@ pub const all_features = blk: { |
| 458 | 460 | .description = "LZCNT instructions are as fast as most simple integer ops", |
| 459 | 461 | .dependencies = featureSet(&[_]Feature{}), |
| 460 | 462 | }; |
| 463 | result[@enumToInt(Feature.fast_movbe)] = .{ | |
| 464 | .llvm_name = "fast-movbe", | |
| 465 | .description = "Prefer a movbe over a single-use load + bswap / single-use bswap + store", | |
| 466 | .dependencies = featureSet(&[_]Feature{}), | |
| 467 | }; | |
| 461 | 468 | result[@enumToInt(Feature.fast_scalar_fsqrt)] = .{ |
| 462 | 469 | .llvm_name = "fast-scalar-fsqrt", |
| 463 | 470 | .description = "Scalar SQRT is fast (disable Newton-Raphson)", |
| ... | ... | @@ -473,9 +480,14 @@ pub const all_features = blk: { |
| 473 | 480 | .description = "SHLD can be used as a faster rotate", |
| 474 | 481 | .dependencies = featureSet(&[_]Feature{}), |
| 475 | 482 | }; |
| 476 | result[@enumToInt(Feature.fast_variable_shuffle)] = .{ | |
| 477 | .llvm_name = "fast-variable-shuffle", | |
| 478 | .description = "Shuffles with variable masks are fast", | |
| 483 | result[@enumToInt(Feature.fast_variable_crosslane_shuffle)] = .{ | |
| 484 | .llvm_name = "fast-variable-crosslane-shuffle", | |
| 485 | .description = "Cross-lane shuffles with variable masks are fast", | |
| 486 | .dependencies = featureSet(&[_]Feature{}), | |
| 487 | }; | |
| 488 | result[@enumToInt(Feature.fast_variable_perlane_shuffle)] = .{ | |
| 489 | .llvm_name = "fast-variable-perlane-shuffle", | |
| 490 | .description = "Per-lane shuffles with variable masks are fast", | |
| 479 | 491 | .dependencies = featureSet(&[_]Feature{}), |
| 480 | 492 | }; |
| 481 | 493 | result[@enumToInt(Feature.fast_vector_fsqrt)] = .{ |
| ... | ... | @@ -1010,26 +1022,27 @@ pub const cpu = struct { |
| 1010 | 1022 | .features = featureSet(&[_]Feature{ |
| 1011 | 1023 | .@"64bit", |
| 1012 | 1024 | .adx, |
| 1013 | .aes, | |
| 1014 | 1025 | .avxvnni, |
| 1015 | 1026 | .bmi, |
| 1016 | 1027 | .bmi2, |
| 1017 | 1028 | .cldemote, |
| 1018 | 1029 | .clflushopt, |
| 1030 | .clwb, | |
| 1019 | 1031 | .cmov, |
| 1020 | 1032 | .cx16, |
| 1021 | .ermsb, | |
| 1022 | 1033 | .f16c, |
| 1023 | 1034 | .false_deps_popcnt, |
| 1024 | 1035 | .fast_15bytenop, |
| 1025 | 1036 | .fast_gather, |
| 1026 | 1037 | .fast_scalar_fsqrt, |
| 1027 | 1038 | .fast_shld_rotate, |
| 1028 | .fast_variable_shuffle, | |
| 1039 | .fast_variable_crosslane_shuffle, | |
| 1040 | .fast_variable_perlane_shuffle, | |
| 1029 | 1041 | .fast_vector_fsqrt, |
| 1030 | 1042 | .fma, |
| 1031 | 1043 | .fsgsbase, |
| 1032 | 1044 | .fxsr, |
| 1045 | .gfni, | |
| 1033 | 1046 | .hreset, |
| 1034 | 1047 | .idivq_to_divl, |
| 1035 | 1048 | .invpcid, |
| ... | ... | @@ -1037,19 +1050,27 @@ pub const cpu = struct { |
| 1037 | 1050 | .macrofusion, |
| 1038 | 1051 | .mmx, |
| 1039 | 1052 | .movbe, |
| 1053 | .movdir64b, | |
| 1054 | .movdiri, | |
| 1040 | 1055 | .nopl, |
| 1041 | .pclmul, | |
| 1056 | .pconfig, | |
| 1057 | .pku, | |
| 1042 | 1058 | .popcnt, |
| 1043 | 1059 | .prfchw, |
| 1044 | 1060 | .ptwrite, |
| 1061 | .rdpid, | |
| 1045 | 1062 | .rdrnd, |
| 1046 | 1063 | .rdseed, |
| 1047 | 1064 | .sahf, |
| 1048 | 1065 | .serialize, |
| 1049 | .sgx, | |
| 1066 | .sha, | |
| 1067 | .shstk, | |
| 1050 | 1068 | .slow_3ops_lea, |
| 1069 | .vaes, | |
| 1070 | .vpclmulqdq, | |
| 1051 | 1071 | .vzeroupper, |
| 1052 | 1072 | .waitpkg, |
| 1073 | .widekl, | |
| 1053 | 1074 | .x87, |
| 1054 | 1075 | .xsavec, |
| 1055 | 1076 | .xsaveopt, |
| ... | ... | @@ -1292,6 +1313,7 @@ pub const cpu = struct { |
| 1292 | 1313 | .f16c, |
| 1293 | 1314 | .fast_11bytenop, |
| 1294 | 1315 | .fast_bextr, |
| 1316 | .fast_movbe, | |
| 1295 | 1317 | .fast_scalar_shift_masks, |
| 1296 | 1318 | .fma, |
| 1297 | 1319 | .fxsr, |
| ... | ... | @@ -1324,6 +1346,7 @@ pub const cpu = struct { |
| 1324 | 1346 | .f16c, |
| 1325 | 1347 | .fast_11bytenop, |
| 1326 | 1348 | .fast_bextr, |
| 1349 | .fast_movbe, | |
| 1327 | 1350 | .fast_scalar_shift_masks, |
| 1328 | 1351 | .fma, |
| 1329 | 1352 | .fsgsbase, |
| ... | ... | @@ -1359,6 +1382,7 @@ pub const cpu = struct { |
| 1359 | 1382 | .f16c, |
| 1360 | 1383 | .fast_11bytenop, |
| 1361 | 1384 | .fast_bextr, |
| 1385 | .fast_movbe, | |
| 1362 | 1386 | .fast_scalar_shift_masks, |
| 1363 | 1387 | .fma, |
| 1364 | 1388 | .fsgsbase, |
| ... | ... | @@ -1424,7 +1448,8 @@ pub const cpu = struct { |
| 1424 | 1448 | .fast_15bytenop, |
| 1425 | 1449 | .fast_scalar_fsqrt, |
| 1426 | 1450 | .fast_shld_rotate, |
| 1427 | .fast_variable_shuffle, | |
| 1451 | .fast_variable_crosslane_shuffle, | |
| 1452 | .fast_variable_perlane_shuffle, | |
| 1428 | 1453 | .fma, |
| 1429 | 1454 | .fsgsbase, |
| 1430 | 1455 | .fxsr, |
| ... | ... | @@ -1485,6 +1510,7 @@ pub const cpu = struct { |
| 1485 | 1510 | .fast_bextr, |
| 1486 | 1511 | .fast_hops, |
| 1487 | 1512 | .fast_lzcnt, |
| 1513 | .fast_movbe, | |
| 1488 | 1514 | .fast_scalar_shift_masks, |
| 1489 | 1515 | .fast_vector_shift_masks, |
| 1490 | 1516 | .fxsr, |
| ... | ... | @@ -1548,7 +1574,8 @@ pub const cpu = struct { |
| 1548 | 1574 | .fast_gather, |
| 1549 | 1575 | .fast_scalar_fsqrt, |
| 1550 | 1576 | .fast_shld_rotate, |
| 1551 | .fast_variable_shuffle, | |
| 1577 | .fast_variable_crosslane_shuffle, | |
| 1578 | .fast_variable_perlane_shuffle, | |
| 1552 | 1579 | .fast_vector_fsqrt, |
| 1553 | 1580 | .fsgsbase, |
| 1554 | 1581 | .fxsr, |
| ... | ... | @@ -1567,7 +1594,6 @@ pub const cpu = struct { |
| 1567 | 1594 | .rdrnd, |
| 1568 | 1595 | .rdseed, |
| 1569 | 1596 | .sahf, |
| 1570 | .sgx, | |
| 1571 | 1597 | .sha, |
| 1572 | 1598 | .slow_3ops_lea, |
| 1573 | 1599 | .vzeroupper, |
| ... | ... | @@ -1601,7 +1627,8 @@ pub const cpu = struct { |
| 1601 | 1627 | .fast_gather, |
| 1602 | 1628 | .fast_scalar_fsqrt, |
| 1603 | 1629 | .fast_shld_rotate, |
| 1604 | .fast_variable_shuffle, | |
| 1630 | .fast_variable_crosslane_shuffle, | |
| 1631 | .fast_variable_perlane_shuffle, | |
| 1605 | 1632 | .fast_vector_fsqrt, |
| 1606 | 1633 | .fsgsbase, |
| 1607 | 1634 | .fxsr, |
| ... | ... | @@ -1652,7 +1679,8 @@ pub const cpu = struct { |
| 1652 | 1679 | .fast_gather, |
| 1653 | 1680 | .fast_scalar_fsqrt, |
| 1654 | 1681 | .fast_shld_rotate, |
| 1655 | .fast_variable_shuffle, | |
| 1682 | .fast_variable_crosslane_shuffle, | |
| 1683 | .fast_variable_perlane_shuffle, | |
| 1656 | 1684 | .fast_vector_fsqrt, |
| 1657 | 1685 | .fsgsbase, |
| 1658 | 1686 | .fxsr, |
| ... | ... | @@ -1714,7 +1742,8 @@ pub const cpu = struct { |
| 1714 | 1742 | .fast_15bytenop, |
| 1715 | 1743 | .fast_scalar_fsqrt, |
| 1716 | 1744 | .fast_shld_rotate, |
| 1717 | .fast_variable_shuffle, | |
| 1745 | .fast_variable_crosslane_shuffle, | |
| 1746 | .fast_variable_perlane_shuffle, | |
| 1718 | 1747 | .fma, |
| 1719 | 1748 | .fsgsbase, |
| 1720 | 1749 | .fxsr, |
| ... | ... | @@ -1844,6 +1873,7 @@ pub const cpu = struct { |
| 1844 | 1873 | .cmov, |
| 1845 | 1874 | .cx16, |
| 1846 | 1875 | .false_deps_popcnt, |
| 1876 | .fast_movbe, | |
| 1847 | 1877 | .fsgsbase, |
| 1848 | 1878 | .fxsr, |
| 1849 | 1879 | .mmx, |
| ... | ... | @@ -1877,6 +1907,7 @@ pub const cpu = struct { |
| 1877 | 1907 | .clflushopt, |
| 1878 | 1908 | .cmov, |
| 1879 | 1909 | .cx16, |
| 1910 | .fast_movbe, | |
| 1880 | 1911 | .fsgsbase, |
| 1881 | 1912 | .fxsr, |
| 1882 | 1913 | .mmx, |
| ... | ... | @@ -1890,7 +1921,6 @@ pub const cpu = struct { |
| 1890 | 1921 | .rdrnd, |
| 1891 | 1922 | .rdseed, |
| 1892 | 1923 | .sahf, |
| 1893 | .sgx, | |
| 1894 | 1924 | .sha, |
| 1895 | 1925 | .slow_incdec, |
| 1896 | 1926 | .slow_lea, |
| ... | ... | @@ -1921,7 +1951,8 @@ pub const cpu = struct { |
| 1921 | 1951 | .fast_15bytenop, |
| 1922 | 1952 | .fast_scalar_fsqrt, |
| 1923 | 1953 | .fast_shld_rotate, |
| 1924 | .fast_variable_shuffle, | |
| 1954 | .fast_variable_crosslane_shuffle, | |
| 1955 | .fast_variable_perlane_shuffle, | |
| 1925 | 1956 | .fma, |
| 1926 | 1957 | .fsgsbase, |
| 1927 | 1958 | .fxsr, |
| ... | ... | @@ -1960,7 +1991,6 @@ pub const cpu = struct { |
| 1960 | 1991 | .bmi, |
| 1961 | 1992 | .bmi2, |
| 1962 | 1993 | .clflushopt, |
| 1963 | .clwb, | |
| 1964 | 1994 | .cmov, |
| 1965 | 1995 | .cx16, |
| 1966 | 1996 | .ermsb, |
| ... | ... | @@ -1968,7 +1998,8 @@ pub const cpu = struct { |
| 1968 | 1998 | .fast_gather, |
| 1969 | 1999 | .fast_scalar_fsqrt, |
| 1970 | 2000 | .fast_shld_rotate, |
| 1971 | .fast_variable_shuffle, | |
| 2001 | .fast_variable_crosslane_shuffle, | |
| 2002 | .fast_variable_perlane_shuffle, | |
| 1972 | 2003 | .fast_vector_fsqrt, |
| 1973 | 2004 | .fsgsbase, |
| 1974 | 2005 | .fsrm, |
| ... | ... | @@ -1989,7 +2020,6 @@ pub const cpu = struct { |
| 1989 | 2020 | .rdrnd, |
| 1990 | 2021 | .rdseed, |
| 1991 | 2022 | .sahf, |
| 1992 | .sgx, | |
| 1993 | 2023 | .sha, |
| 1994 | 2024 | .slow_3ops_lea, |
| 1995 | 2025 | .vaes, |
| ... | ... | @@ -2027,7 +2057,8 @@ pub const cpu = struct { |
| 2027 | 2057 | .fast_gather, |
| 2028 | 2058 | .fast_scalar_fsqrt, |
| 2029 | 2059 | .fast_shld_rotate, |
| 2030 | .fast_variable_shuffle, | |
| 2060 | .fast_variable_crosslane_shuffle, | |
| 2061 | .fast_variable_perlane_shuffle, | |
| 2031 | 2062 | .fast_vector_fsqrt, |
| 2032 | 2063 | .fsgsbase, |
| 2033 | 2064 | .fsrm, |
| ... | ... | @@ -2049,7 +2080,6 @@ pub const cpu = struct { |
| 2049 | 2080 | .rdrnd, |
| 2050 | 2081 | .rdseed, |
| 2051 | 2082 | .sahf, |
| 2052 | .sgx, | |
| 2053 | 2083 | .sha, |
| 2054 | 2084 | .slow_3ops_lea, |
| 2055 | 2085 | .vaes, |
| ... | ... | @@ -2175,6 +2205,7 @@ pub const cpu = struct { |
| 2175 | 2205 | .cmov, |
| 2176 | 2206 | .cx16, |
| 2177 | 2207 | .fast_gather, |
| 2208 | .fast_movbe, | |
| 2178 | 2209 | .fsgsbase, |
| 2179 | 2210 | .fxsr, |
| 2180 | 2211 | .idivq_to_divl, |
| ... | ... | @@ -2214,6 +2245,7 @@ pub const cpu = struct { |
| 2214 | 2245 | .cmov, |
| 2215 | 2246 | .cx16, |
| 2216 | 2247 | .fast_gather, |
| 2248 | .fast_movbe, | |
| 2217 | 2249 | .fsgsbase, |
| 2218 | 2250 | .fxsr, |
| 2219 | 2251 | .idivq_to_divl, |
| ... | ... | @@ -2472,6 +2504,64 @@ pub const cpu = struct { |
| 2472 | 2504 | .x87, |
| 2473 | 2505 | }), |
| 2474 | 2506 | }; |
| 2507 | pub const rocketlake = CpuModel{ | |
| 2508 | .name = "rocketlake", | |
| 2509 | .llvm_name = "rocketlake", | |
| 2510 | .features = featureSet(&[_]Feature{ | |
| 2511 | .@"64bit", | |
| 2512 | .adx, | |
| 2513 | .avx512bitalg, | |
| 2514 | .avx512cd, | |
| 2515 | .avx512dq, | |
| 2516 | .avx512ifma, | |
| 2517 | .avx512vbmi, | |
| 2518 | .avx512vbmi2, | |
| 2519 | .avx512vl, | |
| 2520 | .avx512vnni, | |
| 2521 | .avx512vpopcntdq, | |
| 2522 | .bmi, | |
| 2523 | .bmi2, | |
| 2524 | .clflushopt, | |
| 2525 | .cmov, | |
| 2526 | .cx16, | |
| 2527 | .ermsb, | |
| 2528 | .fast_15bytenop, | |
| 2529 | .fast_gather, | |
| 2530 | .fast_scalar_fsqrt, | |
| 2531 | .fast_shld_rotate, | |
| 2532 | .fast_variable_crosslane_shuffle, | |
| 2533 | .fast_variable_perlane_shuffle, | |
| 2534 | .fast_vector_fsqrt, | |
| 2535 | .fsgsbase, | |
| 2536 | .fsrm, | |
| 2537 | .fxsr, | |
| 2538 | .gfni, | |
| 2539 | .idivq_to_divl, | |
| 2540 | .invpcid, | |
| 2541 | .lzcnt, | |
| 2542 | .macrofusion, | |
| 2543 | .mmx, | |
| 2544 | .movbe, | |
| 2545 | .nopl, | |
| 2546 | .pku, | |
| 2547 | .popcnt, | |
| 2548 | .prefer_256_bit, | |
| 2549 | .prfchw, | |
| 2550 | .rdpid, | |
| 2551 | .rdrnd, | |
| 2552 | .rdseed, | |
| 2553 | .sahf, | |
| 2554 | .sha, | |
| 2555 | .slow_3ops_lea, | |
| 2556 | .vaes, | |
| 2557 | .vpclmulqdq, | |
| 2558 | .vzeroupper, | |
| 2559 | .x87, | |
| 2560 | .xsavec, | |
| 2561 | .xsaveopt, | |
| 2562 | .xsaves, | |
| 2563 | }), | |
| 2564 | }; | |
| 2475 | 2565 | pub const sandybridge = CpuModel{ |
| 2476 | 2566 | .name = "sandybridge", |
| 2477 | 2567 | .llvm_name = "sandybridge", |
| ... | ... | @@ -2532,7 +2622,8 @@ pub const cpu = struct { |
| 2532 | 2622 | .fast_gather, |
| 2533 | 2623 | .fast_scalar_fsqrt, |
| 2534 | 2624 | .fast_shld_rotate, |
| 2535 | .fast_variable_shuffle, | |
| 2625 | .fast_variable_crosslane_shuffle, | |
| 2626 | .fast_variable_perlane_shuffle, | |
| 2536 | 2627 | .fast_vector_fsqrt, |
| 2537 | 2628 | .fsgsbase, |
| 2538 | 2629 | .fsrm, |
| ... | ... | @@ -2558,7 +2649,6 @@ pub const cpu = struct { |
| 2558 | 2649 | .rdseed, |
| 2559 | 2650 | .sahf, |
| 2560 | 2651 | .serialize, |
| 2561 | .sgx, | |
| 2562 | 2652 | .sha, |
| 2563 | 2653 | .shstk, |
| 2564 | 2654 | .slow_3ops_lea, |
| ... | ... | @@ -2584,6 +2674,7 @@ pub const cpu = struct { |
| 2584 | 2674 | .cx16, |
| 2585 | 2675 | .false_deps_popcnt, |
| 2586 | 2676 | .fast_7bytenop, |
| 2677 | .fast_movbe, | |
| 2587 | 2678 | .fxsr, |
| 2588 | 2679 | .idivq_to_divl, |
| 2589 | 2680 | .mmx, |
| ... | ... | @@ -2626,7 +2717,8 @@ pub const cpu = struct { |
| 2626 | 2717 | .fast_gather, |
| 2627 | 2718 | .fast_scalar_fsqrt, |
| 2628 | 2719 | .fast_shld_rotate, |
| 2629 | .fast_variable_shuffle, | |
| 2720 | .fast_variable_crosslane_shuffle, | |
| 2721 | .fast_variable_perlane_shuffle, | |
| 2630 | 2722 | .fast_vector_fsqrt, |
| 2631 | 2723 | .fsgsbase, |
| 2632 | 2724 | .fxsr, |
| ... | ... | @@ -2673,7 +2765,8 @@ pub const cpu = struct { |
| 2673 | 2765 | .fast_gather, |
| 2674 | 2766 | .fast_scalar_fsqrt, |
| 2675 | 2767 | .fast_shld_rotate, |
| 2676 | .fast_variable_shuffle, | |
| 2768 | .fast_variable_crosslane_shuffle, | |
| 2769 | .fast_variable_perlane_shuffle, | |
| 2677 | 2770 | .fast_vector_fsqrt, |
| 2678 | 2771 | .fma, |
| 2679 | 2772 | .fsgsbase, |
| ... | ... | @@ -2691,7 +2784,6 @@ pub const cpu = struct { |
| 2691 | 2784 | .rdrnd, |
| 2692 | 2785 | .rdseed, |
| 2693 | 2786 | .sahf, |
| 2694 | .sgx, | |
| 2695 | 2787 | .slow_3ops_lea, |
| 2696 | 2788 | .vzeroupper, |
| 2697 | 2789 | .x87, |
| ... | ... | @@ -2723,7 +2815,8 @@ pub const cpu = struct { |
| 2723 | 2815 | .fast_gather, |
| 2724 | 2816 | .fast_scalar_fsqrt, |
| 2725 | 2817 | .fast_shld_rotate, |
| 2726 | .fast_variable_shuffle, | |
| 2818 | .fast_variable_crosslane_shuffle, | |
| 2819 | .fast_variable_perlane_shuffle, | |
| 2727 | 2820 | .fast_vector_fsqrt, |
| 2728 | 2821 | .fsgsbase, |
| 2729 | 2822 | .fxsr, |
| ... | ... | @@ -2759,6 +2852,7 @@ pub const cpu = struct { |
| 2759 | 2852 | .cx16, |
| 2760 | 2853 | .false_deps_popcnt, |
| 2761 | 2854 | .fast_7bytenop, |
| 2855 | .fast_movbe, | |
| 2762 | 2856 | .fxsr, |
| 2763 | 2857 | .idivq_to_divl, |
| 2764 | 2858 | .mmx, |
| ... | ... | @@ -2805,7 +2899,8 @@ pub const cpu = struct { |
| 2805 | 2899 | .fast_gather, |
| 2806 | 2900 | .fast_scalar_fsqrt, |
| 2807 | 2901 | .fast_shld_rotate, |
| 2808 | .fast_variable_shuffle, | |
| 2902 | .fast_variable_crosslane_shuffle, | |
| 2903 | .fast_variable_perlane_shuffle, | |
| 2809 | 2904 | .fast_vector_fsqrt, |
| 2810 | 2905 | .fsgsbase, |
| 2811 | 2906 | .fsrm, |
| ... | ... | @@ -2828,7 +2923,6 @@ pub const cpu = struct { |
| 2828 | 2923 | .rdrnd, |
| 2829 | 2924 | .rdseed, |
| 2830 | 2925 | .sahf, |
| 2831 | .sgx, | |
| 2832 | 2926 | .sha, |
| 2833 | 2927 | .shstk, |
| 2834 | 2928 | .slow_3ops_lea, |
| ... | ... | @@ -2851,6 +2945,7 @@ pub const cpu = struct { |
| 2851 | 2945 | .clwb, |
| 2852 | 2946 | .cmov, |
| 2853 | 2947 | .cx16, |
| 2948 | .fast_movbe, | |
| 2854 | 2949 | .fsgsbase, |
| 2855 | 2950 | .fxsr, |
| 2856 | 2951 | .gfni, |
| ... | ... | @@ -2865,7 +2960,6 @@ pub const cpu = struct { |
| 2865 | 2960 | .rdrnd, |
| 2866 | 2961 | .rdseed, |
| 2867 | 2962 | .sahf, |
| 2868 | .sgx, | |
| 2869 | 2963 | .sha, |
| 2870 | 2964 | .slow_incdec, |
| 2871 | 2965 | .slow_lea, |
| ... | ... | @@ -2978,7 +3072,8 @@ pub const cpu = struct { |
| 2978 | 3072 | .fast_15bytenop, |
| 2979 | 3073 | .fast_scalar_fsqrt, |
| 2980 | 3074 | .fast_shld_rotate, |
| 2981 | .fast_variable_shuffle, | |
| 3075 | .fast_variable_crosslane_shuffle, | |
| 3076 | .fast_variable_perlane_shuffle, | |
| 2982 | 3077 | .fma, |
| 2983 | 3078 | .fxsr, |
| 2984 | 3079 | .idivq_to_divl, |
| ... | ... | @@ -3013,7 +3108,8 @@ pub const cpu = struct { |
| 3013 | 3108 | .fast_gather, |
| 3014 | 3109 | .fast_scalar_fsqrt, |
| 3015 | 3110 | .fast_shld_rotate, |
| 3016 | .fast_variable_shuffle, | |
| 3111 | .fast_variable_crosslane_shuffle, | |
| 3112 | .fast_variable_perlane_shuffle, | |
| 3017 | 3113 | .fast_vector_fsqrt, |
| 3018 | 3114 | .fxsr, |
| 3019 | 3115 | .idivq_to_divl, |
| ... | ... | @@ -3065,6 +3161,7 @@ pub const cpu = struct { |
| 3065 | 3161 | .fast_15bytenop, |
| 3066 | 3162 | .fast_bextr, |
| 3067 | 3163 | .fast_lzcnt, |
| 3164 | .fast_movbe, | |
| 3068 | 3165 | .fast_scalar_shift_masks, |
| 3069 | 3166 | .fma, |
| 3070 | 3167 | .fsgsbase, |
| ... | ... | @@ -3110,6 +3207,7 @@ pub const cpu = struct { |
| 3110 | 3207 | .fast_15bytenop, |
| 3111 | 3208 | .fast_bextr, |
| 3112 | 3209 | .fast_lzcnt, |
| 3210 | .fast_movbe, | |
| 3113 | 3211 | .fast_scalar_shift_masks, |
| 3114 | 3212 | .fma, |
| 3115 | 3213 | .fsgsbase, |
| ... | ... | @@ -3156,13 +3254,16 @@ pub const cpu = struct { |
| 3156 | 3254 | .fast_15bytenop, |
| 3157 | 3255 | .fast_bextr, |
| 3158 | 3256 | .fast_lzcnt, |
| 3257 | .fast_movbe, | |
| 3159 | 3258 | .fast_scalar_shift_masks, |
| 3259 | .fast_variable_perlane_shuffle, | |
| 3160 | 3260 | .fma, |
| 3161 | 3261 | .fsgsbase, |
| 3162 | 3262 | .fsrm, |
| 3163 | 3263 | .fxsr, |
| 3164 | 3264 | .invpcid, |
| 3165 | 3265 | .lzcnt, |
| 3266 | .macrofusion, | |
| 3166 | 3267 | .mmx, |
| 3167 | 3268 | .movbe, |
| 3168 | 3269 | .mwaitx, |
tools/update_cpu_features.zig-2| ... | ... | @@ -219,7 +219,6 @@ const llvm_targets = [_]LlvmTarget{ |
| 219 | 219 | "use_postra_scheduler", |
| 220 | 220 | "use_reciprocal_square_root", |
| 221 | 221 | "v8a", |
| 222 | "zcz_fp", | |
| 223 | 222 | }, |
| 224 | 223 | }, |
| 225 | 224 | .{ |
| ... | ... | @@ -236,7 +235,6 @@ const llvm_targets = [_]LlvmTarget{ |
| 236 | 235 | "slow_paired_128", |
| 237 | 236 | "use_postra_scheduler", |
| 238 | 237 | "v8a", |
| 239 | "zcz_fp", | |
| 240 | 238 | }, |
| 241 | 239 | }, |
| 242 | 240 | .{ |