Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(383)

Side by Side Diff: test/cctest/test-assembler-arm.cc

Issue 2620343002: [ARM] Add vand, vorr NEON instructions. (Closed)
Patch Set: Created 3 years, 11 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View unified diff | Download patch
OLDNEW
1 // Copyright 2012 the V8 project authors. All rights reserved. 1 // Copyright 2012 the V8 project authors. All rights reserved.
2 // Redistribution and use in source and binary forms, with or without 2 // Redistribution and use in source and binary forms, with or without
3 // modification, are permitted provided that the following conditions are 3 // modification, are permitted provided that the following conditions are
4 // met: 4 // met:
5 // 5 //
6 // * Redistributions of source code must retain the above copyright 6 // * Redistributions of source code must retain the above copyright
7 // notice, this list of conditions and the following disclaimer. 7 // notice, this list of conditions and the following disclaimer.
8 // * Redistributions in binary form must reproduce the above 8 // * Redistributions in binary form must reproduce the above
9 // copyright notice, this list of conditions and the following 9 // copyright notice, this list of conditions and the following
10 // disclaimer in the documentation and/or other materials provided 10 // disclaimer in the documentation and/or other materials provided
(...skipping 1266 matching lines...) Expand 10 before | Expand all | Expand 10 after
1277 uint32_t vmov_from_scalar_s16, vmov_from_scalar_u16; 1277 uint32_t vmov_from_scalar_s16, vmov_from_scalar_u16;
1278 uint32_t vmov_from_scalar_32; 1278 uint32_t vmov_from_scalar_32;
1279 uint32_t vmov[4], vmvn[4]; 1279 uint32_t vmov[4], vmvn[4];
1280 int32_t vcvt_s32_f32[4]; 1280 int32_t vcvt_s32_f32[4];
1281 uint32_t vcvt_u32_f32[4]; 1281 uint32_t vcvt_u32_f32[4];
1282 float vcvt_f32_s32[4], vcvt_f32_u32[4]; 1282 float vcvt_f32_s32[4], vcvt_f32_u32[4];
1283 uint32_t vdup8[4], vdup16[4], vdup32[4]; 1283 uint32_t vdup8[4], vdup16[4], vdup32[4];
1284 float vabsf[4], vnegf[4]; 1284 float vabsf[4], vnegf[4];
1285 uint32_t vabs_s8[4], vabs_s16[4], vabs_s32[4]; 1285 uint32_t vabs_s8[4], vabs_s16[4], vabs_s32[4];
1286 uint32_t vneg_s8[4], vneg_s16[4], vneg_s32[4]; 1286 uint32_t vneg_s8[4], vneg_s16[4], vneg_s32[4];
1287 uint32_t veor[4]; 1287 uint32_t veor[4], vand[4], vorr[4];
1288 float vdupf[4], vaddf[4], vsubf[4], vmulf[4]; 1288 float vdupf[4], vaddf[4], vsubf[4], vmulf[4];
1289 uint32_t vadd8[4], vadd16[4], vadd32[4]; 1289 uint32_t vadd8[4], vadd16[4], vadd32[4];
1290 uint32_t vsub8[4], vsub16[4], vsub32[4]; 1290 uint32_t vsub8[4], vsub16[4], vsub32[4];
1291 uint32_t vmul8[4], vmul16[4], vmul32[4]; 1291 uint32_t vmul8[4], vmul16[4], vmul32[4];
1292 float vrecpe[4], vrecps[4], vrsqrte[4], vrsqrts[4]; 1292 float vrecpe[4], vrecps[4], vrsqrte[4], vrsqrts[4];
1293 uint32_t vtst[4], vceq[4], vceqf[4], vbsl[4]; 1293 uint32_t vtst[4], vceq[4], vceqf[4], vbsl[4];
1294 uint32_t vext[4]; 1294 uint32_t vext[4];
1295 uint32_t vzip8a[4], vzip8b[4], vzip16a[4], vzip16b[4], vzip32a[4], 1295 uint32_t vzip8a[4], vzip8b[4], vzip16a[4], vzip16b[4], vzip32a[4],
1296 vzip32b[4]; 1296 vzip32b[4];
1297 uint32_t vrev64_32[4], vrev64_16[4], vrev64_8[4]; 1297 uint32_t vrev64_32[4], vrev64_16[4], vrev64_8[4];
(...skipping 148 matching lines...) Expand 10 before | Expand all | Expand 10 after
1446 __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vneg_s8)))); 1446 __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vneg_s8))));
1447 __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4)); 1447 __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
1448 __ vneg(Neon16, q1, q0); 1448 __ vneg(Neon16, q1, q0);
1449 __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vneg_s16)))); 1449 __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vneg_s16))));
1450 __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4)); 1450 __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
1451 __ vneg(Neon32, q1, q0); 1451 __ vneg(Neon32, q1, q0);
1452 __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vneg_s32)))); 1452 __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vneg_s32))));
1453 __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4)); 1453 __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
1454 1454
1455 // veor. 1455 // veor.
1456 __ mov(r4, Operand(0x00aa)); 1456 __ mov(r4, Operand(0xaa));
1457 __ vdup(Neon16, q0, r4); 1457 __ vdup(Neon16, q0, r4);
1458 __ mov(r4, Operand(0x0055)); 1458 __ mov(r4, Operand(0x55));
1459 __ vdup(Neon16, q1, r4); 1459 __ vdup(Neon16, q1, r4);
1460 __ veor(q1, q1, q0); 1460 __ veor(q1, q1, q0);
1461 __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, veor)))); 1461 __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, veor))));
1462 __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4)); 1462 __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
1463 // vand.
1464 __ mov(r4, Operand(0xff));
1465 __ vdup(Neon16, q0, r4);
1466 __ mov(r4, Operand(0xfe));
1467 __ vdup(Neon16, q1, r4);
1468 __ vand(q1, q1, q0);
1469 __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vand))));
1470 __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
1471 // vorr.
1472 __ mov(r4, Operand(0xaa));
1473 __ vdup(Neon16, q0, r4);
1474 __ mov(r4, Operand(0x55));
1475 __ vdup(Neon16, q1, r4);
1476 __ vorr(q1, q1, q0);
1477 __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vorr))));
1478 __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
1463 1479
1464 // vadd (float). 1480 // vadd (float).
1465 __ vmov(s4, 1.0); 1481 __ vmov(s4, 1.0);
1466 __ vdup(q0, s4); 1482 __ vdup(q0, s4);
1467 __ vdup(q1, s4); 1483 __ vdup(q1, s4);
1468 __ vadd(q1, q1, q0); 1484 __ vadd(q1, q1, q0);
1469 __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vaddf)))); 1485 __ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vaddf))));
1470 __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4)); 1486 __ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
1471 // vsub (float). 1487 // vsub (float).
1472 __ vmov(s4, 2.0); 1488 __ vmov(s4, 2.0);
(...skipping 296 matching lines...) Expand 10 before | Expand all | Expand 10 after
1769 CHECK_EQ_32X4(vnegf, 1.0, 0.0, -0.0, -1.0); 1785 CHECK_EQ_32X4(vnegf, 1.0, 0.0, -0.0, -1.0);
1770 // src: [0x7f7f7f7f, 0x01010101, 0xffffffff, 0x80808080] 1786 // src: [0x7f7f7f7f, 0x01010101, 0xffffffff, 0x80808080]
1771 CHECK_EQ_32X4(vabs_s8, 0x7f7f7f7fu, 0x01010101u, 0x01010101u, 0x80808080u); 1787 CHECK_EQ_32X4(vabs_s8, 0x7f7f7f7fu, 0x01010101u, 0x01010101u, 0x80808080u);
1772 CHECK_EQ_32X4(vabs_s16, 0x7f7f7f7fu, 0x01010101u, 0x00010001u, 0x7f807f80u); 1788 CHECK_EQ_32X4(vabs_s16, 0x7f7f7f7fu, 0x01010101u, 0x00010001u, 0x7f807f80u);
1773 CHECK_EQ_32X4(vabs_s32, 0x7f7f7f7fu, 0x01010101u, 0x00000001u, 0x7f7f7f80u); 1789 CHECK_EQ_32X4(vabs_s32, 0x7f7f7f7fu, 0x01010101u, 0x00000001u, 0x7f7f7f80u);
1774 CHECK_EQ_32X4(vneg_s8, 0x81818181u, 0xffffffffu, 0x01010101u, 0x80808080u); 1790 CHECK_EQ_32X4(vneg_s8, 0x81818181u, 0xffffffffu, 0x01010101u, 0x80808080u);
1775 CHECK_EQ_32X4(vneg_s16, 0x80818081u, 0xfefffeffu, 0x00010001u, 0x7f807f80u); 1791 CHECK_EQ_32X4(vneg_s16, 0x80818081u, 0xfefffeffu, 0x00010001u, 0x7f807f80u);
1776 CHECK_EQ_32X4(vneg_s32, 0x80808081u, 0xfefefeffu, 0x00000001u, 0x7f7f7f80u); 1792 CHECK_EQ_32X4(vneg_s32, 0x80808081u, 0xfefefeffu, 0x00000001u, 0x7f7f7f80u);
1777 1793
1778 CHECK_EQ_SPLAT(veor, 0x00ff00ffu); 1794 CHECK_EQ_SPLAT(veor, 0x00ff00ffu);
1795 CHECK_EQ_SPLAT(vand, 0x00fe00feu);
1796 CHECK_EQ_SPLAT(vorr, 0x00ff00ffu);
1779 CHECK_EQ_SPLAT(vaddf, 2.0); 1797 CHECK_EQ_SPLAT(vaddf, 2.0);
1780 CHECK_EQ_SPLAT(vsubf, -1.0); 1798 CHECK_EQ_SPLAT(vsubf, -1.0);
1781 CHECK_EQ_SPLAT(vmulf, 4.0); 1799 CHECK_EQ_SPLAT(vmulf, 4.0);
1782 CHECK_EQ_SPLAT(vrecpe, 0.5f); // 1 / 2 1800 CHECK_EQ_SPLAT(vrecpe, 0.5f); // 1 / 2
1783 CHECK_EQ_SPLAT(vrecps, -1.0f); // 2 - (2 * 1.5) 1801 CHECK_EQ_SPLAT(vrecps, -1.0f); // 2 - (2 * 1.5)
1784 CHECK_EQ_SPLAT(vrsqrte, 0.5f); // 1 / sqrt(4) 1802 CHECK_EQ_SPLAT(vrsqrte, 0.5f); // 1 / sqrt(4)
1785 CHECK_EQ_SPLAT(vrsqrts, -1.0f); // (3 - (2 * 2.5)) / 2 1803 CHECK_EQ_SPLAT(vrsqrts, -1.0f); // (3 - (2 * 2.5)) / 2
1786 CHECK_EQ_SPLAT(vceqf, 0xffffffffu); 1804 CHECK_EQ_SPLAT(vceqf, 0xffffffffu);
1787 CHECK_EQ_SPLAT(vadd8, 0x03030303u); 1805 CHECK_EQ_SPLAT(vadd8, 0x03030303u);
1788 CHECK_EQ_SPLAT(vadd16, 0x00030003u); 1806 CHECK_EQ_SPLAT(vadd16, 0x00030003u);
(...skipping 1707 matching lines...) Expand 10 before | Expand all | Expand 10 after
3496 HandleScope scope(isolate); 3514 HandleScope scope(isolate);
3497 3515
3498 Assembler assm(isolate, NULL, 0); 3516 Assembler assm(isolate, NULL, 0);
3499 __ mov(r0, Operand(isolate->factory()->infinity_value())); 3517 __ mov(r0, Operand(isolate->factory()->infinity_value()));
3500 __ BlockConstPoolFor(1019); 3518 __ BlockConstPoolFor(1019);
3501 for (int i = 0; i < 1019; ++i) __ nop(); 3519 for (int i = 0; i < 1019; ++i) __ nop();
3502 __ vldr(d0, MemOperand(r0, 0)); 3520 __ vldr(d0, MemOperand(r0, 0));
3503 } 3521 }
3504 3522
3505 #undef __ 3523 #undef __
OLDNEW

Powered by Google App Engine
This is Rietveld 408576698