[PULL,v2,04/44] Hexagon (target/hexagon) Add v68 HVX instructions

Message ID	20230518200411.271148-5-tsimpson@quicinc.com
State	New
Headers	show Return-Path: <qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org> From: Taylor Simpson <tsimpson@quicinc.com> To: qemu-devel@nongnu.org Cc: tsimpson@quicinc.com, richard.henderson@linaro.org, philmd@linaro.org, peter.maydell@linaro.org, bcain@quicinc.com, quic_mathbern@quicinc.com, stefanha@redhat.com, ale@rev.ng, anjo@rev.ng, quic_mliebel@quicinc.com Subject: [PULL v2 04/44] Hexagon (target/hexagon) Add v68 HVX instructions Date: Thu, 18 May 2023 13:03:31 -0700 Message-Id: <20230518200411.271148-5-tsimpson@quicinc.com> In-Reply-To: <20230518200411.271148-1-tsimpson@quicinc.com> References: <20230518200411.271148-1-tsimpson@quicinc.com> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: base64 Received-SPF: pass client-ip=205.220.180.131; envelope-from=tsimpson@qualcomm.com; helo=mx0b-0031df01.pphosted.com X-Spam_score_int: -17 X-Spam_score: -1.8 X-Spam_bar: - X-Spam_report: (-1.8 / 5.0 requ) BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, HEADER_FROM_DIFFERENT_DOMAINS=0.25, SPF_PASS=-0.001, T_SCC_BODY_TEXT_LINE=-0.01, T_SPF_HELO_TEMPERROR=0.01 autolearn=no autolearn_force=no X-Spam_action: no action Precedence: list Errors-To: qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org Sender: qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org
Series	[PULL,v2,01/44] Hexagon (target/hexagon) Add support for v68/v69/v71/v73 \| expand [PULL,v2,01/44] Hexagon (target/hexagon) Add support for v68/v69/v71/v73 [PULL,v2,02/44] Hexagon (target/hexagon) Add v68 scalar instructions [PULL,v2,03/44] Hexagon (tests/tcg/hexagon) Add v68 scalar tests [PULL,v2,04/44] Hexagon (target/hexagon) Add v68 HVX instructions [PULL,v2,05/44] Hexagon (tests/tcg/hexagon) Add v68 HVX tests [PULL,v2,06/44] Hexagon (target/hexagon) Add v69 HVX instructions [PULL,v2,07/44] Hexagon (tests/tcg/hexagon) Add v69 HVX tests [PULL,v2,08/44] Hexagon (target/hexagon) Add v73 scalar instructions [PULL,v2,09/44] Hexagon (tests/tcg/hexagon) Add v73 scalar tests [PULL,v2,10/44] meson.build Add CONFIG_HEXAGON_IDEF_PARSER [PULL,v2,11/44] Hexagon (target/hexagon) Add DisasContext arg to gen_log_reg_write [PULL,v2,12/44] Hexagon (target/hexagon) Add overrides for loop setup instructions [PULL,v2,13/44] Hexagon (target/hexagon) Add overrides for allocframe/deallocframe [PULL,v2,14/44] Hexagon (target/hexagon) Add overrides for clr[tf]new [PULL,v2,15/44] Hexagon (target/hexagon) Remove log_reg_write from op_helper.[ch] [PULL,v2,16/44] Hexagon (target/hexagon) Eliminate uses of log_pred_write function [PULL,v2,17/44] Hexagon (target/hexagon) Clean up pred_written usage [PULL,v2,18/44] Hexagon (target/hexagon) Don't overlap dest writes with source reads [PULL,v2,19/44] Hexagon (target/hexagon) Mark registers as read during packet analysis [PULL,v2,20/44] Hexagon (target/hexagon) Short-circuit packet register writes [PULL,v2,21/44] Hexagon (target/hexagon) Short-circuit packet predicate writes [PULL,v2,22/44] Hexagon (target/hexagon) Short-circuit packet HVX writes [PULL,v2,23/44] Hexagon (target/hexagon) Short-circuit more HVX single instruction packets [PULL,v2,24/44] Hexagon (target/hexagon) Add overrides for disabled idef-parser insns [PULL,v2,25/44] Hexagon (target/hexagon) Make special new_value for USR [PULL,v2,26/44] Hexagon (target/hexagon) Move new_value to DisasContext [PULL,v2,27/44] Hexagon (target/hexagon) Move new_pred_value to DisasContext [PULL,v2,28/44] Hexagon (target/hexagon) Move pred_written to DisasContext [PULL,v2,29/44] Hexagon (target/hexagon) Move pkt_has_store_s1 to DisasContext [PULL,v2,30/44] Hexagon (target/hexagon) Move items to DisasContext [PULL,v2,31/44] Hexagon (target/hexagon) Additional instructions handled by idef-parser [PULL,v2,32/44] target/hexagon: fix = vs. == mishap [PULL,v2,33/44] Hexagon (target/hexagon/*.py): raise exception on reg parsing error [PULL,v2,34/44] Hexagon: list available CPUs with `-cpu help` [PULL,v2,35/44] Hexagon: append eflags to unknown cpu model string [PULL,v2,36/44] Hexagon (iclass): update J4_hintjumpr slot constraints [PULL,v2,37/44] Hexagon (decode): look for pkts with multiple insns at the same slot [PULL,v2,38/44] Remove test_vshuff from hvx_misc tests [PULL,v2,39/44] gdbstub: only send stop-reply packets when allowed to [PULL,v2,40/44] gdbstub: add test for untimely stop-reply packets [PULL,v2,41/44] Hexagon: add core gdbstub xml data for LLDB [PULL,v2,42/44] Hexagon (gdbstub): fix p3:0 read and write via stub [PULL,v2,43/44] Hexagon (gdbstub): add HVX support [PULL,v2,44/44] Hexagon (linux-user/hexagon): handle breakpoints

diff --git a/target/hexagon/mmvec/macros.h b/target/hexagon/mmvec/macros.h index 1201d778d0..a655634fd1 100644 --- a/target/hexagon/mmvec/macros.h +++ b/target/hexagon/mmvec/macros.h @@ -1,5 +1,5 @@ /* - * Copyright(c) 2019-2022 Qualcomm Innovation Center, Inc. All Rights Reserved. + * Copyright(c) 2019-2023 Qualcomm Innovation Center, Inc. All Rights Reserved. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -346,4 +346,11 @@ #define fUARCH_NOTE_PUMP_2X() #define IV1DEAD() + +#define fGET10BIT(COE, VAL, POS) \ + do { \ + COE = (sextract32(VAL, 24 + 2 * POS, 2) << 8) | \ + extract32(VAL, POS * 8, 8); \ + } while (0); + #endif diff --git a/target/hexagon/imported/mmvec/encode_ext.def b/target/hexagon/imported/mmvec/encode_ext.def index 6fbbe2c422..b9b62fef8d 100644 --- a/target/hexagon/imported/mmvec/encode_ext.def +++ b/target/hexagon/imported/mmvec/encode_ext.def @@ -1,5 +1,5 @@ /* - * Copyright(c) 2019-2021 Qualcomm Innovation Center, Inc. All Rights Reserved. + * Copyright(c) 2019-2023 Qualcomm Innovation Center, Inc. All Rights Reserved. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -730,6 +730,8 @@ DEF_ENC(V6_vmaxb, ICLASS_CJ" 1 111 001 vvvvv PP 0 uuuuu 101 ddddd") // DEF_ENC(V6_vsatuwuh, ICLASS_CJ" 1 111 001 vvvvv PP 0 uuuuu 110 ddddd") // DEF_ENC(V6_vdealb4w, ICLASS_CJ" 1 111 001 vvvvv PP 0 uuuuu 111 ddddd") // +DEF_ENC(V6_v6mpyvubs10_vxx, ICLASS_CJ" 1 111 001 vvvvv PP 1 uuuuu 0ii xxxxx") +DEF_ENC(V6_v6mpyhubs10_vxx, ICLASS_CJ" 1 111 001 vvvvv PP 1 uuuuu 1ii xxxxx") DEF_ENC(V6_vmpyowh_rnd, ICLASS_CJ" 1 111 010 vvvvv PP 0 uuuuu 000 ddddd") // DEF_ENC(V6_vshuffeb, ICLASS_CJ" 1 111 010 vvvvv PP 0 uuuuu 001 ddddd") // @@ -740,6 +742,10 @@ DEF_ENC(V6_vshufoeh, ICLASS_CJ" 1 111 010 vvvvv PP 0 uuuuu 101 ddddd") // DEF_ENC(V6_vshufoeb, ICLASS_CJ" 1 111 010 vvvvv PP 0 uuuuu 110 ddddd") // DEF_ENC(V6_vcombine, ICLASS_CJ" 1 111 010 vvvvv PP 0 uuuuu 111 ddddd") // +DEF_ENC(V6_v6mpyvubs10, ICLASS_CJ" 1 111 010 vvvvv PP 1 uuuuu 0ii ddddd") +DEF_ENC(V6_v6mpyhubs10, ICLASS_CJ" 1 111 010 vvvvv PP 1 uuuuu 1ii ddddd") + + DEF_ENC(V6_vmpyieoh, ICLASS_CJ" 1 111 011 vvvvv PP 0 uuuuu 000 ddddd") // DEF_ENC(V6_vadduwsat, ICLASS_CJ" 1 111 011 vvvvv PP 0 uuuuu 001 ddddd") // DEF_ENC(V6_vsathub, ICLASS_CJ" 1 111 011 vvvvv PP 0 uuuuu 010 ddddd") // diff --git a/target/hexagon/imported/mmvec/ext.idef b/target/hexagon/imported/mmvec/ext.idef index 8ca5a606e1..c0d169fd4f 100644 --- a/target/hexagon/imported/mmvec/ext.idef +++ b/target/hexagon/imported/mmvec/ext.idef @@ -1,5 +1,5 @@ /* - * Copyright(c) 2019-2021 Qualcomm Innovation Center, Inc. All Rights Reserved. + * Copyright(c) 2019-2023 Qualcomm Innovation Center, Inc. All Rights Reserved. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -116,6 +116,10 @@ ITERATOR_INSN_MPY_SLOT_LATE(WIDTH,TAG, SYNTAX2,DESCR,CODE) EXTINSN(V6_##TAG, SYNTAX, ATTRIBS(A_EXTENSION,A_CVI,A_CVI_VX_DV), \ DESCR, DO_FOR_EACH_CODE(WIDTH, CODE)) +#define ITERATOR_INSN_MPY_SLOT_DOUBLE_VEC_VX_FWD(WIDTH,TAG,SYNTAX,DESCR,CODE) \ +EXTINSN(V6_##TAG, SYNTAX, ATTRIBS(A_EXTENSION,A_CVI,A_CVI_VX_DV), \ +DESCR, DO_FOR_EACH_CODE(WIDTH, CODE)) + #define ITERATOR_INSN2_MPY_SLOT_DOUBLE_VEC(WIDTH,TAG,SYNTAX,SYNTAX2,DESCR,CODE) \ ITERATOR_INSN_MPY_SLOT_DOUBLE_VEC(WIDTH,TAG,SYNTAX2,DESCR,CODE) @@ -2507,6 +2511,281 @@ EXTINSN(V6_vscattermhw , "vscatter(Rt32,Mu2,Vvv32.w).h=Vw32", ATTRIBS(A_EXTENSIO }) +ITERATOR_INSN_MPY_SLOT_DOUBLE_VEC_VX_FWD(32, v6mpyvubs10_vxx, "Vxx32.w+=v6mpy(Vuu32.ub,Vvv32.b,#u2):v", "", + fHIDE(size2s_t c00;) + fGET10BIT(c00, VvvV.v[0].uw[i], 0) + fHIDE(size2s_t c01;) + fGET10BIT(c01, VvvV.v[0].uw[i], 1) + fHIDE(size2s_t c02;) + fGET10BIT(c02, VvvV.v[0].uw[i], 2) + + fHIDE(size2s_t c10;) + fGET10BIT(c10, VvvV.v[1].uw[i], 0) + fHIDE(size2s_t c11;) + fGET10BIT(c11, VvvV.v[1].uw[i], 1) + fHIDE(size2s_t c12;) + fGET10BIT(c12, VvvV.v[1].uw[i], 2) + + if (uiV == 0) { + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c10); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c11); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[1].uw[i]), c12); + + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c00); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c01); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c02); + + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c10); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c11); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c12); + + } else if (uiV == 1) { + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c00); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c01); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[1].uw[i]), c02); + + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c10); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c11); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[1].uw[i]), c12); + + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c00); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c01); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c02); + + } else if (uiV == 2) { + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c10); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c11); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c12); + + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[0].uw[i]), c00); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c01); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c02); + + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[0].uw[i]), c10); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c11); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c12); + + } else if (uiV == 3) { + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c00); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c01); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c02); + + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c10); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c11); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c12); + + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[0].uw[i]), c00); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c01); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c02); + } +) +ITERATOR_INSN_MPY_SLOT_DOUBLE_VEC_VX_FWD(32, v6mpyhubs10_vxx, "Vxx32.w+=v6mpy(Vuu32.ub,Vvv32.b,#u2):h", "", + fHIDE(size2s_t c00;) + fGET10BIT(c00, VvvV.v[0].uw[i], 0) + fHIDE(size2s_t c01;) + fGET10BIT(c01, VvvV.v[0].uw[i], 1) + fHIDE(size2s_t c02;) + fGET10BIT(c02, VvvV.v[0].uw[i], 2) + fHIDE(size2s_t c10;) + fGET10BIT(c10, VvvV.v[1].uw[i], 0) + fHIDE(size2s_t c11;) + fGET10BIT(c11, VvvV.v[1].uw[i], 1) + fHIDE(size2s_t c12;) + fGET10BIT(c12, VvvV.v[1].uw[i], 2) + + if (uiV == 0) { + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[1].uw[i]), c10); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c11); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c12); + + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c00); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c01); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c02); + + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c10); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c11); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c12); + + } else if (uiV == 1) { + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[1].uw[i]), c00); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c01); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c02); + + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[1].uw[i]), c10); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c11); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c12); + + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c00); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c01); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c02); + + } else if (uiV == 2) { + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c10); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c11); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c12); + + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c00); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c01); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[0].uw[i]), c02); + + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c10); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c11); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[0].uw[i]), c12); + + } else if (uiV == 3) { + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c00); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c01); + VxxV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c02); + + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c10); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c11); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c12); + + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c00); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c01); + VxxV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[0].uw[i]), c02); + } +) + + +ITERATOR_INSN_MPY_SLOT_DOUBLE_VEC(32, v6mpyvubs10, "Vdd32.w=v6mpy(Vuu32.ub,Vvv32.b,#u2):v", "", + fHIDE(short c00;) + fGET10BIT(c00, VvvV.v[0].uw[i], 0) + fHIDE(short c01;) + fGET10BIT(c01, VvvV.v[0].uw[i], 1) + fHIDE(short c02;) + fGET10BIT(c02, VvvV.v[0].uw[i], 2) + fHIDE(short c10;) + fGET10BIT(c10, VvvV.v[1].uw[i], 0) + fHIDE(short c11;) + fGET10BIT(c11, VvvV.v[1].uw[i], 1) + fHIDE(short c12;) + fGET10BIT(c12, VvvV.v[1].uw[i], 2) + + + + if (uiV == 0) { + VddV.v[1].w[i] = fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c10); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c11); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[1].uw[i]), c12); + + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c00); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c01); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c02); + + VddV.v[0].w[i] = fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c10); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c11); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c12); + + } else if (uiV == 1) { + VddV.v[1].w[i] = fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c00); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c01); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[1].uw[i]), c02); + + VddV.v[0].w[i] = fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c10); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c11); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[1].uw[i]), c12); + + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c00); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c01); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c02); + + } else if (uiV == 2) { + VddV.v[1].w[i] = fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c10); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c11); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c12); + + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[0].uw[i]), c00); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c01); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c02); + + VddV.v[0].w[i] = fMPY16US(fGETUBYTE(0,VuuV.v[0].uw[i]), c10); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c11); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c12); + + } else if (uiV == 3) { + VddV.v[1].w[i] = fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c00); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c01); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c02); + + VddV.v[0].w[i] = fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c10); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c11); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c12); + + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[0].uw[i]), c00); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c01); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c02); + } +) + +ITERATOR_INSN_MPY_SLOT_DOUBLE_VEC(32, v6mpyhubs10, "Vdd32.w=v6mpy(Vuu32.ub,Vvv32.b,#u2):h", "", + fHIDE(short c00;) + fGET10BIT(c00, VvvV.v[0].uw[i], 0) + fHIDE(short c01;) + fGET10BIT(c01, VvvV.v[0].uw[i], 1) + fHIDE(short c02;) + fGET10BIT(c02, VvvV.v[0].uw[i], 2) + fHIDE(short c10;) + fGET10BIT(c10, VvvV.v[1].uw[i], 0) + fHIDE(short c11;) + fGET10BIT(c11, VvvV.v[1].uw[i], 1) + fHIDE(short c12;) + fGET10BIT(c12, VvvV.v[1].uw[i], 2) + + if (uiV == 0) { + VddV.v[1].w[i] = fMPY16US(fGETUBYTE(3,VuuV.v[1].uw[i]), c10); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c11); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c12); + + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c00); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c01); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c02); + + VddV.v[0].w[i] = fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c10); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c11); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c12); + + } else if (uiV == 1) { + VddV.v[1].w[i] = fMPY16US(fGETUBYTE(3,VuuV.v[1].uw[i]), c00); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c01); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c02); + + VddV.v[0].w[i] = fMPY16US(fGETUBYTE(3,VuuV.v[1].uw[i]), c10); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c11); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c12); + + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[1].uw[i]), c00); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c01); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c02); + + } else if (uiV == 2) { + VddV.v[1].w[i] = fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c10); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c11); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c12); + + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c00); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c01); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[0].uw[i]), c02); + + VddV.v[0].w[i] = fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c10); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c11); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[0].uw[i]), c12); + + } else if (uiV == 3) { + VddV.v[1].w[i] = fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c00); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c01); + VddV.v[1].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c02); + + VddV.v[0].w[i] = fMPY16US(fGETUBYTE(1,VuuV.v[1].uw[i]), c10); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(3,VuuV.v[0].uw[i]), c11); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(1,VuuV.v[0].uw[i]), c12); + + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[1].uw[i]), c00); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(2,VuuV.v[0].uw[i]), c01); + VddV.v[0].w[i] += fMPY16US(fGETUBYTE(0,VuuV.v[0].uw[i]), c02); + } +) + EXTINSN(V6_vscattermhwq, "if (Qs4) vscatter(Rt32,Mu2,Vvv32.w).h=Vw32", ATTRIBS(A_EXTENSION,A_CVI,A_CVI_SCATTER,A_CVI_VA_DV,A_CVI_VM,A_MEMLIKE), "Scatter halfwords conditional", {

[PULL,v2,04/44] Hexagon (target/hexagon) Add v68 HVX instructions

Commit Message

Patch