@scrypted/libav 1.0.180 → 1.0.181

package/openh264/codec/common/loongarch/intra_pred_com_lasx.c

@@ -0,0 +1,112 @@
+/*!
+ **********************************************************************************
+ * Copyright (c) 2022 Loongson Technology Corporation Limited
+ * Contributed by Lu Wang <wanglu@loongson.cn>
+ *
+ * \copy
+ *     Copyright (c)  2009-2013, Cisco Systems
+ *     All rights reserved.
+ *
+ *     Redistribution and use in source and binary forms, with or without
+ *     modification, are permitted provided that the following conditions
+ *     are met:
+ *
+ *        * Redistributions of source code must retain the above copyright
+ *          notice, this list of conditions and the following disclaimer.
+ *
+ *        * Redistributions in binary form must reproduce the above copyright
+ *          notice, this list of conditions and the following disclaimer in
+ *          the documentation and/or other materials provided with the
+ *          distribution.
+ *
+ *     THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *     "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *     LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
+ *     FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
+ *     COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
+ *     INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
+ *     BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ *     LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+ *     CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ *     LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
+ *     ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ *     POSSIBILITY OF SUCH DAMAGE.
+ *
+ * \file    intra_pred_com_lasx.c
+ *
+ * \brief   Loongson optimization
+ *
+ * \date    13/10/2022 Created
+ *
+ **********************************************************************************
+ */
+
+#include <stdint.h>
+#include "loongson_intrinsics.h"
+
+void WelsIChromaPredV_lasx (uint8_t* pPred, uint8_t* pRef, const int32_t kiStride) {
+  __m256i vec_kuiSrc64 = __lasx_xvldrepl_d (pRef - kiStride, 0);
+
+  __lasx_xvst(vec_kuiSrc64, pPred, 0);
+  __lasx_xvst(vec_kuiSrc64, pPred, 32);
+}
+
+void WelsIChromaPredH_lasx (uint8_t* pPred, uint8_t* pRef, const int32_t kiStride) {
+  __m256i vec_kuiSrc0, vec_kuiSrc1;
+  int32_t iStride_x2 = (kiStride << 1);
+  int32_t iStride_x3 = (kiStride << 1) + kiStride;
+  int32_t iStride_x4 = (kiStride << 2);
+
+  pRef -= 1;
+  vec_kuiSrc0 = __lasx_xvldrepl_b(pRef + kiStride, 0);
+  vec_kuiSrc1 = __lasx_xvldrepl_b(pRef, 0);
+  vec_kuiSrc0 = __lasx_xvilvl_d(vec_kuiSrc0, vec_kuiSrc1);
+  __lasx_xvst(vec_kuiSrc0, pPred, 0);
+
+  vec_kuiSrc0 = __lasx_xvldrepl_b(pRef + iStride_x3, 0);
+  vec_kuiSrc1 = __lasx_xvldrepl_b(pRef + iStride_x2, 0);
+  vec_kuiSrc0 = __lasx_xvilvl_d(vec_kuiSrc0, vec_kuiSrc1);
+  __lasx_xvst(vec_kuiSrc0, pPred, 16);
+
+  pRef += iStride_x4;
+  vec_kuiSrc0 = __lasx_xvldrepl_b(pRef + kiStride, 0);
+  vec_kuiSrc1 = __lasx_xvldrepl_b(pRef, 0);
+  vec_kuiSrc0 = __lasx_xvilvl_d(vec_kuiSrc0, vec_kuiSrc1);
+  __lasx_xvst(vec_kuiSrc0, pPred, 32);
+
+  vec_kuiSrc0 = __lasx_xvldrepl_b(pRef + iStride_x3, 0);
+  vec_kuiSrc1 = __lasx_xvldrepl_b(pRef + iStride_x2, 0);
+  vec_kuiSrc0 = __lasx_xvilvl_d(vec_kuiSrc0, vec_kuiSrc1);
+  __lasx_xvst(vec_kuiSrc0, pPred, 48);
+}
+
+void WelsIChromaPredDc_lasx (uint8_t* pPred, uint8_t* pRef, const int32_t kiStride) {
+  const int32_t kuiL1 = kiStride - 1;
+  const int32_t kuiL2 = kuiL1 + kiStride;
+  const int32_t kuiL3 = kuiL2 + kiStride;
+  const int32_t kuiL4 = kuiL3 + kiStride;
+  const int32_t kuiL5 = kuiL4 + kiStride;
+  const int32_t kuiL6 = kuiL5 + kiStride;
+  const int32_t kuiL7 = kuiL6 + kiStride;
+  /*caculate the iMean value*/
+  const uint8_t kuiMean1 = (pRef[-kiStride] + pRef[1 - kiStride] + pRef[2 - kiStride] +
+                           pRef[3 - kiStride] + pRef[-1] + pRef[kuiL1] + pRef[kuiL2] +
+                           pRef[kuiL3] + 4) >> 3;
+  const uint32_t kuiSum2 = pRef[4 - kiStride] + pRef[5 - kiStride] + pRef[6 - kiStride]
+                           + pRef[7 - kiStride];
+  const uint32_t kuiSum3 = pRef[kuiL4] + pRef[kuiL5] + pRef[kuiL6] + pRef[kuiL7];
+  const uint8_t kuiMean2 = (kuiSum2 + 2) >> 2;
+  const uint8_t kuiMean3 = (kuiSum3 + 2) >> 2;
+  const uint8_t kuiMean4 = (kuiSum2 + kuiSum3 + 4) >> 3;
+
+  const uint8_t kuiTopMean[8] = {kuiMean1, kuiMean1, kuiMean1, kuiMean1, kuiMean2,
+                                 kuiMean2, kuiMean2, kuiMean2};
+  const uint8_t kuiBottomMean[8] = {kuiMean3, kuiMean3, kuiMean3, kuiMean3, kuiMean4,
+                                    kuiMean4, kuiMean4, kuiMean4};
+
+  __m256i vec_kuiTopMean64 = __lasx_xvldrepl_d(kuiTopMean, 0);
+  __m256i vec_kuiBottomMean64 = __lasx_xvldrepl_d(kuiBottomMean, 0);
+
+  __lasx_xvst(vec_kuiTopMean64, pPred, 0);
+  __lasx_xvst(vec_kuiBottomMean64, pPred, 32);
+}

package/openh264/codec/common/loongarch/intra_pred_com_lsx.c

@@ -0,0 +1,91 @@
+/*!
+ **********************************************************************************
+ * Copyright (c) 2022 Loongson Technology Corporation Limited
+ * Contributed by Lu Wang <wanglu@loongson.cn>
+ *
+ * \copy
+ *     Copyright (c)  2009-2013, Cisco Systems
+ *     All rights reserved.
+ *
+ *     Redistribution and use in source and binary forms, with or without
+ *     modification, are permitted provided that the following conditions
+ *     are met:
+ *
+ *        * Redistributions of source code must retain the above copyright
+ *          notice, this list of conditions and the following disclaimer.
+ *
+ *        * Redistributions in binary form must reproduce the above copyright
+ *          notice, this list of conditions and the following disclaimer in
+ *          the documentation and/or other materials provided with the
+ *          distribution.
+ *
+ *     THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *     "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *     LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
+ *     FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
+ *     COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
+ *     INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
+ *     BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ *     LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+ *     CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ *     LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
+ *     ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ *     POSSIBILITY OF SUCH DAMAGE.
+ *
+ *
+ * \file    intra_pred_com_lsx.c
+ *
+ * \brief   Loongson optimization
+ *
+ * \date    03/03/2022 Created
+ *
+ *************************************************************************************
+ */
+#include <stdint.h>
+#include "loongson_intrinsics.h"
+
+void WelsI16x16LumaPredV_lsx (uint8_t* pPred, uint8_t* pRef, const int32_t kiStride) {
+  const int8_t* kpSrc = (int8_t*)&pRef[-kiStride];
+  const uint64_t kuiT1 = *(uint64_t*)kpSrc;
+  const uint64_t kuiT2 = *(uint64_t*)(kpSrc + 8);
+  uint8_t* pDst = pPred;
+  __m128i kuiT_vec, kuiT1_vec, kuiT2_vec;
+
+  kuiT1_vec = __lsx_vreplgr2vr_d(kuiT1);
+  kuiT2_vec = __lsx_vreplgr2vr_d(kuiT2);
+  kuiT_vec = __lsx_vpackev_d(kuiT2_vec, kuiT1_vec);
+
+  __lsx_vst(kuiT_vec, pDst, 0);
+  __lsx_vstx(kuiT_vec, pDst, 16);
+  __lsx_vstx(kuiT_vec, pDst, 32);
+  __lsx_vstx(kuiT_vec, pDst, 48);
+  __lsx_vstx(kuiT_vec, pDst, 64);
+  __lsx_vstx(kuiT_vec, pDst, 80);
+  __lsx_vstx(kuiT_vec, pDst, 96);
+  __lsx_vstx(kuiT_vec, pDst, 112);
+  __lsx_vstx(kuiT_vec, pDst, 128);
+  __lsx_vstx(kuiT_vec, pDst, 144);
+  __lsx_vstx(kuiT_vec, pDst, 160);
+  __lsx_vstx(kuiT_vec, pDst, 176);
+  __lsx_vstx(kuiT_vec, pDst, 192);
+  __lsx_vstx(kuiT_vec, pDst, 208);
+  __lsx_vstx(kuiT_vec, pDst, 224);
+  __lsx_vstx(kuiT_vec, pDst, 240);
+}
+
+void WelsI16x16LumaPredH_lsx (uint8_t* pPred, uint8_t* pRef, const int32_t kiStride) {
+  int32_t iStridex15 = (kiStride << 4) - kiStride;
+  int32_t iPredStride = 16;
+  int32_t iPredStridex15 = 240; //(iPredStride<<4)-iPredStride;
+  uint8_t i = 15;
+  __m128i kuiV64_vec;
+
+  do {
+    const uint8_t kuiSrc8 = pRef[iStridex15 - 1];
+    kuiV64_vec = __lsx_vreplgr2vr_b(kuiSrc8);
+    __lsx_vstx(kuiV64_vec, pPred, iPredStridex15);
+
+    iStridex15 -= kiStride;
+    iPredStridex15 -= iPredStride;
+  } while (i-- > 0);
+}

package/openh264/codec/common/loongarch/mc_chroma_lsx.c

@@ -0,0 +1,192 @@
+/*!
+ **********************************************************************************
+ * Copyright (c) 2021 Loongson Technology Corporation Limited
+ * Contributed by Lu Wang <wanglu@loongson.cn>
+ *
+ * \copy
+ *     Copyright (c)  2013, Cisco Systems
+ *     All rights reserved.
+ *
+ *     Redistribution and use in source and binary forms, with or without
+ *     modification, are permitted provided that the following conditions
+ *     are met:
+ *
+ *        * Redistributions of source code must retain the above copyright
+ *          notice, this list of conditions and the following disclaimer.
+ *
+ *        * Redistributions in binary form must reproduce the above copyright
+ *          notice, this list of conditions and the following disclaimer in
+ *          the documentation and/or other materials provided with the
+ *          distribution.
+ *
+ *     THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *     "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *     LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
+ *     FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
+ *     COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
+ *     INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
+ *     BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ *     LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+ *     CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ *     LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
+ *     ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ *     POSSIBILITY OF SUCH DAMAGE.
+ *
+ * \file    mc_chroma_lsx.c
+ *
+ * \brief   Loongson optimization
+ *
+ * \date    12/23/2021 Created
+ *
+ **********************************************************************************
+ */
+
+#include "stdint.h"
+#include "loongson_intrinsics.h"
+
+void McCopyWidthEq4_lsx(const uint8_t *pSrc, int iSrcStride,
+                         uint8_t *pDst, int iDstStride, int iHeight) {
+  __m128i src0, src1;
+  int iSrcStride_x2 = iSrcStride << 1;
+  int iDstStride_x2 = iDstStride << 1;
+  iHeight = iHeight >> 1;
+  for (int i = 0; i < iHeight; i++) {
+    DUP2_ARG2(__lsx_vld, pSrc, 0, pSrc + iSrcStride, 0, src0, src1);
+    __lsx_vstelm_w(src0, pDst, 0, 0);
+    __lsx_vstelm_w(src1, pDst + iDstStride, 0, 0);
+    pSrc += iSrcStride_x2;
+    pDst += iDstStride_x2;
+  }
+}
+
+void McCopyWidthEq8_lsx(const uint8_t *pSrc, int iSrcStride,
+                        uint8_t *pDst, int iDstStride, int iHeight) {
+  __m128i src0, src1;
+  int iSrcStride_x2 = iSrcStride << 1;
+  int iDstStride_x2 = iDstStride << 1;
+  iHeight = iHeight >> 1;
+  for (int i = 0; i < iHeight; i++) {
+    DUP2_ARG2(__lsx_vld, pSrc, 0, pSrc + iSrcStride, 0, src0, src1);
+    __lsx_vstelm_d(src0, pDst, 0, 0);
+    __lsx_vstelm_d(src1, pDst + iDstStride, 0, 0);
+    pSrc += iSrcStride_x2;
+    pDst += iDstStride_x2;
+  }
+}
+
+void McCopyWidthEq16_lsx(const uint8_t *pSrc, int iSrcStride,
+                         uint8_t *pDst, int iDstStride, int iHeight) {
+  __m128i src0, src1;
+  int iSrcStride_x2 = iSrcStride << 1;
+  int iDstStride_x2 = iDstStride << 1;
+  iHeight = iHeight >> 1;
+  for (int i = 0; i < iHeight; i++) {
+    DUP2_ARG2(__lsx_vld, pSrc, 0, pSrc + iSrcStride, 0, src0, src1);
+    __lsx_vst(src0, pDst, 0);
+    __lsx_vstx(src1, pDst, iDstStride);
+    pSrc += iSrcStride_x2;
+    pDst += iDstStride_x2;
+  }
+}
+
+void McChromaWidthEq4_lsx(const uint8_t *pSrc, int32_t iSrcStride, uint8_t *pDst,
+                          int32_t iDstStride, const uint8_t *pABCD, int32_t iHeight) {
+  int32_t i;
+  int8_t iA, iB, iC, iD;
+  uint16_t iFillData;
+  int32_t iSrcStride_x2 = iSrcStride << 1;
+  int32_t iDstStride_x2 = iDstStride << 1;
+  __m128i paramAB, paramCD, paramConst;
+  __m128i src0, src1, src2, src0Even, src1Even, src2Even;
+  __m128i out0, outNext0, out1, outNext1;
+  __m128i shift;
+
+  iA = pABCD[0];
+  iB = pABCD[1];
+  iC = pABCD[2];
+  iD = pABCD[3];
+
+  iFillData = (iB << 8) | iA;
+  paramAB = __lsx_vreplgr2vr_h(iFillData);
+  iFillData = (iD << 8) | iC;
+  paramCD = __lsx_vreplgr2vr_h(iFillData);
+  paramConst = __lsx_vreplgr2vr_h(32);
+  shift = __lsx_vreplgr2vr_h(6);
+  iHeight = iHeight >> 1;
+
+  for (i = 0; i < iHeight; i++) {
+    DUP2_ARG2(__lsx_vld, pSrc, 0, pSrc + iSrcStride, 0, src0, src1);
+    src2 = __lsx_vld(pSrc + iSrcStride_x2, 0);
+    pSrc += 1;
+    DUP2_ARG2(__lsx_vld, pSrc, 0, pSrc + iSrcStride, 0, src0Even, src1Even);
+    src2Even = __lsx_vld(pSrc + iSrcStride_x2, 0);
+    pSrc -= 1;
+
+    DUP2_ARG2(__lsx_vilvl_w, src0Even, src0, src1Even, src1, src0, src1);
+    src2 = __lsx_vilvl_w(src2Even, src2);
+    DUP4_ARG2(__lsx_vdp2_h_bu, src0, paramAB, src1, paramCD, src1, paramAB, src2,
+              paramCD, out0, outNext0, out1, outNext1);
+    DUP4_ARG2(__lsx_vadd_h, out0, outNext0, out0, paramConst, out1, outNext1, out1,
+              paramConst, out0, out0, out1, out1);
+    DUP2_ARG2(__lsx_vsrl_h, out0, shift, out1, shift, out0, out1);
+    DUP2_ARG2(__lsx_vsrl_d, out0, paramConst, out1, paramConst, outNext0, outNext1);
+    DUP2_ARG2(__lsx_vpackev_b, outNext0, out0, outNext1, out1, out0, out1);
+    __lsx_vstelm_w(out0, pDst, 0, 0);
+    __lsx_vstelm_w(out1, pDst + iDstStride, 0, 0);
+
+    pDst += iDstStride_x2;
+    pSrc += iSrcStride_x2;
+  }
+}
+
+void McChromaWidthEq8_lsx(const uint8_t *pSrc, int32_t iSrcStride, uint8_t *pDst,
+                          int32_t iDstStride, const uint8_t *pABCD, int32_t iHeight) {
+  int32_t i;
+  int8_t iA, iB, iC, iD;
+  uint16_t iFillData;
+  int32_t iSrcStride_x2 = iSrcStride << 1;
+  int32_t iDstStride_x2 = iDstStride << 1;
+  __m128i paramAB, paramCD, paramConst;
+  __m128i src0, src1, src2, src0Even, src1Even, src2Even;
+  __m128i out0, outNext0, out1, outNext1;
+  __m128i shift;
+
+  iA = pABCD[0];
+  iB = pABCD[1];
+  iC = pABCD[2];
+  iD = pABCD[3];
+
+  iFillData = (iB << 8) | iA;
+  paramAB = __lsx_vreplgr2vr_h(iFillData);
+  iFillData = (iD << 8) | iC;
+  paramCD = __lsx_vreplgr2vr_h(iFillData);
+  paramConst = __lsx_vreplgr2vr_h(32);
+  shift = __lsx_vreplgr2vr_h(6);
+  iHeight = iHeight >> 1;
+
+  for (i = 0; i < iHeight; i++) {
+    DUP2_ARG2(__lsx_vld, pSrc, 0, pSrc + iSrcStride, 0, src0, src1);
+    src2 = __lsx_vld(pSrc + iSrcStride_x2, 0);
+    pSrc += 1;
+    DUP2_ARG2(__lsx_vld, pSrc, 0, pSrc + iSrcStride, 0, src0Even, src1Even);
+    src2Even = __lsx_vld(pSrc + iSrcStride_x2, 0);
+    pSrc -= 1;
+
+    DUP2_ARG2(__lsx_vilvl_d, src0Even, src0, src1Even, src1, src0, src1);
+    src2 = __lsx_vilvl_d(src2Even, src2);
+    DUP4_ARG2(__lsx_vdp2_h_bu, src0, paramAB, src1, paramCD, src1, paramAB, src2,
+              paramCD, out0, outNext0, out1, outNext1);
+    DUP4_ARG2(__lsx_vadd_h, out0, outNext0, out0, paramConst, out1, outNext1, out1,
+              paramConst, out0, out0, out1, out1);
+    DUP2_ARG2(__lsx_vsrl_h, out0, shift, out1, shift, out0, out1);
+
+    DUP2_ARG2(__lsx_vilvh_d, out0, out0, out1, out1, outNext0, outNext1);
+    DUP2_ARG2(__lsx_vpackev_b, outNext0, out0, outNext1, out1, out0, out1);
+
+    __lsx_vstelm_d(out0, pDst, 0, 0);
+    __lsx_vstelm_d(out1, pDst + iDstStride, 0, 0);
+
+    pDst += iDstStride_x2;
+    pSrc += iSrcStride_x2;
+  }
+}