// ===== REFERENCES =====
// [1] duck.ac 用户 saffah_cc_v41_agg1，提交 #103276 <https://duck.ac/submission/103276>
// 用途：**本文件的引擎正文取自该提交**（= 本账号在 mmmd1k 上的最好件，37.740261 ms），
// 本件对它的唯一改动是下面的"七乘积直写 C 象限"（其余一字不动）。
// [2] duck.ac 用户 saffah_cc_v41_agg1，提交 #104048 <https://duck.ac/submission/104048>
// （mmml1k 上的同一把刀，判题机 −0.92%）用途：参考了"双目的地 leaf 直接把乘积写进两个
// C 象限、取消 cmb/scratch"的思想；本件在 mmmd1k 上独立实现并单变量定价。
// [3] 队友协调代理（同工作区）2026-09-28 的判题机读数：本題 combine 消融 = 2.41 Mcyc、
// M 平面 1575 页 = 0.87 Mcyc、页单价 555 ticks/页。用途：作为本刀收益侧的算术依据。
// ======================
// ===== 思路 =====
// 正式提交（非试验性；正确性已在本地逐位验证）。
//
// 一、机理：本題 n=1024/BASE=512 下，7 个 Strassen 乘积原本是"4 个直写 C 象限 + 3 个写 M 平面，
// 最后由一趟 deferred combine 把 3 个 M 平面按 ± 混进 4 个象限"。那把 combine 判题机消融价
// **2.41 Mcyc**，而 3 个 M 平面又占 **1575 页**（页单价实测 555 ticks/页 ⇒ 0.87 Mcyc）。
// 本件把 7 个乘积**全部直接写进自己的 1~2 个 C 象限**：C_i += s_i * acc（s_i ∈ {0,±1}），
// 于是 **M 平面与整趟 combine 一并消失**。收益 ≈ 2.41 + 0.87 − 4.2 MB 额外流量 ≈ **−2.8 Mcyc**。
// 二、实现要点（单变量、可逐行核对）
// 1) mikro 收尾由 12 条 NT store 改成**双目的地 FMA 累加**：每个目的地 6 行 × (2 load + 2 FMA + 2 store)；
// 符号取自一张 {−1,0,+1} 向量表并作为 FMA 的内存乘数 ⇒ **收尾零分支**。
// 2) op=0（某象限的首写者）用 "vxorpd 清零 + s=+1 累加" 实现**覆盖**语义 ⇒ 与旧件一样对"C 初值非 0"稳健。
// 3) 象限表（0=C11 1=C12 2=C21 3=C22）：Q1/O1={0,2,1,0,0,3,0}/{0,0,0,1,−1,1,1}、
// Q2/O2={3,3,3,2,1,−1,−1}/{0,−1,1,1,1,0,0}；叶序改为自然序 **0..6**（唯一满足
// "每个象限的首写者必须是 op=0 的那一个"的简单拓扑序）；单个目的地的两叶其第二目的地指向一个
// 64 字节假 sink（带 `g_d2on=0` 守卫，偏移为 0）。
// 4) 旧 `pending`/combine 不再上膛（`early=1`）⇒ 相关机器码保留但从不执行。
// 三、正确性（本地，与 [1] 同 A/B 对拍）
// * `A=B=e00`：maxdiff **0**；`A=B=I`：**恰好是单位阵**（offcount=0）；
// * 随机 A/B（n=1024 全矩阵）：与 [1] 的输出 **maxdiff = 0.000e+00（逐位相同）**；
// * 4 组 (op1,op2) 组合的 mikro 收尾在隔离台逐位核对通过；点积抽检 2.3e-14（容差 6.99e-10）。
// 四、本次为**单变量提交**：相对 [1] 只改"7 个乘积的落点 + 收尾形态"，分块/打包/叶内循环一字不动。
// ================
// ===== REFERENCES =====
// [1] duck.ac 用户 saffah_codex_6s_agg2，提交 #103194 <https://duck.ac/submission/103194>
// 用途：**本文件的引擎正文逐字节取自该提交的公开源码**（本题当前榜首，37.718412 ms）。
// 该件本身又是 [2] / [3] / [4] 的累积，其对本代理的增量是三处：
// (a) `#pragma GCC target("avx2,fma")` → 追加 `,tune=skylake`；
// (b) 只读旋钮 `static const/volatile int` → `static constexpr int`（编译期特化）；
// (c) `kernel_plain` 的 k 循环由 4 行展开改成 8 行展开，并在第 5 个 k 之后
// 追加 4 条 `prefetcht0 768..960(%[b])`（B 面板的下一组 4 个 cache line）。
// [2] duck.ac 用户 saffah_codex_6s_agg2，提交 #103029 <https://duck.ac/submission/103029>
// 用途：其"前沿受限预取"改动随 [1] 一起被本文件继承 ——
// `pending_prefetch()` 里加一条 `if(pending.prog && pos/BASE >= pending.rowlim) return;`
// （推迟的 combine 的软件预取不得越过最后一个叶已写完的行前沿）。
// [3] duck.ac 用户 saffah_codex_6s_agg2，提交 #103028 <https://duck.ac/submission/103028>
// 用途：`tune=skylake` / constexpr 特化 / 64 字节循环对齐（`#103194` 的自述里声明其父件
// 为 #103028）—— 本文件经由 [1] 继承。
// [4] duck.ac 用户 saffah_cc_v41_agg1，提交 #103010 <https://duck.ac/submission/103010>
// 用途：**引擎的原始正文**（本账号自己，本代理的前一代最好件）。其本身是姊妹题 mmmd4k
// 达标题 #101193 <https://duck.ac/submission/101193> 的 n-generic 移植 + 两把本代理的刀
// （滞后 combine 改"前沿受限 + 提前一叶装载"；叶序遍历 lperm={3,1,2,0,5,4,6}）。
// 本地归档：problems/mmmd1k/work/d1_sub_v1.cpp。
// [5] duck.ac 用户 saffah_cc_v41_agg1，提交 #101320 <https://duck.ac/submission/101320>
// 用途：本代理更早一代在本题上的最好件（扁平 5 层分块 6x8 GEMM，41.673395 ms），
// 仅作为换引擎后的正确性对拍基线。
// [6] Goto, K. and van de Geijn, R., "Anatomy of High-Performance Matrix Multiplication",
// ACM TOMS 34(3), 2008. 用途：block packing / 面板布局概念（经 [4] 间接沿用）。
// ======================
// ===== 思路 =====
// 正式提交（非试验性）。
//
// 一、本件的性质：**[1] 的正文 + 本代理自己的增量刀**（见二）。
// 为什么要以 [1] 为基座：本代理的 #103010 与 [1] 的谱系同源（[1] 的父件就是我们的 #103010），
// 所以在它上面做增量是**零风险的等价基座**，不需要重新取证。
// ⚠ 本代理**不**单独提交 [1] 的纯复刻：`37.718412 > 0.99*T = 37.339503`，
// 纯复刻是"已知不达标"的件（RULES 1.4）。增量刀是本提交的全部价值。
//
// 二、本代理的增量刀（与 [1] 的差异逐条列出，均为单变量）
// (A) **B 面板预取距离扫描**（BRIEF §2.18.622：prefetch 的第一件事是把"距离"当变量扫，
// 距离 0 = "没有预取"的对照）。本题的 B 面板分块在**同一次 kernel 的 5 个 r 调用之间
// 被完全复用**（每个 j 的 B 块只有 32 KB，r 循环 5 次全打在 L1/L2 上）
// ⇒ 硬件顺序预取已经够用，那 4 条 `prefetcht0` 极可能是**白占取数端口**的。
// 本件的取值由判题机 probe 决定（见文件末尾的实测记录）。
// (B) **循环对齐 `.p2align`**：notes 里列为"仍未扫的轴"。
//
// 三、正确性
// 本件与 [1] 在算术、访存顺序、微内核上**只有 (A)(B) 两处差异**，两处都不改变任何数值路径
// （预取是对读的提示；`.p2align` 是填充）⇒ 输出与 [1] 逐位相同。
// 本机对拍：n=1024 与长双精度参照（容差 3n²·DBL_EPS = 6.985e-10）+ A=B=I 恒等。
// ================
#pragma GCC optimize("O3,unroll-loops,no-strict-aliasing")
#pragma GCC target("avx2,fma,tune=skylake")
#include <immintrin.h>
#define MD9_T2BASE 1
#include <stddef.h>
#ifndef BASE
#define BASE 512
#endif
#define NTMIN g_NTMIN
#ifndef UPPER_PAGE_PAD
#define UPPER_PAGE_PAD 7680
#endif
#define LEAF_PAD g_LP
#ifndef SUMNT
#define SUMNT 1024
#endif
static constexpr int MR=6,NR=8;
alignas(4096) static double workspace[48*1024*1024];
alignas(64) static double ap[(128+MR)*(BASE+24)],ap2[(128+MR)*(BASE+24)],bp[BASE*(BASE+24)],bp2[BASE*(BASE+24)];
/* m2s/md5 fused-pack support: a zero row (for pad rows and for pure copies) and the
two sign masks (+0.0 / -0.0) used to apply Strassen's operand sign in-register. */
alignas(64) static double pk_zero[BASE+8];
static constexpr int g_abl=0;
static constexpr int g_MC=30, g_LP=960, g_NTMIN=512, g_PERIOD=1;
static constexpr int g_early=1, g_PSTEP=2, g_SF=1, g_PBC=0;
static int g_leafprod=0;
static constexpr int g_STEP=32;
static const int g_lperm[7]={0,1,2,3,4,5,6};
/* ===== sm3_dl: direct-leaf (dual-destination) support ===================== */
alignas(32) static double sv_tab[12]={0,0,0,0, 1,1,1,1, -1,-1,-1,-1};
static const double *g_sv1=sv_tab+4,*g_sv2=sv_tab;
static const double *g_svp_tab[3]={sv_tab+8,sv_tab,sv_tab+4}; /* [op+1]: -1 -> minus, 0 -> zero, +1 -> plus */
alignas(64) static double sink_dummy[8]={0,0,0,0,0,0,0,0};
static double *g_cq[4]={0,0,0,0};
static double *g_sink=0, *g_c2=0;
static long g_stride=0, g_stride2=0;
static int g_op1=1, g_op2=0, g_d2on=0;
static const int DL_Q1[7]={0,2,1,0,0,3,0};
static const int DL_O1[7]={0,0,0,1,-1,1,1};
static const int DL_Q2[7]={3,3,3,2,1,-1,-1};
static const int DL_O2[7]={0,-1,1,1,1,0,0};
alignas(64) static double pk_mp[4]={0.0,0.0,0.0,0.0}, pk_mn[4]={-0.0,-0.0,-0.0,-0.0};
struct Mat { const double *x; int ld; const double *y=nullptr; int ly=0; int sign=0; };
static inline __m256d ld(Mat a,int i,int j) {
__m256d x=_mm256_load_pd(a.x+(size_t)i*a.ld+j);
if(a.sign) {
__m256d y=_mm256_load_pd(a.y+(size_t)i*a.ly+j);
x=a.sign>0?_mm256_add_pd(x,y):_mm256_sub_pd(x,y);
}
return x;
}
static inline double scalar(Mat a,int i,int k) {
double x=a.x[(size_t)i*a.ld+k];
if(a.sign)x+=a.sign>0?a.y[(size_t)i*a.ly+k]:-a.y[(size_t)i*a.ly+k];
return x;
}
template<int SG> static __attribute__((always_inline)) inline void kernel(int n,const double *a,const double *b,double *c,double *c2,int ldc,int rows,
const double *pks,const double *pky,double *pkd) {
(void)c2;
alignas(32) double tmp[MR*8];
double *out=rows==MR?c:tmp;
ptrdiff_t stride=rows==MR?ldc*8:64;
int k=n;
asm volatile (
"vxorpd %%ymm0, %%ymm0, %%ymm0\n\t"
"vxorpd %%ymm1, %%ymm1, %%ymm1\n\t"
"vxorpd %%ymm2, %%ymm2, %%ymm2\n\t"
"vxorpd %%ymm3, %%ymm3, %%ymm3\n\t"
"vxorpd %%ymm4, %%ymm4, %%ymm4\n\t"
"vxorpd %%ymm5, %%ymm5, %%ymm5\n\t"
"vxorpd %%ymm6, %%ymm6, %%ymm6\n\t"
"vxorpd %%ymm7, %%ymm7, %%ymm7\n\t"
"vxorpd %%ymm8, %%ymm8, %%ymm8\n\t"
"vxorpd %%ymm9, %%ymm9, %%ymm9\n\t"
"vxorpd %%ymm10, %%ymm10, %%ymm10\n\t"
"vxorpd %%ymm11, %%ymm11, %%ymm11\n\t"
".p2align 6\n\t"
"1:\n\t"
".if %c[sg] != 0\n\t"
"prefetcht0 512(%[b])\n\t"
"prefetcht0 576(%[b])\n\t"
"prefetcht0 640(%[b])\n\t"
"prefetcht0 704(%[b])\n\t"
"vmovapd 0(%[b]), %%ymm12\n\t"
"vmovapd 32(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 64(%[b]), %%ymm12\n\t"
"vmovapd 96(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 128(%[b]), %%ymm12\n\t"
"vmovapd 160(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 192(%[b]), %%ymm12\n\t"
"vmovapd 224(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"prefetcht0 768(%[b])\n\t"
"prefetcht0 832(%[b])\n\t"
"prefetcht0 896(%[b])\n\t"
"prefetcht0 960(%[b])\n\t"
"vmovapd 256(%[b]), %%ymm12\n\t"
"vmovapd 288(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+32(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+32(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+32(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+32(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+32(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+32(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 320(%[b]), %%ymm12\n\t"
"vmovapd 352(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+40(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+40(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+40(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+40(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+40(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+40(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 384(%[b]), %%ymm12\n\t"
"vmovapd 416(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+48(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+48(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+48(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+48(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+48(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+48(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 448(%[b]), %%ymm12\n\t"
"vmovapd 480(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+56(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+56(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+56(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+56(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+56(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+56(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovupd 0(%[pks]), %%ymm15\n\t"
".if %c[sg] == 1\n\t"
"vaddpd 0(%[pky]), %%ymm15, %%ymm15\n\t"
".elseif %c[sg] == -1\n\t"
"vsubpd 0(%[pky]), %%ymm15, %%ymm15\n\t"
".endif\n\t"
"vmovupd %%ymm15, 0(%[pkd])\n\t"
"add $32, %[pks]\n\t"
"add $32, %[pky]\n\t"
"add $32, %[pkd]\n\t"
"add $64, %[a]\n\t"
"add $512, %[b]\n\t"
"sub $8, %[k]\n\t"
".else\n\t"
"prefetcht0 512(%[b])\n\t"
"prefetcht0 576(%[b])\n\t"
"prefetcht0 640(%[b])\n\t"
"prefetcht0 704(%[b])\n\t"
"vmovapd 0(%[b]), %%ymm12\n\t"
"vmovapd 32(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 64(%[b]), %%ymm12\n\t"
"vmovapd 96(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 128(%[b]), %%ymm12\n\t"
"vmovapd 160(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 192(%[b]), %%ymm12\n\t"
"vmovapd 224(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"add $32, %[a]\n\t"
"add $256, %[b]\n\t"
"sub $4, %[k]\n\t"
".endif\n\t"
"jnz 1b\n\t"
"test %[op1], %[op1]\n\t"
"jz .LSD1s_%=\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm0, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm1, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm2, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm3, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm4, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm5, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm6, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm7, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm8, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm9, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm10, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm11, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"jmp .LSD1e_%=\n\t"
".LSD1s_%=:\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm0, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm1, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm2, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm3, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm4, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm5, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm6, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm7, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm8, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm9, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm10, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm11, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
".LSD1e_%=:\n\t"
"test %[op2], %[op2]\n\t"
"jz .LSD2s_%=\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm0, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm1, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm2, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm3, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm4, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm5, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm6, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm7, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm8, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm9, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm10, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm11, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"jmp .LSD2e_%=\n\t"
".LSD2s_%=:\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm0, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm1, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm2, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm3, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm4, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm5, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm6, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm7, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm8, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm9, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm10, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm11, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
".LSD2e_%=:\n\t"
: [a] "+&r"(a), [b] "+&r"(b), [out] "+&r"(out), [out2] "+&r"(c2), [k] "+&r"(k),
[pks] "+&r"(pks), [pky] "+&r"(pky), [pkd] "+&r"(pkd)
: [stride] "r"(g_stride), [stride2] "r"(g_stride2), [sg] "i"(SG), [op1] "r"(g_op1), [op2] "r"(g_op2), [sv1] "r"(g_sv1), [sv2] "r"(g_sv2), [o0] "i"(0*(BASE+24)*8), [o1] "i"(1*(BASE+24)*8), [o2] "i"(2*(BASE+24)*8), [o3] "i"(3*(BASE+24)*8), [o4] "i"(4*(BASE+24)*8), [o5] "i"(5*(BASE+24)*8)
: "memory", "cc", "ymm0","ymm1","ymm2","ymm3","ymm4","ymm5","ymm6","ymm7","ymm8","ymm9","ymm10","ymm11","ymm12","ymm13","ymm14","ymm15");
if(rows!=MR)for(int i=0;i<rows;i++)for(int j=0;j<8;j+=4)
_mm256_store_pd(c+(size_t)i*ldc+j,_mm256_load_pd(tmp+i*8+j));
}
static __attribute__((always_inline)) inline void kernel_plain(int n,const double *a,const double *b,double *c,double *c2,int ldc,int rows) {
(void)c2;
alignas(32) double tmp[MR*8];
double *out=rows==MR?c:tmp;
ptrdiff_t stride=rows==MR?ldc*8:64;
int k=n;
asm volatile (
"vxorpd %%ymm0, %%ymm0, %%ymm0\n\t"
"vxorpd %%ymm1, %%ymm1, %%ymm1\n\t"
"vxorpd %%ymm2, %%ymm2, %%ymm2\n\t"
"vxorpd %%ymm3, %%ymm3, %%ymm3\n\t"
"vxorpd %%ymm4, %%ymm4, %%ymm4\n\t"
"vxorpd %%ymm5, %%ymm5, %%ymm5\n\t"
"vxorpd %%ymm6, %%ymm6, %%ymm6\n\t"
"vxorpd %%ymm7, %%ymm7, %%ymm7\n\t"
"vxorpd %%ymm8, %%ymm8, %%ymm8\n\t"
"vxorpd %%ymm9, %%ymm9, %%ymm9\n\t"
"vxorpd %%ymm10, %%ymm10, %%ymm10\n\t"
"vxorpd %%ymm11, %%ymm11, %%ymm11\n\t"
".p2align 6\n\t"
"1:\n\t"
"prefetcht0 512(%[b])\n\t"
"prefetcht0 576(%[b])\n\t"
"prefetcht0 640(%[b])\n\t"
"prefetcht0 704(%[b])\n\t"
"vmovapd 0(%[b]), %%ymm12\n\t"
"vmovapd 32(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 64(%[b]), %%ymm12\n\t"
"vmovapd 96(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 128(%[b]), %%ymm12\n\t"
"vmovapd 160(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 192(%[b]), %%ymm12\n\t"
"vmovapd 224(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"prefetcht0 768(%[b])\n\t"
"prefetcht0 832(%[b])\n\t"
"prefetcht0 896(%[b])\n\t"
"prefetcht0 960(%[b])\n\t"
"vmovapd 256(%[b]), %%ymm12\n\t"
"vmovapd 288(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+32(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+32(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+32(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+32(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+32(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+32(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 320(%[b]), %%ymm12\n\t"
"vmovapd 352(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+40(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+40(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+40(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+40(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+40(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+40(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 384(%[b]), %%ymm12\n\t"
"vmovapd 416(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+48(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+48(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+48(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+48(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+48(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+48(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"vmovapd 448(%[b]), %%ymm12\n\t"
"vmovapd 480(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+56(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+56(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vbroadcastsd %c[o2]+56(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm4\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm5\n\t"
"vbroadcastsd %c[o3]+56(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm6\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm7\n\t"
"vbroadcastsd %c[o4]+56(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm8\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm9\n\t"
"vbroadcastsd %c[o5]+56(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm10\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm11\n\t"
"add $64, %[a]\n\t"
"add $512, %[b]\n\t"
"sub $8, %[k]\n\t"
"jnz 1b\n\t"
"test %[op1], %[op1]\n\t"
"jz .LSD1s_%=\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm0, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm1, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm2, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm3, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm4, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm5, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm6, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm7, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm8, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm9, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm10, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm11, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"jmp .LSD1e_%=\n\t"
".LSD1s_%=:\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm0, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm1, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm2, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm3, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm4, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm5, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm6, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm7, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm8, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm9, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm10, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm11, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
".LSD1e_%=:\n\t"
"test %[op2], %[op2]\n\t"
"jz .LSD2s_%=\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm0, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm1, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm2, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm3, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm4, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm5, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm6, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm7, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm8, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm9, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm10, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm11, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"jmp .LSD2e_%=\n\t"
".LSD2s_%=:\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm0, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm1, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm2, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm3, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm4, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm5, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm6, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm7, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm8, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm9, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm10, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm11, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
".LSD2e_%=:\n\t"
: [a] "+&r"(a), [b] "+&r"(b), [out] "+&r"(out), [out2] "+&r"(c2), [k] "+&r"(k)
: [stride] "r"(g_stride), [stride2] "r"(g_stride2), [op1] "r"(g_op1), [op2] "r"(g_op2), [sv1] "r"(g_sv1), [sv2] "r"(g_sv2), [o0] "i"(0*(BASE+24)*8), [o1] "i"(1*(BASE+24)*8), [o2] "i"(2*(BASE+24)*8), [o3] "i"(3*(BASE+24)*8), [o4] "i"(4*(BASE+24)*8), [o5] "i"(5*(BASE+24)*8)
: "memory", "cc", "ymm0","ymm1","ymm2","ymm3","ymm4","ymm5","ymm6","ymm7","ymm8","ymm9","ymm10","ymm11","ymm12","ymm13","ymm14","ymm15");
if(rows!=MR)for(int i=0;i<rows;i++)for(int j=0;j<8;j+=4)
_mm256_store_pd(c+(size_t)i*ldc+j,_mm256_load_pd(tmp+i*8+j));
}
/* ===== mdA_dp2: deferred h=1024 combine (Winograd form, exact same arithmetic) ===== */
struct Pend2 { double *c; int ldc; const double *m; int h; int pz; int pos; bool active; };
static Pend2 pend2={0,0,0,0,0,0,false};
static constexpr int STEP2=16;
static inline void pend2_prefetch(void){
if(!pend2.active)return;
int i=pend2.pos/pend2.h, j=pend2.pos%pend2.h;
const double *c=pend2.c+(size_t)i*pend2.ldc+j;
const double *m=pend2.m+pend2.pos;
for(int t=0;t<STEP2;t+=8) {
_mm_prefetch((const char*)(c+t),_MM_HINT_T1);
_mm_prefetch((const char*)(c+pend2.h+t),_MM_HINT_T1);
_mm_prefetch((const char*)(c+(size_t)pend2.h*pend2.ldc+t),_MM_HINT_T1);
_mm_prefetch((const char*)(c+(size_t)pend2.h*pend2.ldc+pend2.h+t),_MM_HINT_T1);
_mm_prefetch((const char*)(m+t),_MM_HINT_T1);
_mm_prefetch((const char*)(m+pend2.pz+t),_MM_HINT_T1);
_mm_prefetch((const char*)(m+2*pend2.pz+t),_MM_HINT_T1);
}
}
static inline void pend2_step(void){
int i=pend2.pos/pend2.h, j=pend2.pos%pend2.h;
for(int t=0;t<STEP2;t+=4) {
double *c11=pend2.c+(size_t)i*pend2.ldc+j+t,*c12=c11+pend2.h,
*c21=c11+(size_t)pend2.h*pend2.ldc,*c22=c21+pend2.h;
const double *m=pend2.m+pend2.pos+t;
__m256d m1=_mm256_load_pd(c11),m2=_mm256_load_pd(c21),
m3=_mm256_load_pd(c12),m4=_mm256_load_pd(c22),
m5=_mm256_load_pd(m),m6=_mm256_load_pd(m+pend2.pz),m7=_mm256_load_pd(m+2*pend2.pz);
__m256d u2=_mm256_add_pd(m1,m6),u3=_mm256_add_pd(u2,m7),u4=_mm256_add_pd(u2,m5);
__m256d r11=_mm256_add_pd(m1,m2),r12=_mm256_add_pd(u4,m3),
r21=_mm256_sub_pd(u3,m4),r22=_mm256_add_pd(u3,m5);
_mm256_store_pd(c11,r11);_mm256_store_pd(c12,r12);
_mm256_store_pd(c21,r21);_mm256_store_pd(c22,r22);
}
pend2.pos+=STEP2;
if(pend2.pos>=pend2.h*pend2.h)pend2.active=false;
}
static void pend2_finish(void){ while(pend2.active)pend2_step(); }
struct Pending { double *c; int ldc; const double *m; int pos=0; bool active=false; int rowlim=0; int prog=0; };
static Pending pending;
alignas(4096) static double pending_work[3*(BASE*BASE+16384)+16384];
static unsigned pending_tick=0;
#define STEP g_STEP
#define PERIOD g_PERIOD
static inline void pending_prefetch() {
if(!pending.active)return;
// Do not repeatedly fetch the next row before its producer has finished it.
if(pending.prog && (int)(pending.pos/BASE)>=pending.rowlim)return;
int i=pending.pos/BASE,j=pending.pos%BASE;
const double *c=pending.c+(size_t)i*pending.ldc+j;
const double *m=pending.m+pending.pos;
for(int t=0;t<STEP;t+=8) {
_mm_prefetch((const char*)(c+t),_MM_HINT_T1);
_mm_prefetch((const char*)(c+BASE+t),_MM_HINT_T1);
_mm_prefetch((const char*)(c+(size_t)BASE*pending.ldc+t),_MM_HINT_T1);
_mm_prefetch((const char*)(c+(size_t)BASE*pending.ldc+BASE+t),_MM_HINT_T1);
_mm_prefetch((const char*)(m+t),_MM_HINT_T1);
_mm_prefetch((const char*)(m+(BASE*BASE+LEAF_PAD)+t),_MM_HINT_T1);
_mm_prefetch((const char*)(m+2*(BASE*BASE+LEAF_PAD)+t),_MM_HINT_T1);
}
}
static inline void pending_step() {
{ int lim=pending.prog?pending.rowlim:BASE;
if((int)(pending.pos/BASE)>=lim) return; }
int i=pending.pos/BASE,j=pending.pos%BASE;
for(int t=0;t<STEP;t+=4) {
double *c11=pending.c+(size_t)i*pending.ldc+j+t,
*c12=c11+BASE,*c21=c11+(size_t)BASE*pending.ldc,*c22=c21+BASE;
const double *m=pending.m+pending.pos+t;
__m256d m1=_mm256_load_pd(c11),m2=_mm256_load_pd(c21),
m3=_mm256_load_pd(c12),m4=_mm256_load_pd(c22),
m5=_mm256_load_pd(m),m6=_mm256_load_pd(m+(BASE*BASE+LEAF_PAD)),m7=_mm256_load_pd(m+2*(BASE*BASE+LEAF_PAD));
__m256d r11=_mm256_add_pd(_mm256_sub_pd(_mm256_add_pd(m1,m4),m5),m7),
r12=_mm256_add_pd(m3,m5),r21=_mm256_add_pd(m2,m4),
r22=_mm256_add_pd(_mm256_add_pd(_mm256_sub_pd(m1,m2),m3),m6);
_mm256_store_pd(c11,r11);_mm256_store_pd(c12,r12);
_mm256_store_pd(c21,r21);_mm256_store_pd(c22,r22);
}
pending.pos+=STEP;
if(pending.pos==BASE*BASE)pending.active=false;
}
static void pending_finish() {
pending.prog=0;
while(pending.active)pending_step();
}
static __attribute__((always_inline)) inline void kernel2(int n,const double *a,const double *b,double *c,int ldc) {
int k=n;
double *out=c;
ptrdiff_t stride=(ptrdiff_t)ldc*8;
asm volatile (
"vxorpd %%ymm0, %%ymm0, %%ymm0\n\t"
"vxorpd %%ymm1, %%ymm1, %%ymm1\n\t"
"vxorpd %%ymm2, %%ymm2, %%ymm2\n\t"
"vxorpd %%ymm3, %%ymm3, %%ymm3\n\t"
".p2align 6\n\t" "1:\n\t"
"prefetcht0 512(%[b])\n\t"
"prefetcht0 576(%[b])\n\t"
"prefetcht0 640(%[b])\n\t"
"prefetcht0 704(%[b])\n\t"
"vmovapd 0(%[b]), %%ymm12\n\t"
"vmovapd 32(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+0(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vmovapd 64(%[b]), %%ymm12\n\t"
"vmovapd 96(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+8(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vmovapd 128(%[b]), %%ymm12\n\t"
"vmovapd 160(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+16(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"vmovapd 192(%[b]), %%ymm12\n\t"
"vmovapd 224(%[b]), %%ymm13\n\t"
"vbroadcastsd %c[o0]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm0\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm1\n\t"
"vbroadcastsd %c[o1]+24(%[a]), %%ymm14\n\t"
"vfmadd231pd %%ymm12, %%ymm14, %%ymm2\n\t"
"vfmadd231pd %%ymm13, %%ymm14, %%ymm3\n\t"
"add $32, %[a]\n\t"
"add $256, %[b]\n\t"
"sub $4, %[k]\n\t"
"jnz 1b\n\t"
"vmovntpd %%ymm0, (%[out])\n\t"
"vmovntpd %%ymm1, 32(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vmovntpd %%ymm2, (%[out])\n\t"
"vmovntpd %%ymm3, 32(%[out])\n\t"
: [a] "+&r"(a), [b] "+&r"(b), [out] "+&r"(out), [k] "+&r"(k)
: [stride] "r"(stride), [o0] "i"(0*(BASE+24)*8), [o1] "i"(1*(BASE+24)*8)
: "memory", "cc", "ymm0","ymm1","ymm2","ymm3","ymm12","ymm13","ymm14");
}
static __attribute__((noinline)) void kernel_tail16(int n,const double *a,const double *b,double *out,double *out2,int ldc) {
const double *b1=b+8*(n+1);
ptrdiff_t stride=(ptrdiff_t)ldc*8;
int kk=n;
asm volatile (
"vxorpd %%ymm0, %%ymm0, %%ymm0\n\t"
"vxorpd %%ymm1, %%ymm1, %%ymm1\n\t"
"vxorpd %%ymm2, %%ymm2, %%ymm2\n\t"
"vxorpd %%ymm3, %%ymm3, %%ymm3\n\t"
"vxorpd %%ymm4, %%ymm4, %%ymm4\n\t"
"vxorpd %%ymm5, %%ymm5, %%ymm5\n\t"
"vxorpd %%ymm6, %%ymm6, %%ymm6\n\t"
"vxorpd %%ymm7, %%ymm7, %%ymm7\n\t"
".p2align 6\n\t"
"1:\n\t"
"prefetcht0 512(%[b])\n\t"
"prefetcht0 512(%[b1])\n\t"
"vbroadcastsd %c[o0]+0(%[a]), %%ymm12\n\t"
"vbroadcastsd %c[o1]+0(%[a]), %%ymm13\n\t"
"vmovapd 0(%[b]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm0\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm4\n\t"
"vmovapd 32(%[b]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm1\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm5\n\t"
"vmovapd 0(%[b1]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm2\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm6\n\t"
"vmovapd 32(%[b1]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm3\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm7\n\t"
"prefetcht0 576(%[b])\n\t"
"prefetcht0 576(%[b1])\n\t"
"vbroadcastsd %c[o0]+8(%[a]), %%ymm12\n\t"
"vbroadcastsd %c[o1]+8(%[a]), %%ymm13\n\t"
"vmovapd 64(%[b]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm0\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm4\n\t"
"vmovapd 96(%[b]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm1\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm5\n\t"
"vmovapd 64(%[b1]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm2\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm6\n\t"
"vmovapd 96(%[b1]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm3\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm7\n\t"
"prefetcht0 640(%[b])\n\t"
"prefetcht0 640(%[b1])\n\t"
"vbroadcastsd %c[o0]+16(%[a]), %%ymm12\n\t"
"vbroadcastsd %c[o1]+16(%[a]), %%ymm13\n\t"
"vmovapd 128(%[b]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm0\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm4\n\t"
"vmovapd 160(%[b]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm1\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm5\n\t"
"vmovapd 128(%[b1]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm2\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm6\n\t"
"vmovapd 160(%[b1]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm3\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm7\n\t"
"prefetcht0 704(%[b])\n\t"
"prefetcht0 704(%[b1])\n\t"
"vbroadcastsd %c[o0]+24(%[a]), %%ymm12\n\t"
"vbroadcastsd %c[o1]+24(%[a]), %%ymm13\n\t"
"vmovapd 192(%[b]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm0\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm4\n\t"
"vmovapd 224(%[b]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm1\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm5\n\t"
"vmovapd 192(%[b1]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm2\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm6\n\t"
"vmovapd 224(%[b1]), %%ymm14\n\t"
"vfmadd231pd %%ymm14, %%ymm12, %%ymm3\n\t"
"vfmadd231pd %%ymm14, %%ymm13, %%ymm7\n\t"
"add $32, %[a]\n\t"
"add $256, %[b]\n\t"
"add $256, %[b1]\n\t"
"sub $4, %[kk]\n\t"
"jnz 1b\n\t"
"test %[op1], %[op1]\n\t"
"jz .LSD1s_%=\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm0, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm1, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"vmovapd 64(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm2, %%ymm12\n\t"
"vmovapd %%ymm12, 64(%[out])\n\t"
"vmovapd 96(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm3, %%ymm12\n\t"
"vmovapd %%ymm12, 96(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vmovapd 0(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm4, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vmovapd 32(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm5, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"vmovapd 64(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm6, %%ymm12\n\t"
"vmovapd %%ymm12, 64(%[out])\n\t"
"vmovapd 96(%[out]), %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm7, %%ymm12\n\t"
"vmovapd %%ymm12, 96(%[out])\n\t"
"jmp .LSD1e_%=\n\t"
".LSD1s_%=:\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm0, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm1, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm2, %%ymm12\n\t"
"vmovapd %%ymm12, 64(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm3, %%ymm12\n\t"
"vmovapd %%ymm12, 96(%[out])\n\t"
"add %[stride], %[out]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm4, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm5, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm6, %%ymm12\n\t"
"vmovapd %%ymm12, 64(%[out])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv1]), %%ymm7, %%ymm12\n\t"
"vmovapd %%ymm12, 96(%[out])\n\t"
".LSD1e_%=:\n\t"
"test %[op2], %[op2]\n\t"
"jz .LSD2s_%=\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm0, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm1, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"vmovapd 64(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm2, %%ymm12\n\t"
"vmovapd %%ymm12, 64(%[out2])\n\t"
"vmovapd 96(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm3, %%ymm12\n\t"
"vmovapd %%ymm12, 96(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vmovapd 0(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm4, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vmovapd 32(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm5, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"vmovapd 64(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm6, %%ymm12\n\t"
"vmovapd %%ymm12, 64(%[out2])\n\t"
"vmovapd 96(%[out2]), %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm7, %%ymm12\n\t"
"vmovapd %%ymm12, 96(%[out2])\n\t"
"jmp .LSD2e_%=\n\t"
".LSD2s_%=:\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm0, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm1, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm2, %%ymm12\n\t"
"vmovapd %%ymm12, 64(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm3, %%ymm12\n\t"
"vmovapd %%ymm12, 96(%[out2])\n\t"
"add %[stride2], %[out2]\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm4, %%ymm12\n\t"
"vmovapd %%ymm12, 0(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm5, %%ymm12\n\t"
"vmovapd %%ymm12, 32(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm6, %%ymm12\n\t"
"vmovapd %%ymm12, 64(%[out2])\n\t"
"vxorpd %%ymm12, %%ymm12, %%ymm12\n\t"
"vfmadd231pd (%[sv2]), %%ymm7, %%ymm12\n\t"
"vmovapd %%ymm12, 96(%[out2])\n\t"
".LSD2e_%=:\n\t"
: [a] "+&r"(a),[b] "+&r"(b),[b1] "+&r"(b1),[out] "+&r"(out),[out2] "+&r"(out2),[kk] "+&r"(kk)
: [stride] "r"(g_stride),[stride2] "r"(g_stride2),[op1] "r"(g_op1),[op2] "r"(g_op2),[sv1] "r"(g_sv1),[sv2] "r"(g_sv2),[o0] "i"(0),[o1] "i"((BASE+24)*8)
: "memory","cc","ymm0","ymm1","ymm2","ymm3","ymm4","ymm5","ymm6","ymm7","ymm8","ymm9","ymm10","ymm11","ymm12","ymm13","ymm14","ymm15");
}
/* ===== packB carry (g_PBC) =================================================
leaf() packs its whole B panel into bp *serially* before any kernel runs, so
that DRAM burst (2.1 MB write + 2.1-4.2 MB read per leaf, x7) never overlaps
the FMA-bound kernels. Measured pool on the judge: 5.39 Mcyc = 4.05%.
Same template the engine already uses for the A panel: build the NEXT leaf's
B panel a little at a time from inside this leaf's kernel loop.
g_PBC=0 restores the original code path exactly (BP==bp, g_pbp==0). */
static double *g_bp[2]={bp,bp2};
static int g_bpi=0, g_pbp=0;
struct PackJob { Mat b{0,0}; double *dst=0; int k=0, n=0; bool active=false; };
static PackJob pj;
static __attribute__((noinline)) void packjob_step(void){
int k=pj.k,n=pj.n;
for(int j=0;j<n;j+=8){
double *p=pj.dst+(size_t)j*(n+1)+k*8;
_mm256_store_pd(p+0,ld(pj.b,k+0,j));
_mm256_store_pd(p+4,ld(pj.b,k+0,j+4));
_mm256_store_pd(p+8,ld(pj.b,k+1,j));
_mm256_store_pd(p+12,ld(pj.b,k+1,j+4));
}
pj.k+=2;
if(pj.k>=n)pj.active=false;
}
static void packjob_finish(void){ while(pj.active)packjob_step(); }
/* reset the carry's persistent state: REQUIRED between arms, otherwise a pbc=1 arm
leaks g_pbp=1 into the next pbc=0 arm and that arm silently skips its B pack. */
static void pbc_reset(void){ g_bpi=0; g_pbp=0; pj.active=false; }
static inline void packjob_start(Mat b,double *dst,int n){
pj.b=b; pj.dst=dst; pj.k=0; pj.n=n; pj.active=true;
}
static void leaf(int n,Mat a,Mat b,double *c,int ldc) {
if(g_abl&2){for(int i=0;i<n;i++)for(int j=0;j<n;j+=4)_mm256_store_pd(c+(size_t)i*ldc+j,_mm256_setzero_pd());return;}
const int mc_=g_MC;
#define MC mc_
double *BP = g_PBC ? g_bp[g_bpi] : bp;
if(!g_pbp && !(g_abl&4))for(int k=0;k<n;k+=2)for(int j=0;j<n;j+=8) {
double *p=BP+(size_t)j*(n+1)+k*8;
_mm256_store_pd(p+0,ld(b,k+0,j));_mm256_store_pd(p+4,ld(b,k+0,j+4));
_mm256_store_pd(p+8,ld(b,k+1,j));_mm256_store_pd(p+12,ld(b,k+1,j+4));
}
pending_prefetch();
double *pkb[2]={ap,ap2};
int cur=0;
{ /* pack the first ic-block the ordinary way */
int mc0=n<MC?n:MC;
for(int r=0;r<mc0;r+=MR) {
int rows=mc0-r<MR?mc0-r:MR;
for(int i=0;i<rows;i++)for(int k=0;k<n;k+=4)
_mm256_store_pd(pkb[0]+(size_t)(r+i)*(BASE+24)+k,ld(a,r+i,k));
for(int i=rows;i<MR;i++)for(int k=0;k<n;k+=4)
_mm256_store_pd(pkb[0]+(size_t)(r+i)*(BASE+24)+k,_mm256_setzero_pd());
}
}
for(int ic=0;ic<n;) {
int mc=n-ic<=MC+MR?n-ic:MC;
if(mc==MC+2) {
for(int j=0;j<n;j+=16) {
for(int jj=j;jj<j+16;jj+=NR)for(int r=0;r<MC;r+=MR) {
kernel_plain(n,pkb[cur]+(size_t)r*(BASE+24),BP+(size_t)jj*(n+1),c+(size_t)(ic+r)*ldc+jj,g_c2+(size_t)((ic+r)*ldc+jj)*g_d2on,ldc,MR);
if(pj.active)packjob_step();
if(pending.active && (++pending_tick%PERIOD)==0) {for(int ps=0;ps<g_PSTEP;ps++)pending_step();pending_prefetch();}
if(pend2.active) {pend2_step();pend2_prefetch();}
}
kernel_tail16(n,pkb[cur]+(size_t)MC*(BASE+24),BP+(size_t)j*(n+1),c+(size_t)(ic+MC)*ldc+j,g_c2+(size_t)((ic+MC)*ldc+j)*g_d2on,ldc);
}
ic+=mc;
if(g_leafprod)pending.rowlim=ic;
continue;
}
/* ---- the NEXT ic-block's A panel is carried by this block's kernel calls ---- */
const int ic2=ic+mc;
const int nxt=(ic2<n)?(n-ic2<=MC+MR?n-ic2:MC):0;
const int rows2=((nxt+MR-1)/MR)*MR; /* rows incl. the zero pad rows */
const int nch=rows2*2; /* 2 chunks per row */
const int qd=n/2; /* doubles per chunk */
int ck=0; int skip=(n/NR)*((mc+MR-1)/MR)-nch;
for(int j=0;j<n;j+=NR)for(int r=0;r<mc;r+=MR) {
if(mc-r==2) { /* md9_t2: 2-row tail body (was 6-row body + tmp copy) */
kernel2(n,pkb[cur]+(size_t)r*(BASE+24),BP+(size_t)j*(n+1),c+(size_t)(ic+r)*ldc+j,ldc);
continue;
}
const double *pks=pk_zero,*pky=pk_zero,*pkm=pk_mp;
double *pkd=pk_zero;
int pkc=0;
if(ck<nch && --skip<0) {
const int row=ck/2, off=(ck%2)*qd;
if(row<nxt) {
pks=a.x+(size_t)(ic2+row)*a.ld+(size_t)off;
pky=a.sign?(a.y+(size_t)(ic2+row)*a.ly+(size_t)off):pk_zero;
pkm=a.sign>0?pk_mp:pk_mn;
}
pkd=pkb[cur^1]+(size_t)row*(BASE+24)+(size_t)off;
pkc=8;
ck++;
}
if(!pkc) kernel_plain(n,pkb[cur]+(size_t)r*(BASE+24),BP+(size_t)j*(n+1),c+(size_t)(ic+r)*ldc+j,g_c2+(size_t)((ic+r)*ldc+j)*g_d2on,ldc,mc-r<MR?mc-r:MR);
else if(a.sign==0) kernel<2>(n,pkb[cur]+(size_t)r*(BASE+24),BP+(size_t)j*(n+1),c+(size_t)(ic+r)*ldc+j,g_c2+(size_t)((ic+r)*ldc+j)*g_d2on,ldc,mc-r<MR?mc-r:MR,pks,pky,pkd);
else if(a.sign>0) kernel<1>(n,pkb[cur]+(size_t)r*(BASE+24),BP+(size_t)j*(n+1),c+(size_t)(ic+r)*ldc+j,g_c2+(size_t)((ic+r)*ldc+j)*g_d2on,ldc,mc-r<MR?mc-r:MR,pks,pky,pkd);
else kernel<-1>(n,pkb[cur]+(size_t)r*(BASE+24),BP+(size_t)j*(n+1),c+(size_t)(ic+r)*ldc+j,g_c2+(size_t)((ic+r)*ldc+j)*g_d2on,ldc,mc-r<MR?mc-r:MR,pks,pky,pkd);
if(pj.active)packjob_step();
if(pending.active && (++pending_tick%PERIOD)==0) {
for(int ps=0;ps<g_PSTEP;ps++)pending_step(); pending_prefetch();
}
if(pend2.active) { pend2_step(); pend2_prefetch(); }
}
/* ---- finish whatever the kernel loop did not carry ---- */
for(int cc=ck;cc<nch;cc++) {
const int row=cc/2, off=(cc%2)*qd;
if(row<nxt) {
for(int k=off;k<off+qd;k+=4)
_mm256_store_pd(pkb[cur^1]+(size_t)row*(BASE+24)+k, ld(a,ic2+row,k));
} else {
for(int k=off;k<off+qd;k+=4)
_mm256_store_pd(pkb[cur^1]+(size_t)row*(BASE+24)+k, _mm256_setzero_pd());
}
}
cur^=1;
ic+=mc;
if(g_leafprod)pending.rowlim=ic;
}
if(g_SF && n>=NTMIN)_mm_sfence();
}
static void sum(int h,double *dst,Mat a,Mat b,int sign) {
for(int i=0;i<h;i++)for(int j=0;j<h;j+=8) {
__m256d v=sign>0?_mm256_add_pd(ld(a,i,j),ld(b,i,j)):_mm256_sub_pd(ld(a,i,j),ld(b,i,j));
__m256d w=sign>0?_mm256_add_pd(ld(a,i,j+4),ld(b,i,j+4)):_mm256_sub_pd(ld(a,i,j+4),ld(b,i,j+4));
asm volatile("" : "+x"(v), "+x"(w) : : "memory");
if(h>=SUMNT) {
_mm256_stream_pd(dst+(size_t)i*h+j,v);_mm256_stream_pd(dst+(size_t)i*h+j+4,w);
} else {
_mm256_store_pd(dst+(size_t)i*h+j,v);_mm256_store_pd(dst+(size_t)i*h+j+4,w);
}
}
if(h>=SUMNT)_mm_sfence();
}
static void combine(int h,double *c,int ldc,const double *m,bool wino=false) {
size_t z=(size_t)h*h+(wino?UPPER_PAGE_PAD:0);
for(int i=0;i<h;i++)for(int j=0;j<h;j+=4) {
size_t t=(size_t)i*h+j;
double *c11=c+(size_t)i*ldc+j,*c12=c11+h,*c21=c+(size_t)(i+h)*ldc+j,*c22=c21+h;
__m256d m1=_mm256_load_pd(c11),m2=_mm256_load_pd(c21),m3=_mm256_load_pd(c12),m4=_mm256_load_pd(c22),
m5=_mm256_load_pd(m+t),m6=_mm256_load_pd(m+z+t),m7=_mm256_load_pd(m+2*z+t);
__m256d r11,r12,r21,r22;
if(wino) {
__m256d u2=_mm256_add_pd(m1,m6),u3=_mm256_add_pd(u2,m7),u4=_mm256_add_pd(u2,m5);
r11=_mm256_add_pd(m1,m2);r12=_mm256_add_pd(u4,m3);r21=_mm256_sub_pd(u3,m4);r22=_mm256_add_pd(u3,m5);
} else {
r11=_mm256_add_pd(_mm256_sub_pd(_mm256_add_pd(m1,m4),m5),m7);r12=_mm256_add_pd(m3,m5);r21=_mm256_add_pd(m2,m4);
r22=_mm256_add_pd(_mm256_add_pd(_mm256_sub_pd(m1,m2),m3),m6);
}
_mm256_store_pd(c11,r11);_mm256_store_pd(c12,r12);_mm256_store_pd(c21,r21);_mm256_store_pd(c22,r22);
}
}
static void pre(int h,Mat a,double *s1,double *s2,double *s4,double *s3,int ld3,bool right) {
for(int i=0;i<h;i++)for(int j=0;j<h;j+=8) {
__m256d a0=ld(a,i,j),a1=ld(a,i,j+4),b0=ld(a,i,j+h),b1=ld(a,i,j+h+4),
c0=ld(a,i+h,j),c1=ld(a,i+h,j+4),d0=ld(a,i+h,j+h),d1=ld(a,i+h,j+h+4);
__m256d p0,p1,q0,q1,r0,r1,t0,t1;
if(!right) {
p0=_mm256_add_pd(c0,d0);p1=_mm256_add_pd(c1,d1);
q0=_mm256_sub_pd(p0,a0);q1=_mm256_sub_pd(p1,a1);
r0=_mm256_sub_pd(b0,q0);r1=_mm256_sub_pd(b1,q1);
t0=_mm256_sub_pd(a0,c0);t1=_mm256_sub_pd(a1,c1);
}else {
p0=_mm256_sub_pd(b0,a0);p1=_mm256_sub_pd(b1,a1);
q0=_mm256_sub_pd(d0,p0);q1=_mm256_sub_pd(d1,p1);
r0=_mm256_sub_pd(q0,c0);r1=_mm256_sub_pd(q1,c1);
t0=_mm256_sub_pd(d0,b0);t1=_mm256_sub_pd(d1,b1);
}
size_t off=(size_t)i*h+j,off3=(size_t)i*ld3+j;
_mm256_stream_pd(s1+off,p0);_mm256_stream_pd(s1+off+4,p1);
_mm256_stream_pd(s2+off,q0);_mm256_stream_pd(s2+off+4,q1);
_mm256_stream_pd(s4+off,r0);_mm256_stream_pd(s4+off+4,r1);
_mm256_stream_pd(s3+off3,t0);_mm256_stream_pd(s3+off3+4,t1);
}
_mm_sfence();
}
static void rec(int n,Mat a,Mat b,double *c,int ldc,double *work) {
if(n<=BASE){leaf(n,a,b,c,ldc);return;}
int h=n/2;size_t z=(size_t)h*h;
Mat aa[4]={{a.x,a.ld},{a.x+h,a.ld},{a.x+(size_t)h*a.ld,a.ld},{a.x+(size_t)h*a.ld+h,a.ld}};
Mat bb[4]={{b.x,b.ld},{b.x+h,b.ld},{b.x+(size_t)h*b.ld,b.ld},{b.x+(size_t)h*b.ld+h,b.ld}};
double *m=n==2*BASE?work+LEAF_PAD:work;
if(n==2*BASE && pending.active && pending.m==m)m=pending_work+LEAF_PAD;
size_t pz=z+(n==2*BASE?LEAF_PAD:UPPER_PAGE_PAD);
double *p=work+3*z,*q=p+z,*sub=q+z;
double *out[7]={c,c+(size_t)h*ldc,c+h,c+(size_t)h*ldc+h,m,m+pz,m+2*pz};
if(n>BASE*2) {
pending_finish();
double *pa0=m+3*pz,*pa1=pa0+pz,*pa2=pa1+pz,*pb0=pa2+pz,*pb1=pb0+pz,*pb2=pb1+pz;
sub=pb2+pz;
pre(h,a,pa0,pa1,pa2,c,ldc,false);
pre(h,b,pb0,pb1,pb2,c+h,ldc,true);
/* mdA_dp2: child order [1,3,4,5,6,0,2] -- constraint "6 before 0 and 2" is the only
hard one; this order puts the three m-panel writers (4,5,6) as late as possible,
which IS the deferral window for the previous sibling's deferred h=1024 combine. */
rec(h,aa[1],bb[2],out[1],ldc,sub);
rec(h,aa[3],{pb2,h},out[3],ldc,sub);
pend2_finish(); /* barrier: next call writes m, the deferred products */
rec(h,{pa0,h},{pb0,h},out[4],h,sub);
rec(h,{pa1,h},{pb1,h},out[5],h,sub);
rec(h,{c,ldc},{c+h,ldc},out[6],h,sub);
rec(h,aa[0],bb[0],out[0],ldc,sub);
rec(h,{pa2,h},bb[3],out[2],ldc,sub);
pending_finish();
pend2_finish(); /* finish an older deferred one */
pend2={c,ldc,m,h,(int)pz,0,true}; /* defer THIS combine (wino form);
m-plane stride = the engine's pz */
return;
}
const int ai[7]={0,2,0,3,0,2,1},aj[7]={3,3,-1,-1,1,0,3},as[7]={1,1,0,0,1,-1,-1};
const int bi[7]={0,0,1,2,3,0,2},bj[7]={3,-1,3,0,-1,1,3},bs[7]={1,0,-1,-1,0,1,1};
const int early = 1; /* sm3_dl: keep the deferred combine disarmed */
/* sm3_dl: the 7 products write their C quadrants directly (no M planes,
no deferred combine). Quadrants: 0=C11 1=C12 2=C21 3=C22. */
g_cq[0]=c; g_cq[1]=c+h; g_cq[2]=c+(size_t)h*ldc; g_cq[3]=c+(size_t)h*ldc+h;
g_stride=(long)ldc*8; g_stride2=0; g_sink=sink_dummy;
for(int r=0;r<7;r++) {
Mat x=aa[ai[r]],y=bb[bi[r]];
if(h<=BASE) {
/* mdE_perm: leaf-order permutation (outputs independent => bit-exact).
lperm = {2,4,5,1,3,6,0} A-adjacency 7 / B-adjacency 5 (natural order: 3/3). */
const int q=g_lperm[r];
Mat x=aa[ai[q]],y=bb[bi[q]];
if(as[q]){x.y=aa[aj[q]].x;x.ly=a.ld;x.sign=as[q];}
if(bs[q]){y.y=bb[bj[q]].x;y.ly=b.ld;y.sign=bs[q];}
if(g_PBC && h==BASE){
packjob_finish();
if(r<6){
const int q2=g_lperm[r+1];
Mat y2=bb[bi[q2]];
if(bs[q2]){y2.y=bb[bj[q2]].x;y2.ly=b.ld;y2.sign=bs[q2];}
packjob_start(y2,g_bp[(r+1)&1],h);
}
g_bpi=r&1; g_pbp=(r>0)?1:0;
}
{
int qq=q;
/* op=0 means overwrite: zero-start + sv=+1 (robust, same code shape) */
g_sv1=g_svp_tab[(DL_O1[qq]==0)?2:(DL_O1[qq]+1)];
g_sv2=g_svp_tab[(DL_O2[qq]==0)?2:(DL_O2[qq]+1)];
g_op1=DL_O1[qq]?1:0; g_op2=DL_O2[qq]?1:0;
g_stride2=(DL_Q2[qq]<0)?0:g_stride;
g_c2=(DL_Q2[qq]<0)?g_sink:g_cq[DL_Q2[qq]];
g_d2on=(DL_Q2[qq]<0)?0:1; /* 0 => the dummy sink gets no offset */
leaf(h,x,y,g_cq[DL_Q1[qq]],ldc);
}
} else {
if(as[r]){sum(h,p,x,aa[aj[r]],as[r]);x={p,h};}
if(bs[r]){sum(h,q,y,bb[bj[r]],bs[r]);y={q,h};}
rec(h,x,y,out[r],r<4?ldc:h,sub);
}
}
if(h==BASE) {
pending_finish();
if(!early) pending={c,ldc,m,0,true};
} else {
pending_finish();
combine(h,c,ldc,m);
}
}
void matrix_multiply(int n,const double *a,const double *b,double *c) {
rec(n,{a,n},{b,n},c,n,workspace);
pending_finish();
pend2_finish();
_mm_sfence();
}
| Compilation | N/A | N/A | Compile OK | Score: N/A | 显示更多 |
| Testcase #1 | 40.061 ms | 10 MB + 308 KB | Accepted | Score: 100 | 显示更多 |