atl_mm4x3x8p.c
来自「基于Blas CLapck的.用过的人知道是干啥的」· C语言 代码 · 共 342 行 · 第 1/2 页
C
342 行
rC0_0 += m0; m0 = rA0 * rB1; rC1_0 += m1; m1 = rA1 * rB1; ATL_pfl1R(pB0+16); rC2_0 += m2; m2 = rA2 * rB1; rC3_0 += m3; m3 = rA3 * rB1; rB1 = pB0[KB +6]; rC0_1 += m0; m0 = rA0 * rB2; rA0 = pA0[ 6]; rC1_1 += m1; m1 = rA1 * rB2; rA1 = pA0[KB +6]; rC2_1 += m2; m2 = rA2 * rB2; rA2 = pA0[KB2+6]; rC3_1 += m3; m3 = rA3 * rB2; rB2 = pB0[KB2+6]; rC0_2 += m0; m0 = ra0 * rb0; rA3 = pA0[KB3+6]; rC1_2 += m1; m1 = ra1 * rb0; rC2_2 += m2; m2 = ra2 * rb0; rC3_2 += m3; m3 = ra3 * rb0; rb0 = pB0[7]; rC0_0 += m0; m0 = ra0 * rb1; rC1_0 += m1; m1 = ra1 * rb1; ATL_pfl1R(pB0+KB+16); rC2_0 += m2; m2 = ra2 * rb1; rC3_0 += m3; m3 = ra3 * rb1; rb1 = pB0[KB +7]; rC0_1 += m0; m0 = ra0 * rb2; ra0 = pA0[7]; rC1_1 += m1; m1 = ra1 * rb2; ra1 = pA0[KB +7]; rC2_1 += m2; m2 = ra2 * rb2; ra2 = pA0[KB2+7]; rC3_1 += m3; m3 = ra3 * rb2; rb2 = pB0[KB2+7]; rC0_2 += m0; m0 = rA0 * rB0; ra3 = pA0[KB3+7]; rC1_2 += m1; m1 = rA1 * rB0; rC2_2 += m2; m2 = rA2 * rB0; rC3_2 += m3; m3 = rA3 * rB0; rB0 = pB0[8]; rC0_0 += m0; m0 = rA0 * rB1; rC1_0 += m1; m1 = rA1 * rB1; ATL_pfl1R(pB0+KB2+16); rC2_0 += m2; m2 = rA2 * rB1; rC3_0 += m3; m3 = rA3 * rB1; rB1 = pB0[KB +8]; rC0_1 += m0; m0 = rA0 * rB2; rA0 = pA0[ 8]; rC1_1 += m1; m1 = rA1 * rB2; rA1 = pA0[KB +8]; rC2_1 += m2; m2 = rA2 * rB2; rA2 = pA0[KB2+8]; rC3_1 += m3; m3 = rA3 * rB2; rB2 = pB0[KB2+8]; rC0_2 += m0; m0 = ra0 * rb0; rA3 = pA0[KB3+8]; rC1_2 += m1; m1 = ra1 * rb0; rC2_2 += m2; m2 = ra2 * rb0; rC3_2 += m3; m3 = ra3 * rb0; rb0 = pB0[9]; rC0_0 += m0; m0 = ra0 * rb1; rC1_0 += m1; m1 = ra1 * rb1; ATL_pfl1R(pB0+KB3+16); rC2_0 += m2; m2 = ra2 * rb1; rC3_0 += m3; m3 = ra3 * rb1; rb1 = pB0[KB +9]; rC0_1 += m0; m0 = ra0 * rb2; ra0 = pA0[9]; rC1_1 += m1; m1 = ra1 * rb2; ra1 = pA0[KB +9]; rC2_1 += m2; m2 = ra2 * rb2; ra2 = pA0[KB2+9]; rC3_1 += m3; m3 = ra3 * rb2; rb2 = pB0[KB2+9]; rC0_2 += m0; m0 = rA0 * rB0; ra3 = pA0[KB3+9]; pA0 += 8; rC1_2 += m1; m1 = rA1 * rB0; rC2_2 += m2; m2 = rA2 * rB0; rC3_2 += m3; m3 = rA3 * rB0; rB0 = pB0[10]; pB0 += 8; } rC0_0 += m0; m0 = rA0 * rB1; rC1_0 += m1; m1 = rA1 * rB1; ATL_pfl1R(pA0+KB4); rC2_0 += m2; m2 = rA2 * rB1; rC3_0 += m3; m3 = rA3 * rB1; rB1 = pB0[KB +2]; rC0_1 += m0; m0 = rA0 * rB2; rA0 = pA0[ 2]; rC1_1 += m1; m1 = rA1 * rB2; rA1 = pA0[KB +2]; rC2_1 += m2; m2 = rA2 * rB2; rA2 = pA0[KB2+2]; rC3_1 += m3; m3 = rA3 * rB2; rB2 = pB0[KB2+2]; rC0_2 += m0; m0 = ra0 * rb0; rA3 = pA0[KB3+2]; rC1_2 += m1; m1 = ra1 * rb0; rC2_2 += m2; m2 = ra2 * rb0; rC3_2 += m3; m3 = ra3 * rb0; rb0 = pB0[3]; rC0_0 += m0; m0 = ra0 * rb1; rC1_0 += m1; m1 = ra1 * rb1; ATL_pfl1R(pA0+KB5); rC2_0 += m2; m2 = ra2 * rb1; rC3_0 += m3; m3 = ra3 * rb1; rb1 = pB0[KB +3]; rC0_1 += m0; m0 = ra0 * rb2; ra0 = pA0[3]; rC1_1 += m1; m1 = ra1 * rb2; ra1 = pA0[KB +3]; rC2_1 += m2; m2 = ra2 * rb2; ra2 = pA0[KB2+3]; rC3_1 += m3; m3 = ra3 * rb2; rb2 = pB0[KB2+3]; rC0_2 += m0; m0 = rA0 * rB0; ra3 = pA0[KB3+3]; rC1_2 += m1; m1 = rA1 * rB0; rC2_2 += m2; m2 = rA2 * rB0; rC3_2 += m3; m3 = rA3 * rB0; rB0 = pB0[4]; rC0_0 += m0; m0 = rA0 * rB1; rC1_0 += m1; m1 = rA1 * rB1; ATL_pfl1R(pA0+KB6); rC2_0 += m2; m2 = rA2 * rB1; rC3_0 += m3; m3 = rA3 * rB1; rB1 = pB0[KB +4]; rC0_1 += m0; m0 = rA0 * rB2; rA0 = pA0[ 4]; rC1_1 += m1; m1 = rA1 * rB2; rA1 = pA0[KB +4]; rC2_1 += m2; m2 = rA2 * rB2; rA2 = pA0[KB2+4]; rC3_1 += m3; m3 = rA3 * rB2; rB2 = pB0[KB2+4]; rC0_2 += m0; m0 = ra0 * rb0; rA3 = pA0[KB3+4]; rC1_2 += m1; m1 = ra1 * rb0; rC2_2 += m2; m2 = ra2 * rb0; rC3_2 += m3; m3 = ra3 * rb0; rb0 = pB0[5]; rC0_0 += m0; m0 = ra0 * rb1; rC1_0 += m1; m1 = ra1 * rb1; ATL_pfl1R(pA0+KB7); rC2_0 += m2; m2 = ra2 * rb1; rC3_0 += m3; m3 = ra3 * rb1; rb1 = pB0[KB +5]; rC0_1 += m0; m0 = ra0 * rb2; ra0 = pA0[5]; rC1_1 += m1; m1 = ra1 * rb2; ra1 = pA0[KB +5]; rC2_1 += m2; m2 = ra2 * rb2; ra2 = pA0[KB2+5]; rC3_1 += m3; m3 = ra3 * rb2; rb2 = pB0[KB2+5]; rC0_2 += m0; m0 = rA0 * rB0; ra3 = pA0[KB3+5]; rC1_2 += m1; m1 = rA1 * rB0; rC2_2 += m2; m2 = rA2 * rB0; rC3_2 += m3; m3 = rA3 * rB0; rB0 = pB0[6]; rC0_0 += m0; m0 = rA0 * rB1; rC1_0 += m1; m1 = rA1 * rB1; ATL_pfl1R(pB0-KB+8); rC2_0 += m2; m2 = rA2 * rB1; rC3_0 += m3; m3 = rA3 * rB1; rB1 = pB0[KB +6]; rC0_1 += m0; m0 = rA0 * rB2; rA0 = pA0[ 6]; rC1_1 += m1; m1 = rA1 * rB2; rA1 = pA0[KB +6]; rC2_1 += m2; m2 = rA2 * rB2; rA2 = pA0[KB2+6]; rC3_1 += m3; m3 = rA3 * rB2; rB2 = pB0[KB2+6]; rC0_2 += m0; m0 = ra0 * rb0; rA3 = pA0[KB3+6]; rC1_2 += m1; m1 = ra1 * rb0; rC2_2 += m2; m2 = ra2 * rb0; rC3_2 += m3; m3 = ra3 * rb0; rb0 = pB0[7]; rC0_0 += m0; m0 = ra0 * rb1; rC1_0 += m1; m1 = ra1 * rb1; ATL_pfl1R(pB0+8); rC2_0 += m2; m2 = ra2 * rb1; rC3_0 += m3; m3 = ra3 * rb1; rb1 = pB0[KB +7]; rC0_1 += m0; m0 = ra0 * rb2; ra0 = pA0[7]; rC1_1 += m1; m1 = ra1 * rb2; ra1 = pA0[KB +7]; rC2_1 += m2; m2 = ra2 * rb2; ra2 = pA0[KB2+7]; rC3_1 += m3; m3 = ra3 * rb2; rb2 = pB0[KB2+7]; rC0_2 += m0; m0 = rA0 * rB0; ra3 = pA0[KB3+7]; rC1_2 += m1; m1 = rA1 * rB0; pA0 += incAm; rC2_2 += m2; m2 = rA2 * rB0; ATL_pfl1R(pB0+KB+8); rC3_2 += m3; m3 = rA3 * rB0; pB0 += incBm; rC0_0 += m0; m0 = rA0 * rB1; ATL_pfl1W(pC0); rC1_0 += m1; m1 = rA1 * rB1; rC2_0 += m2; m2 = rA2 * rB1; ATL_pfl1W(pC1); rC3_0 += m3; m3 = rA3 * rB1; rC0_1 += m0; m0 = rA0 * rB2; ATL_pfl1W(pC2); rC1_1 += m1; m1 = rA1 * rB2; rC2_1 += m2; m2 = rA2 * rB2; ATL_pfl1R(pC0+(4 SHIFT)); rC3_1 += m3; m3 = rA3 * rB2; rC0_2 += m0; m0 = ra0 * rb0; ATL_pfl1R(pC1+(4 SHIFT)); rC1_2 += m1; m1 = ra1 * rb0; rC2_2 += m2; m2 = ra2 * rb0; ATL_pfl1R(pC2+(4 SHIFT)); rC3_2 += m3; m3 = ra3 * rb0; rC0_0 += m0; m0 = ra0 * rb1; ATL_pfl1R(pB0+8); rC1_0 += m1; m1 = ra1 * rb1; rC2_0 += m2; m2 = ra2 * rb1; rC3_0 += m3; m3 = ra3 * rb1; rC0_1 += m0; m0 = ra0 * rb2; rC1_1 += m1; m1 = ra1 * rb2; ATL_pfl1R(pB0+KB2+8); rC2_1 += m2; m2 = ra2 * rb2; rC3_1 += m3; m3 = ra3 * rb2; rC0_2 += m0; rC1_2 += m1; rC2_2 += m2; rC3_2 += m3; ATL_pfl1R(pB0+KB2+8); #ifdef TREAL *pC0 = rC0_0; pC0[1] = rC1_0; pC0[2] = rC2_0; pC0[3] = rC3_0; pC0 += 4; *pC1 = rC0_1; pC1[1] = rC1_1; pC1[2] = rC2_1; pC1[3] = rC3_1; pC1 += 4; *pC2 = rC0_2; pC2[1] = rC1_2; pC2[2] = rC2_2; pC2[3] = rC3_2; pC2 += 4; #else *pC0 = rC0_0; pC0[2] = rC1_0; pC0[4] = rC2_0; pC0[6] = rC3_0; pC0 += 8; *pC1 = rC0_1; pC1[2] = rC1_1; pC1[4] = rC2_1; pC1[6] = rC3_1; pC1 += 8; *pC2 = rC0_2; pC2[2] = rC1_2; pC2[4] = rC2_2; pC2[6] = rC3_2; pC2 += 8; #endif } while(pA0 != stM); pC0 += incCn; pC1 += incCn; pC2 += incCn; pA0 += incAn; pB0 += incBn; } while(pB0 != stN);}#undef incBn
⌨️ 快捷键说明
复制代码Ctrl + C
搜索代码Ctrl + F
全屏模式F11
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?