Implemented the final stage of the Montgomery modular multiplication, i.e.

addition of AB and M then reduction by right-shift.

3 files changed, 331 insertions, 108 deletions

diff --git a/bench/tb_square.v b/bench/tb_square.v
index d35a5cc..733e741 100644
--- a/bench/tb_square.v
+++ b/bench/tb_square.v
@@ -50,87 +50,87 @@ module tb_square;
     //
     initial begin
         //
-        T1[ 0] = 18'h0f13e; T1[ 1] = 18'h0daf6; T1[ 2] = 18'h0aaa9; T1[ 3] = 18'h0c2c2;
-        T1[ 4] = 18'h0fc5f; T1[ 5] = 18'h12164; T1[ 6] = 18'h14375; T1[ 7] = 18'h15615;
-        T1[ 8] = 18'h0d8e2; T1[ 9] = 18'h0ec15; T1[10] = 18'h17c46; T1[11] = 18'h0c922;
-        T1[12] = 18'h08f00; T1[13] = 18'h152f9; T1[14] = 18'h0b0b6; T1[15] = 18'h0ce87;
-        T1[16] = 18'h178f2; T1[17] = 18'h09efb; T1[18] = 18'h0409d; T1[19] = 18'h11104;
-        T1[20] = 18'h0b4a6; T1[21] = 18'h158a6; T1[22] = 18'h0514e; T1[23] = 18'h0ec55;
-        T1[24] = 18'h11e73; T1[25] = 18'h11ddd; T1[26] = 18'h07bd4; T1[27] = 18'h0638b;
-        T1[28] = 18'h0e805; T1[29] = 18'h11c4f; T1[30] = 18'h0a2eb; T1[31] = 18'h05454;
+        T1[ 0] = 18'h191c5; T1[ 1] = 18'h1a118; T1[ 2] = 18'h06e06; T1[ 3] = 18'h0ea69;
+        T1[ 4] = 18'h12944; T1[ 5] = 18'h0c242; T1[ 6] = 18'h0fc64; T1[ 7] = 18'h14efe;
+        T1[ 8] = 18'h113da; T1[ 9] = 18'h06ff7; T1[10] = 18'h0ef0d; T1[11] = 18'h18581;
+        T1[12] = 18'h1a62c; T1[13] = 18'h052b7; T1[14] = 18'h114f7; T1[15] = 18'h1c53e;
+        T1[16] = 18'h0c63e; T1[17] = 18'h0dd14; T1[18] = 18'h0fba8; T1[19] = 18'h1b8e6;
+        T1[20] = 18'h0d944; T1[21] = 18'h10292; T1[22] = 18'h0d276; T1[23] = 18'h027b1;
+        T1[24] = 18'h0c0c7; T1[25] = 18'h100a9; T1[26] = 18'h0a9ab; T1[27] = 18'h0e696;
+        T1[28] = 18'h10798; T1[29] = 18'h0ae91; T1[30] = 18'h08d4d; T1[31] = 18'h0080b;
         //
-        T2[ 0] = 18'h1a479; T2[ 1] = 18'h102f5; T2[ 2] = 18'h10e72; T2[ 3] = 18'h120b1;
-        T2[ 4] = 18'h169cd; T2[ 5] = 18'h1d0c4; T2[ 6] = 18'h11462; T2[ 7] = 18'h12015;
-        T2[ 8] = 18'h16fca; T2[ 9] = 18'h1044f; T2[10] = 18'h122b4; T2[11] = 18'h10a5a;
-        T2[12] = 18'h12620; T2[13] = 18'h0e01a; T2[14] = 18'h095cd; T2[15] = 18'h1278a;
-        T2[16] = 18'h10763; T2[17] = 18'h09fe7; T2[18] = 18'h0d35c; T2[19] = 18'h10e24;
-        T2[20] = 18'h1527d; T2[21] = 18'h115b3; T2[22] = 18'h05443; T2[23] = 18'h1190a;
-        T2[24] = 18'h0fcc3; T2[25] = 18'h115e2; T2[26] = 18'h0a398; T2[27] = 18'h0608d;
-        T2[28] = 18'h13075; T2[29] = 18'h0d816; T2[30] = 18'h0bb4c; T2[31] = 18'h04e8a;
+        T2[ 0] = 18'h1193b; T2[ 1] = 18'h0de9c; T2[ 2] = 18'h0b993; T2[ 3] = 18'h0d2cd;
+        T2[ 4] = 18'h106ad; T2[ 5] = 18'h076da; T2[ 6] = 18'h10cab; T2[ 7] = 18'h15cd5;
+        T2[ 8] = 18'h15425; T2[ 9] = 18'h16287; T2[10] = 18'h0fd64; T2[11] = 18'h06ee0;
+        T2[12] = 18'h1b0c9; T2[13] = 18'h01a5e; T2[14] = 18'h1855c; T2[15] = 18'h17bf9;
+        T2[16] = 18'h1c83c; T2[17] = 18'h158ed; T2[18] = 18'h086df; T2[19] = 18'h16676;
+        T2[20] = 18'h0a0f8; T2[21] = 18'h14545; T2[22] = 18'h09641; T2[23] = 18'h16863;
+        T2[24] = 18'h17e20; T2[25] = 18'h0d457; T2[26] = 18'h05a9b; T2[27] = 18'h1a4cf;
+        T2[28] = 18'h1582a; T2[29] = 18'h1686c; T2[30] = 18'h1394e; T2[31] = 18'h0bdbc;
         //
-        AB[ 0] = 18'h0be4e; AB[ 1] = 18'h0fed7; AB[ 2] = 18'h09496; AB[ 3] = 18'h07181;
-        AB[ 4] = 18'h0ee73; AB[ 5] = 18'h04692; AB[ 6] = 18'h0141a; AB[ 7] = 18'h0078c;
-        AB[ 8] = 18'h030eb; AB[ 9] = 18'h0217c; AB[10] = 18'h0696f; AB[11] = 18'h0a165;
-        AB[12] = 18'h0b753; AB[13] = 18'h04af9; AB[14] = 18'h0ed7c; AB[15] = 18'h079ce;
-        AB[16] = 18'h0e863; AB[17] = 18'h097df; AB[18] = 18'h07984; AB[19] = 18'h048af;
-        AB[20] = 18'h0197f; AB[21] = 18'h0206a; AB[22] = 18'h027e7; AB[23] = 18'h04b3a;
-        AB[24] = 18'h03312; AB[25] = 18'h03b56; AB[26] = 18'h04487; AB[27] = 18'h0bd6a;
-        AB[28] = 18'h04e4b; AB[29] = 18'h069ca; AB[30] = 18'h0f994; AB[31] = 18'h0dd4e;
-        AB[32] = 18'h1b024; AB[33] = 18'h0127f; AB[34] = 18'h02631; AB[35] = 18'h0186b;
-        AB[36] = 18'h03adb; AB[37] = 18'h05368; AB[38] = 18'h059a5; AB[39] = 18'h002e0;
-        AB[40] = 18'h0b78a; AB[41] = 18'h016f3; AB[42] = 18'h0b58d; AB[43] = 18'h03ddb;
-        AB[44] = 18'h078b0; AB[45] = 18'h0073b; AB[46] = 18'h07337; AB[47] = 18'h0c7b0;
-        AB[48] = 18'h00668; AB[49] = 18'h0106d; AB[50] = 18'h01a44; AB[51] = 18'h05ee3;
-        AB[52] = 18'h0462d; AB[53] = 18'h0fdeb; AB[54] = 18'h05f85; AB[55] = 18'h02af9;
-        AB[56] = 18'h0e1c0; AB[57] = 18'h00989; AB[58] = 18'h01201; AB[59] = 18'h0e194;
-        AB[60] = 18'h07f93; AB[61] = 18'h0e739; AB[62] = 18'h07cf6; AB[63] = 18'h019df;
+        N[ 0] = 18'h00f97; N[ 1] = 18'h018bb; N[ 2] = 18'h08a44; N[ 3] = 18'h00858;
+        N[ 4] = 18'h06647; N[ 5] = 18'h0042c; N[ 6] = 18'h0fa09; N[ 7] = 18'h0c8d3;
+        N[ 8] = 18'h0bbc7; N[ 9] = 18'h0e2dd; N[10] = 18'h017fd; N[11] = 18'h0ef4a;
+        N[12] = 18'h002ef; N[13] = 18'h090c1; N[14] = 18'h032db; N[15] = 18'h028b1;
+        N[16] = 18'h05f0a; N[17] = 18'h0ebfd; N[18] = 18'h017ca; N[19] = 18'h09587;
+        N[20] = 18'h0d266; N[21] = 18'h0563c; N[22] = 18'h041af; N[23] = 18'h0433f;
+        N[24] = 18'h08e83; N[25] = 18'h0bc19; N[26] = 18'h000b2; N[27] = 18'h05b53;
+        N[28] = 18'h00e5d; N[29] = 18'h09bc5; N[30] = 18'h0a822; N[31] = 18'h0efff;
         //
-        N_COEFF[ 0] = 18'h05a97; N_COEFF[ 1] = 18'h0ac69; N_COEFF[ 2] = 18'h0d51e; N_COEFF[ 3] = 18'h07326;
-        N_COEFF[ 4] = 18'h01053; N_COEFF[ 5] = 18'h0f68a; N_COEFF[ 6] = 18'h09c70; N_COEFF[ 7] = 18'h064f7;
-        N_COEFF[ 8] = 18'h01041; N_COEFF[ 9] = 18'h0c2bf; N_COEFF[10] = 18'h0f01f; N_COEFF[11] = 18'h01842;
-        N_COEFF[12] = 18'h0e69a; N_COEFF[13] = 18'h037ea; N_COEFF[14] = 18'h0b4a0; N_COEFF[15] = 18'h0c1ab;
-        N_COEFF[16] = 18'h0bd5b; N_COEFF[17] = 18'h09e5e; N_COEFF[18] = 18'h039bd; N_COEFF[19] = 18'h06430;
-        N_COEFF[20] = 18'h0b460; N_COEFF[21] = 18'h08bd4; N_COEFF[22] = 18'h09fcd; N_COEFF[23] = 18'h05391;
-        N_COEFF[24] = 18'h0fa45; N_COEFF[25] = 18'h08892; N_COEFF[26] = 18'h0732c; N_COEFF[27] = 18'h0baf6;
-        N_COEFF[28] = 18'h067a9; N_COEFF[29] = 18'h0b184; N_COEFF[30] = 18'h02089; N_COEFF[31] = 18'h0297b;
-        N_COEFF[32] = 18'h01810;
+        N_COEFF[ 0] = 18'h09fd9; N_COEFF[ 1] = 18'h0b367; N_COEFF[ 2] = 18'h0e467; N_COEFF[ 3] = 18'h0de24;
+        N_COEFF[ 4] = 18'h02022; N_COEFF[ 5] = 18'h0f0e8; N_COEFF[ 6] = 18'h02919; N_COEFF[ 7] = 18'h09901;
+        N_COEFF[ 8] = 18'h0da43; N_COEFF[ 9] = 18'h0023b; N_COEFF[10] = 18'h0ebf8; N_COEFF[11] = 18'h0f04e;
+        N_COEFF[12] = 18'h0942f; N_COEFF[13] = 18'h029e9; N_COEFF[14] = 18'h07cb0; N_COEFF[15] = 18'h08c25;
+        N_COEFF[16] = 18'h04e60; N_COEFF[17] = 18'h05cdc; N_COEFF[18] = 18'h0dff7; N_COEFF[19] = 18'h0279b;
+        N_COEFF[20] = 18'h0610d; N_COEFF[21] = 18'h0f04a; N_COEFF[22] = 18'h001dc; N_COEFF[23] = 18'h03429;
+        N_COEFF[24] = 18'h0f78c; N_COEFF[25] = 18'h0c3e2; N_COEFF[26] = 18'h00ed8; N_COEFF[27] = 18'h039c0;
+        N_COEFF[28] = 18'h02ac2; N_COEFF[29] = 18'h0f703; N_COEFF[30] = 18'h0c54e; N_COEFF[31] = 18'h022d9;
+        N_COEFF[32] = 18'h0f994;
         //
-        Q[ 0] = 18'h0ac02; Q[ 1] = 18'h0a026; Q[ 2] = 18'h06825; Q[ 3] = 18'h08f06;
-        Q[ 4] = 18'h03783; Q[ 5] = 18'h04cb5; Q[ 6] = 18'h0e8ea; Q[ 7] = 18'h083d2;
-        Q[ 8] = 18'h0fec9; Q[ 9] = 18'h066d9; Q[10] = 18'h0edad; Q[11] = 18'h06c12;
-        Q[12] = 18'h0a5fb; Q[13] = 18'h07295; Q[14] = 18'h06a0c; Q[15] = 18'h081a5;
-        Q[16] = 18'h03493; Q[17] = 18'h0a393; Q[18] = 18'h03da6; Q[19] = 18'h0beb1;
-        Q[20] = 18'h0d138; Q[21] = 18'h02815; Q[22] = 18'h0f191; Q[23] = 18'h03617;
-        Q[24] = 18'h08d4f; Q[25] = 18'h0f641; Q[26] = 18'h00e82; Q[27] = 18'h01774;
-        Q[28] = 18'h0bf39; Q[29] = 18'h0929d; Q[30] = 18'h05273; Q[31] = 18'h0c30a;
-        Q[32] = 18'h0eef3;
+        AB[ 0] = 18'h0c199; AB[ 1] = 18'h0957a; AB[ 2] = 18'h070ad; AB[ 3] = 18'h0e5a6;
+        AB[ 4] = 18'h0fec9; AB[ 5] = 18'h00b73; AB[ 6] = 18'h09c72; AB[ 7] = 18'h0cdf0;
+        AB[ 8] = 18'h08755; AB[ 9] = 18'h07560; AB[10] = 18'h084b1; AB[11] = 18'h0ad3f;
+        AB[12] = 18'h074fe; AB[13] = 18'h04d74; AB[14] = 18'h00e16; AB[15] = 18'h0d3b3;
+        AB[16] = 18'h0d418; AB[17] = 18'h02f12; AB[18] = 18'h0c301; AB[19] = 18'h0be2b;
+        AB[20] = 18'h08222; AB[21] = 18'h0056c; AB[22] = 18'h01c7c; AB[23] = 18'h0bc95;
+        AB[24] = 18'h03427; AB[25] = 18'h0c65a; AB[26] = 18'h089ac; AB[27] = 18'h02117;
+        AB[28] = 18'h0ff7d; AB[29] = 18'h01cde; AB[30] = 18'h02709; AB[31] = 18'h01c56;
+        AB[32] = 18'h0f35a; AB[33] = 18'h08ce6; AB[34] = 18'h0a8e5; AB[35] = 18'h0d6d4;
+        AB[36] = 18'h06868; AB[37] = 18'h09105; AB[38] = 18'h0219e; AB[39] = 18'h0bc40;
+        AB[40] = 18'h00e0a; AB[41] = 18'h07783; AB[42] = 18'h0187a; AB[43] = 18'h0b922;
+        AB[44] = 18'h02609; AB[45] = 18'h0c64b; AB[46] = 18'h06b4b; AB[47] = 18'h04b79;
+        AB[48] = 18'h0fed6; AB[49] = 18'h03eac; AB[50] = 18'h04cac; AB[51] = 18'h0d47d;
+        AB[52] = 18'h045fd; AB[53] = 18'h04fa8; AB[54] = 18'h0597c; AB[55] = 18'h0a10d;
+        AB[56] = 18'h0bf44; AB[57] = 18'h08671; AB[58] = 18'h0112a; AB[59] = 18'h08ccf;
+        AB[60] = 18'h0cae5; AB[61] = 18'h04d94; AB[62] = 18'h0b95a; AB[63] = 18'h00040;
         //
-        N[ 0] = 18'h03ad9; N[ 1] = 18'h046b4; N[ 2] = 18'h0e181; N[ 3] = 18'h0fac7;
-        N[ 4] = 18'h0be72; N[ 5] = 18'h029ab; N[ 6] = 18'h07e51; N[ 7] = 18'h037a8;
-        N[ 8] = 18'h0880c; N[ 9] = 18'h05a7d; N[10] = 18'h043c2; N[11] = 18'h038c9;
-        N[12] = 18'h01275; N[13] = 18'h0aa0d; N[14] = 18'h0c0c1; N[15] = 18'h0d035;
-        N[16] = 18'h04082; N[17] = 18'h0543c; N[18] = 18'h0dcb0; N[19] = 18'h0497c;
-        N[20] = 18'h0b12c; N[21] = 18'h013d4; N[22] = 18'h0b80a; N[23] = 18'h051cf;
-        N[24] = 18'h0286c; N[25] = 18'h0b600; N[26] = 18'h0d838; N[27] = 18'h0af4b;
-        N[28] = 18'h08274; N[29] = 18'h06a07; N[30] = 18'h0beea; N[31] = 18'h0f000;
+        Q[ 0] = 18'h021b1; Q[ 1] = 18'h0d2db; Q[ 2] = 18'h0754b; Q[ 3] = 18'h01fc1;
+        Q[ 4] = 18'h063f7; Q[ 5] = 18'h086e5; Q[ 6] = 18'h0bcea; Q[ 7] = 18'h02260;
+        Q[ 8] = 18'h0c54c; Q[ 9] = 18'h0e298; Q[10] = 18'h05d07; Q[11] = 18'h0f978;
+        Q[12] = 18'h0e742; Q[13] = 18'h0a3f0; Q[14] = 18'h0b31e; Q[15] = 18'h041b7;
+        Q[16] = 18'h06ed9; Q[17] = 18'h03ac5; Q[18] = 18'h0f8eb; Q[19] = 18'h0c619;
+        Q[20] = 18'h067e9; Q[21] = 18'h00350; Q[22] = 18'h00376; Q[23] = 18'h02ebf;
+        Q[24] = 18'h0b125; Q[25] = 18'h05f7d; Q[26] = 18'h0f121; Q[27] = 18'h07ba4;
+        Q[28] = 18'h03050; Q[29] = 18'h0642e; Q[30] = 18'h0c2fc; Q[31] = 18'h0dfcf;
+        Q[32] = 18'h03f9e;
         //
-        M[ 0] = 18'h041b2; M[ 1] = 18'h00128; M[ 2] = 18'h06b69; M[ 3] = 18'h08e7e;
-        M[ 4] = 18'h0118c; M[ 5] = 18'h0b96d; M[ 6] = 18'h0ebe5; M[ 7] = 18'h0f873;
-        M[ 8] = 18'h0cf14; M[ 9] = 18'h0de83; M[10] = 18'h09690; M[11] = 18'h05e9a;
-        M[12] = 18'h048ac; M[13] = 18'h0b506; M[14] = 18'h01283; M[15] = 18'h08631;
-        M[16] = 18'h0179c; M[17] = 18'h06820; M[18] = 18'h0867b; M[19] = 18'h0b750;
-        M[20] = 18'h0e680; M[21] = 18'h0df95; M[22] = 18'h0d818; M[23] = 18'h0b4c5;
-        M[24] = 18'h0cced; M[25] = 18'h0c4a9; M[26] = 18'h0bb78; M[27] = 18'h04295;
-        M[28] = 18'h0b1b4; M[29] = 18'h09635; M[30] = 18'h0066b; M[31] = 18'h022b1;
-        M[32] = 18'h04fdb; M[33] = 18'h0efc8; M[34] = 18'h00a14; M[35] = 18'h04bef;
-        M[36] = 18'h006a1; M[37] = 18'h0f1a6; M[38] = 18'h0fc40; M[39] = 18'h0adb5;
-        M[40] = 18'h06e8f; M[41] = 18'h02c60; M[42] = 18'h083e1; M[43] = 18'h0f862;
-        M[44] = 18'h0da61; M[45] = 18'h0dd3d; M[46] = 18'h03381; M[47] = 18'h09db0;
-        M[48] = 18'h05454; M[49] = 18'h07525; M[50] = 18'h0d9c7; M[51] = 18'h0a361;
-        M[52] = 18'h049e0; M[53] = 18'h0a671; M[54] = 18'h0242e; M[55] = 18'h07cb2;
-        M[56] = 18'h02021; M[57] = 18'h0bde1; M[58] = 18'h025aa; M[59] = 18'h0c615;
-        M[60] = 18'h05645; M[61] = 18'h03b46; M[62] = 18'h065d6; M[63] = 18'h0390d;
-        M[64] = 18'h0e005;
+        M[ 0] = 18'h03e67; M[ 1] = 18'h06a85; M[ 2] = 18'h08f52; M[ 3] = 18'h01a59;
+        M[ 4] = 18'h00136; M[ 5] = 18'h0f48c; M[ 6] = 18'h0638d; M[ 7] = 18'h0320f;
+        M[ 8] = 18'h078aa; M[ 9] = 18'h08a9f; M[10] = 18'h07b4e; M[11] = 18'h052c0;
+        M[12] = 18'h08b01; M[13] = 18'h0b28b; M[14] = 18'h0f1e9; M[15] = 18'h02c4c;
+        M[16] = 18'h02be7; M[17] = 18'h0d0ed; M[18] = 18'h03cfe; M[19] = 18'h041d4;
+        M[20] = 18'h07ddd; M[21] = 18'h0fa93; M[22] = 18'h0e383; M[23] = 18'h0436a;
+        M[24] = 18'h0cbd8; M[25] = 18'h039a5; M[26] = 18'h07653; M[27] = 18'h0dee8;
+        M[28] = 18'h00082; M[29] = 18'h0e321; M[30] = 18'h0d8f6; M[31] = 18'h0e3a9;
+        M[32] = 18'h00ca5; M[33] = 18'h035ed; M[34] = 18'h02b8f; M[35] = 18'h063bd;
+        M[36] = 18'h0ec9f; M[37] = 18'h0b8bb; M[38] = 18'h00389; M[39] = 18'h0ca27;
+        M[40] = 18'h0bea7; M[41] = 18'h0df1e; M[42] = 18'h0d685; M[43] = 18'h0cc1b;
+        M[44] = 18'h036c4; M[45] = 18'h01ce9; M[46] = 18'h0c43b; M[47] = 18'h05f58;
+        M[48] = 18'h02c77; M[49] = 18'h03a12; M[50] = 18'h0eea8; M[51] = 18'h0ac31;
+        M[52] = 18'h05838; M[53] = 18'h093ac; M[54] = 18'h0fd54; M[55] = 18'h06e13;
+        M[56] = 18'h002e2; M[57] = 18'h06af4; M[58] = 18'h0ea18; M[59] = 18'h083b3;
+        M[60] = 18'h059f7; M[61] = 18'h016d3; M[62] = 18'h0c3ad; M[63] = 18'h0dbfc;
+        M[64] = 18'h03ba4;
         //  
     end
     
@@ -151,12 +151,17 @@ module tb_square;
     reg [17:0] mgr_fat_bram_y_din;
     
     reg         mac_fat_bram_xy_ena = 1'b0;
+    reg         mac_fat_bram_xy_ena_aux = 1'b0;
     reg         mac_fat_bram_xy_reg_ena = 1'b0;
+    reg         mac_fat_bram_xy_reg_ena_aux = 1'b0;
     reg  [ 2:0] mac_fat_bram_xy_bank;
     reg  [ 2:0] mac_fat_bram_xy_bank_aux;
     reg  [ 7:0] mac_fat_bram_xy_addr[0:4];
     wire [17:0] mac_fat_bram_x_dout[0:4];
     wire [17:0] mac_fat_bram_y_dout[0:4];
+    wire [ 7:0] mac_fat_bram_xy_addr_aux = mac_fat_bram_xy_addr[4]; // handy for debug
+    wire [17:0] mac_fat_bram_x_dout_aux = mac_fat_bram_x_dout[4];   // handy for debug
+    wire [17:0] mac_fat_bram_y_dout_aux = mac_fat_bram_x_dout[4];   // handy for debug
     
     reg        tb_slim_bram_xy_ena = 1'b0;
     reg [ 1:0] tb_slim_bram_xy_bank;
@@ -201,8 +206,8 @@ module tb_square;
                 .dina   (mgr_fat_bram_x_din),
             
                 .clkb   (clk),
-                .enb    (mac_fat_bram_xy_ena),
-                .regceb (mac_fat_bram_xy_reg_ena),
+                .enb    (z < (NUM_MULTS/2) ? mac_fat_bram_xy_ena     : mac_fat_bram_xy_ena_aux),
+                .regceb (z < (NUM_MULTS/2) ? mac_fat_bram_xy_reg_ena : mac_fat_bram_xy_reg_ena_aux),
                 .addrb  ({(z < (NUM_MULTS/2) ?
                     mac_fat_bram_xy_bank : mac_fat_bram_xy_bank_aux), mac_fat_bram_xy_addr[z]}),
                 .doutb  (mac_fat_bram_x_dout[z])
@@ -217,8 +222,8 @@ module tb_square;
                 .dina   (mgr_fat_bram_y_din),
             
                 .clkb   (clk),
-                .enb    (mac_fat_bram_xy_ena),
-                .regceb (mac_fat_bram_xy_reg_ena),
+                .enb    (z < (NUM_MULTS/2) ? mac_fat_bram_xy_ena     : mac_fat_bram_xy_ena_aux),
+                .regceb (z < (NUM_MULTS/2) ? mac_fat_bram_xy_reg_ena : mac_fat_bram_xy_reg_ena_aux),
                 .addrb  ({z < (NUM_MULTS/2) ?
                     mac_fat_bram_xy_bank : mac_fat_bram_xy_bank_aux, mac_fat_bram_xy_addr[z]}),
                 .doutb  (mac_fat_bram_y_dout[z])
@@ -551,6 +556,24 @@ module tb_square;
 
 
     //
+    // Recombinator Interface
+    //
+    wire [ 2:0] recomb_fat_bram_xy_bank;
+    wire [ 7:0] recomb_fat_bram_xy_addr;
+    wire [17:0] recomb_fat_bram_x_dout;
+    wire [17:0] recomb_fat_bram_y_dout;
+    wire        recomb_fat_bram_xy_dout_valid;
+    wire [ 2:0] recomb_slim_bram_xy_bank;
+    wire [ 7:0] recomb_slim_bram_xy_addr;
+    wire [17:0] recomb_slim_bram_x_dout;
+    wire [17:0] recomb_slim_bram_y_dout;
+    wire        recomb_slim_bram_xy_dout_valid;
+    wire        recomb_rdy;
+    
+    
+
+
+    //
     // FSM Transition Logic
     //
     wire [FSM_STATE_WIDTH-1:0] fsm_state_after_mult_square;
@@ -558,6 +581,9 @@ module tb_square;
     wire [FSM_STATE_WIDTH-1:0] fsm_state_after_mult_rectangle;
     
     
+    //
+    // Slim - Address
+    //
     always @(posedge clk)
         //
         case (fsm_state_next)
@@ -596,7 +622,10 @@ module tb_square;
             assign fat_bram_offset_rom[(z-1)/2] = z[2:0];
         end
     endgenerate    
-        
+
+    //
+    // Fat - Address
+    //
     integer j;
     always @(posedge clk) begin
         //
@@ -648,12 +677,14 @@ module tb_square;
             FSM_STATE_MULT_TRIANGLE_COL_0_BUSY,
             FSM_STATE_MULT_TRIANGLE_COL_N_BUSY:   mac_fat_bram_xy_addr[4] <= mac_fat_bram_xy_addr_next(mac_fat_bram_xy_addr[4], index_last);
             //
-            FSM_STATE_MULT_RECTANGLE_COL_0_INIT:   mac_fat_bram_xy_addr[4] <= {5'd0, 3'd1};
-            FSM_STATE_MULT_RECTANGLE_COL_N_INIT:   mac_fat_bram_xy_addr[4] <= {5'd0, 3'd1};
+            FSM_STATE_MULT_RECTANGLE_COL_0_INIT:   mac_fat_bram_xy_addr[4] <= 8'dX;//{5'd0, 3'd0};
+            FSM_STATE_MULT_RECTANGLE_COL_N_INIT,
             FSM_STATE_MULT_RECTANGLE_COL_0_TRIG,
             FSM_STATE_MULT_RECTANGLE_COL_N_TRIG,
             FSM_STATE_MULT_RECTANGLE_COL_0_BUSY,
-            FSM_STATE_MULT_RECTANGLE_COL_N_BUSY:   mac_fat_bram_xy_addr[4] <= mac_fat_bram_xy_addr_next(mac_fat_bram_xy_addr[4], index_last);
+            FSM_STATE_MULT_RECTANGLE_COL_N_BUSY,
+            FSM_STATE_MULT_RECTANGLE_HOLDOFF:    mac_fat_bram_xy_addr[4] <= recomb_fat_bram_xy_dout_valid ? recomb_fat_bram_xy_addr : 8'dX;//recomb_fat_bram_xy_dout_valid && (recomb_fat_bram_xy_bank == BANK_FAT_ML) ?
+                //mac_fat_bram_xy_addr[4] + 1'b1 : mac_fat_bram_xy_addr[4];
             //
             default:                            mac_fat_bram_xy_addr[4] <= 8'dX;
         endcase
@@ -690,7 +721,29 @@ module tb_square;
             default:                            mac_slim_bram_xy_bank <= 2'bXX;
         endcase
 
-    always @(posedge clk)
+    always @(posedge clk) begin
+        //
+        case (fsm_state_next)
+            FSM_STATE_MULT_SQUARE_COL_0_INIT,
+            FSM_STATE_MULT_SQUARE_COL_N_INIT,
+            FSM_STATE_MULT_SQUARE_COL_0_TRIG,
+            FSM_STATE_MULT_SQUARE_COL_N_TRIG,
+            FSM_STATE_MULT_SQUARE_COL_0_BUSY,
+            FSM_STATE_MULT_SQUARE_COL_N_BUSY:    mac_fat_bram_xy_bank <= BANK_FAT_T1T2;
+            FSM_STATE_MULT_TRIANGLE_COL_0_INIT,
+            FSM_STATE_MULT_TRIANGLE_COL_N_INIT,
+            FSM_STATE_MULT_TRIANGLE_COL_0_TRIG,
+            FSM_STATE_MULT_TRIANGLE_COL_N_TRIG:  mac_fat_bram_xy_bank <= BANK_FAT_ABL;
+            FSM_STATE_MULT_TRIANGLE_COL_0_BUSY,
+            FSM_STATE_MULT_TRIANGLE_COL_N_BUSY:   mac_fat_bram_xy_bank <= BANK_FAT_ABL;
+            FSM_STATE_MULT_RECTANGLE_COL_0_INIT,
+            FSM_STATE_MULT_RECTANGLE_COL_N_INIT,
+            FSM_STATE_MULT_RECTANGLE_COL_0_TRIG,
+            FSM_STATE_MULT_RECTANGLE_COL_N_TRIG,    
+            FSM_STATE_MULT_RECTANGLE_COL_0_BUSY,
+            FSM_STATE_MULT_RECTANGLE_COL_N_BUSY:   mac_fat_bram_xy_bank <= BANK_FAT_N;            
+            default:                             mac_fat_bram_xy_bank <= 3'bXXX;
+        endcase
         //
         case (fsm_state_next)
             FSM_STATE_MULT_SQUARE_COL_0_INIT,
@@ -698,21 +751,31 @@ module tb_square;
             FSM_STATE_MULT_SQUARE_COL_0_TRIG,
             FSM_STATE_MULT_SQUARE_COL_N_TRIG,
             FSM_STATE_MULT_SQUARE_COL_0_BUSY,
-            FSM_STATE_MULT_SQUARE_COL_N_BUSY:   {mac_fat_bram_xy_bank_aux, mac_fat_bram_xy_bank} <= {2{BANK_FAT_T1T2}};
+            FSM_STATE_MULT_SQUARE_COL_N_BUSY:   mac_fat_bram_xy_bank_aux <= BANK_FAT_T1T2;
             FSM_STATE_MULT_TRIANGLE_COL_0_INIT,
             FSM_STATE_MULT_TRIANGLE_COL_N_INIT,
             FSM_STATE_MULT_TRIANGLE_COL_0_TRIG,
-            FSM_STATE_MULT_TRIANGLE_COL_N_TRIG: {mac_fat_bram_xy_bank_aux, mac_fat_bram_xy_bank} <= {BANK_FAT_ABH, BANK_FAT_ABL};
+            FSM_STATE_MULT_TRIANGLE_COL_N_TRIG: mac_fat_bram_xy_bank_aux <= BANK_FAT_ABH;
             FSM_STATE_MULT_TRIANGLE_COL_0_BUSY,
-            FSM_STATE_MULT_TRIANGLE_COL_N_BUSY:   {mac_fat_bram_xy_bank_aux, mac_fat_bram_xy_bank} <= {2{BANK_FAT_ABL}};
+            FSM_STATE_MULT_TRIANGLE_COL_N_BUSY:   mac_fat_bram_xy_bank_aux <= BANK_FAT_ABL;
             FSM_STATE_MULT_RECTANGLE_COL_0_INIT,
             FSM_STATE_MULT_RECTANGLE_COL_N_INIT,
             FSM_STATE_MULT_RECTANGLE_COL_0_TRIG,
             FSM_STATE_MULT_RECTANGLE_COL_N_TRIG,    
             FSM_STATE_MULT_RECTANGLE_COL_0_BUSY,
-            FSM_STATE_MULT_RECTANGLE_COL_N_BUSY:   {mac_fat_bram_xy_bank_aux, mac_fat_bram_xy_bank} <= {2{BANK_FAT_N}};            
-            default:                            {mac_fat_bram_xy_bank_aux, mac_fat_bram_xy_bank} <= {2{3'bXXX}};
+            FSM_STATE_MULT_RECTANGLE_COL_N_BUSY,
+            FSM_STATE_MULT_RECTANGLE_HOLDOFF:   if (recomb_fat_bram_xy_dout_valid)
+                case (recomb_fat_bram_xy_bank)
+                    BANK_FAT_ML: mac_fat_bram_xy_bank_aux <= BANK_FAT_ABL;
+                    BANK_FAT_MH: mac_fat_bram_xy_bank_aux <= BANK_FAT_ABH;
+                    BANK_FAT_EXT: mac_fat_bram_xy_bank_aux <= BANK_FAT_EXT;
+                    default: mac_fat_bram_xy_bank_aux <= 3'bXXX; 
+                 endcase
+                 else mac_fat_bram_xy_bank_aux <= 3'bXXX;
+            default:                            mac_fat_bram_xy_bank_aux <= 3'bXXX;
         endcase
+        //
+    end
 
 
 
@@ -740,7 +803,7 @@ module tb_square;
             default:                              mac_slim_bram_xy_ena <= 1'b0;
         endcase
 
-    always @(posedge clk)
+    always @(posedge clk) begin
         //
         case (fsm_state_next)
             FSM_STATE_MULT_SQUARE_COL_0_INIT,
@@ -763,7 +826,25 @@ module tb_square;
             FSM_STATE_MULT_RECTANGLE_COL_N_BUSY:   mac_fat_bram_xy_ena <= 1'b1;
             default:                            mac_fat_bram_xy_ena <= 1'b0;
         endcase
-
+        //
+        case (fsm_state_next)
+            FSM_STATE_MULT_TRIANGLE_COL_0_INIT,
+            FSM_STATE_MULT_TRIANGLE_COL_N_INIT,
+            FSM_STATE_MULT_TRIANGLE_COL_0_TRIG,
+            FSM_STATE_MULT_TRIANGLE_COL_N_TRIG,
+            FSM_STATE_MULT_TRIANGLE_COL_0_BUSY,
+            FSM_STATE_MULT_TRIANGLE_COL_N_BUSY: mac_fat_bram_xy_ena_aux <= 1'b1;
+            FSM_STATE_MULT_RECTANGLE_COL_0_INIT: mac_fat_bram_xy_ena_aux <= 1'b0;//1'b1;
+            FSM_STATE_MULT_RECTANGLE_COL_N_INIT,
+            FSM_STATE_MULT_RECTANGLE_COL_0_TRIG,
+            FSM_STATE_MULT_RECTANGLE_COL_N_TRIG,
+            FSM_STATE_MULT_RECTANGLE_COL_0_BUSY,
+            FSM_STATE_MULT_RECTANGLE_COL_N_BUSY,
+            FSM_STATE_MULT_RECTANGLE_HOLDOFF:   mac_fat_bram_xy_ena_aux <= recomb_fat_bram_xy_dout_valid;// && (recomb_fat_bram_xy_bank == BANK_FAT_ML);
+            default:                            mac_fat_bram_xy_ena_aux <= 1'b0;
+        endcase
+        //
+    end
 
     always @(posedge clk)
         //
@@ -771,9 +852,9 @@ module tb_square;
         
     always @(posedge clk)
         //
-        mac_fat_bram_xy_reg_ena <= mac_fat_bram_xy_ena;
+        {mac_fat_bram_xy_reg_ena_aux, mac_fat_bram_xy_reg_ena} <= {mac_fat_bram_xy_ena_aux, mac_fat_bram_xy_ena};
           
-    reg ladder_mode = 1'b0; // 0 = X:T1*T2, Y:T2*T2
+    reg ladder_mode = 1'b1; // 0 = X:T1*T2, Y:T2*T2
                             // 1 = X:T1*T2, Y:T2*T1
           
 
@@ -948,18 +1029,6 @@ module tb_square;
         //
     end
     
-    wire [ 2:0] recomb_fat_bram_xy_bank;
-    wire [ 7:0] recomb_fat_bram_xy_addr;
-    wire [17:0] recomb_fat_bram_x_dout;
-    wire [17:0] recomb_fat_bram_y_dout;
-    wire        recomb_fat_bram_xy_dout_valid;
-    wire [ 2:0] recomb_slim_bram_xy_bank;
-    wire [ 7:0] recomb_slim_bram_xy_addr;
-    wire [17:0] recomb_slim_bram_x_dout;
-    wire [17:0] recomb_slim_bram_y_dout;
-    wire        recomb_slim_bram_xy_dout_valid;
-    wire        recomb_rdy;
-    
     modexpng_part_recombinator recomb
     (
         .clk                            (clk),
@@ -1167,5 +1236,144 @@ module tb_square;
     end
     
     
+    
+    //
+    // Reductor
+    //
+    reg reductor_ena = 1'b0;
+    
+    always @(posedge clk)
+        //
+        if (!reductor_ena)
+            case (fsm_state)
+                FSM_STATE_MULT_RECTANGLE_COL_0_INIT: reductor_ena <= 1'b1;  
+            endcase
+        else begin
+        
+        
+        end
+    
+    
+    reg recomb_fat_bram_xy_dout_valid_dly1;
+    reg recomb_fat_bram_xy_dout_valid_dly2;
+    reg recomb_fat_bram_xy_dout_valid_dly3;
+
+    reg [2:0] recomb_fat_bram_xy_bank_dly1;
+    reg [2:0] recomb_fat_bram_xy_bank_dly2;
+    reg [2:0] recomb_fat_bram_xy_bank_dly3;
+
+    reg [7:0] recomb_fat_bram_xy_addr_dly1;
+    reg [7:0] recomb_fat_bram_xy_addr_dly2;
+    reg [7:0] recomb_fat_bram_xy_addr_dly3;
+    
+    always @(posedge clk) begin
+        //
+        recomb_fat_bram_xy_dout_valid_dly1 <= recomb_fat_bram_xy_dout_valid;
+        recomb_fat_bram_xy_dout_valid_dly2 <= recomb_fat_bram_xy_dout_valid_dly1;
+        recomb_fat_bram_xy_dout_valid_dly3 <= recomb_fat_bram_xy_dout_valid_dly2;
+        //
+    end
+    
+    reg [17:0] recomb_fat_bram_x_dout_dly1;
+    reg [17:0] recomb_fat_bram_x_dout_dly2;
+    reg [17:0] recomb_fat_bram_x_dout_dly3;
+    
+    always @(posedge clk) begin
+        //
+        if (recomb_fat_bram_xy_dout_valid) recomb_fat_bram_x_dout_dly1 <= recomb_fat_bram_x_dout;
+        if (recomb_fat_bram_xy_dout_valid_dly1) recomb_fat_bram_x_dout_dly2 <= recomb_fat_bram_x_dout_dly1;
+        if (recomb_fat_bram_xy_dout_valid_dly2) recomb_fat_bram_x_dout_dly3 <= recomb_fat_bram_x_dout_dly2;
+        //
+    end
+    
+    always @(posedge clk) begin
+        //
+        if (recomb_fat_bram_xy_dout_valid) recomb_fat_bram_xy_bank_dly1 <= recomb_fat_bram_xy_bank;
+        if (recomb_fat_bram_xy_dout_valid_dly1) recomb_fat_bram_xy_bank_dly2 <= recomb_fat_bram_xy_bank_dly1;
+        if (recomb_fat_bram_xy_dout_valid_dly2) recomb_fat_bram_xy_bank_dly3 <= recomb_fat_bram_xy_bank_dly2;
+        //
+    end
+    
+    always @(posedge clk) begin
+        //
+        if (recomb_fat_bram_xy_dout_valid) recomb_fat_bram_xy_addr_dly1 <= recomb_fat_bram_xy_addr;
+        if (recomb_fat_bram_xy_dout_valid_dly1) recomb_fat_bram_xy_addr_dly2 <= recomb_fat_bram_xy_addr_dly1;
+        if (recomb_fat_bram_xy_dout_valid_dly2) recomb_fat_bram_xy_addr_dly3 <= recomb_fat_bram_xy_addr_dly2;
+        //
+    end
+    
+   
+    reg [ 1:0] reductor_fat_bram_x_lsb_carry;
+    reg [15:0] reductor_fat_bram_x_lsb_dummy;
+    reg [17:0] reductor_fat_bram_x_lsb_dout;
+
+    reg [17:0] reductor_fat_bram_x_msb_dout;
+    
+    always @(posedge clk)
+        //
+        if (!reductor_ena) begin
+            reductor_fat_bram_x_lsb_carry <= 2'b00;
+        end else if (recomb_fat_bram_xy_dout_valid_dly3) begin
+            
+            case (recomb_fat_bram_xy_bank_dly3)
+                BANK_FAT_ML: {reductor_fat_bram_x_lsb_carry, reductor_fat_bram_x_lsb_dummy} <= recomb_fat_bram_x_dout_dly3 + mac_fat_bram_x_dout_aux + reductor_fat_bram_x_lsb_carry;
+                BANK_FAT_MH:
+                    if (recomb_fat_bram_xy_addr_dly3 == 8'd0)
+                        {reductor_fat_bram_x_lsb_carry, reductor_fat_bram_x_lsb_dummy} <= recomb_fat_bram_x_dout_dly3 + mac_fat_bram_x_dout_aux + reductor_fat_bram_x_lsb_carry;
+                    else if (recomb_fat_bram_xy_addr_dly3 == 8'd1)
+                        reductor_fat_bram_x_msb_dout <= recomb_fat_bram_x_dout_dly3 + mac_fat_bram_x_dout_aux + reductor_fat_bram_x_lsb_carry; 
+                    else
+                        reductor_fat_bram_x_msb_dout <= recomb_fat_bram_x_dout_dly3 + mac_fat_bram_x_dout_aux;
+                BANK_FAT_EXT:
+                    reductor_fat_bram_x_msb_dout <= recomb_fat_bram_x_dout_dly3;
+            endcase
+            // 
+        end
+    /*
+
+
+    reg [17:0] recomb_fat_bram_x_dout_dly1;
+    reg [17:0] recomb_fat_bram_x_dout_dly2;
+
+    reg [ 2:0] recomb_fat_bram_xy_bank_dly1;
+    reg [ 2:0] recomb_fat_bram_xy_bank_dly2;
+    
+    reg [1:0] reductor_fat_bram_x_carry;
+    
+    reg [15:0] reductor_fat_bram_x_dummy;
+    reg [17:0] reductor_fat_bram_x_dout;
+    reg        reductor_fat_bram_xy_dout_valid;
+    
+    always @(posedge clk)
+        //
+        if (reductor_ena) begin
+        
+            if (recomb_fat_bram_xy_dout_valid) begin
+                recomb_fat_bram_x_dout_dly1 <= recomb_fat_bram_x_dout;
+                recomb_fat_bram_xy_bank_dly1 <= recomb_fat_bram_xy_bank;
+            end
+                
+            if (mac_fat_bram_xy_ena_aux) begin
+                recomb_fat_bram_x_dout_dly2 <= recomb_fat_bram_x_dout_dly1;
+                recomb_fat_bram_xy_bank_dly2 <= recomb_fat_bram_xy_bank_dly1;
+            end
+
+            if (mac_fat_bram_xy_reg_ena_aux)
+                case (recomb_fat_bram_xy_bank_dly2)
+                    BANK_FAT_ML: {reductor_fat_bram_x_carry, reductor_fat_bram_x_dummy} <= recomb_fat_bram_x_dout_dly2 + mac_fat_bram_x_dout_aux + reductor_fat_bram_x_carry;
+                endcase
+    
+            //reductor_fat_bram_xy_dout_valid <= mac_fat_bram_xy_reg_ena_aux;
+            
+        end else begin
+        
+            reductor_fat_bram_x_carry <= 2'b00;
+            reductor_fat_bram_xy_dout_valid <= 1'b0;
+        
+        end
+    */
+    
+    
+    
 endmodule
 
diff --git a/rtl/modexpng_parameters.vh b/rtl/modexpng_parameters.vh
index 57eef35..77b57f3 100644
--- a/rtl/modexpng_parameters.vh
+++ b/rtl/modexpng_parameters.vh
@@ -7,10 +7,10 @@ localparam [2:0] BANK_FAT_T1T2      = 3'd0;
 localparam [2:0] BANK_FAT_ABL       = 3'd1;
 localparam [2:0] BANK_FAT_ABH       = 3'd2;
 localparam [2:0] BANK_FAT_N         = 3'd3;
-localparam [2:0] BANK_FAT_ML        = 3'd4;
-localparam [2:0] BANK_FAT_MH        = 3'd5;
+localparam [2:0] BANK_FAT_ML        = 3'd4; // not needed
+localparam [2:0] BANK_FAT_MH        = 3'd5; // not needed
 localparam [2:0] BANK_FAT_EXT       = 3'd6; // 0 -> MH'
-localparam [2:0] BANK_FAT_UNUSED    = 3'd7;
+localparam [2:0] BANK_FAT_UNUSED    = 3'd7; // not needed
 
 localparam [1:0] BANK_SLIM_T1T2     = 2'd0;
 localparam [1:0] BANK_SLIM_N_COEFF  = 2'd1;
diff --git a/rtl/modexpng_part_recombinator.v b/rtl/modexpng_part_recombinator.v
index 567ecd5..957ba8e 100644
--- a/rtl/modexpng_part_recombinator.v
+++ b/rtl/modexpng_part_recombinator.v
@@ -1095,6 +1095,7 @@ module modexpng_part_recombinator
                                 end
                                 //
                             end
+                            //
                         endcase
                 //
                 2'd2:   case (rcmb_xy_dout_valid)
@@ -1103,6 +1104,20 @@ module modexpng_part_recombinator
                             //
                         endcase
                 //
+                2'd3: case (rcmb_xy_dout_valid)
+                                            //
+                                            2'b00: begin
+                                                //
+                                                if (recomb_msb_flag_delay_2) begin
+                                                    //
+                                                    rdy_adv <= ~recomb_msb_flag_delay_1;
+                                                    //
+                                                end
+                                                //
+                                            end
+                                            //
+                                        endcase
+                //
             endcase
             //        
         end