/netgear-R7000-V1.0.7.12_1.2.5/ap/gpl/iserver/libgcrypt-1.5.0/mpi/powerpc32/ |
H A D | mpih-sub1.S | 1 /* PowerPC-32 sub_n -- Subtract two limb vectors of the same length > 0 2 * and store difference in a third limb vector. 51 lwz 8,0(4) # load least significant s1 limb 52 lwz 0,0(5) # load least significant s2 limb 56 Loop: lwzu 8,4(4) # load s1 limb and update s1_ptr 57 lwzu 0,4(5) # load s2 limb and update s2_ptr 58 stwu 7,4(3) # store previous limb in load latency slot 61 Lend: stw 7,4(3) # store ultimate result limb 67 /* Subtract two limb vectors of equal, non-zero length for PowerPC.
|
H A D | mpih-rshift.S | 50 lwz 11,0(4) # load first s1 limb 76 /* Shift a limb right, low level routine. 106 lwz 11,0(4) # load first s1 limb
|
H A D | mpih-lshift.S | 51 lwzu 11,-4(4) # load first s1 limb 77 /* Shift a limb left, low level routine. 105 lwzu %r11,-4(%r4) # load first s1 limb
|
/netgear-R7000-V1.0.7.12_1.2.5/ap/gpl/timemachine/libgcrypt-1.5.0/mpi/powerpc32/ |
H A D | mpih-add1.S | 1 /* PowerPC-32 add_n -- Add two limb vectors of equal, non-zero length. 49 lwz 8,0(4) # load least significant s1 limb 50 lwz 0,0(5) # load least significant s2 limb 54 Loop: lwzu 8,4(4) # load s1 limb and update s1_ptr 55 lwzu 0,4(5) # load s2 limb and update s2_ptr 56 stwu 7,4(3) # store previous limb in load latency slot 59 Lend: stw 7,4(3) # store ultimate result limb 65 /* Add two limb vectors of equal, non-zero length for PowerPC.
|
H A D | mpih-sub1.S | 1 /* PowerPC-32 sub_n -- Subtract two limb vectors of the same length > 0 2 * and store difference in a third limb vector. 51 lwz 8,0(4) # load least significant s1 limb 52 lwz 0,0(5) # load least significant s2 limb 56 Loop: lwzu 8,4(4) # load s1 limb and update s1_ptr 57 lwzu 0,4(5) # load s2 limb and update s2_ptr 58 stwu 7,4(3) # store previous limb in load latency slot 61 Lend: stw 7,4(3) # store ultimate result limb 67 /* Subtract two limb vectors of equal, non-zero length for PowerPC.
|
H A D | mpih-rshift.S | 50 lwz 11,0(4) # load first s1 limb 76 /* Shift a limb right, low level routine. 106 lwz 11,0(4) # load first s1 limb
|
/netgear-R7000-V1.0.7.12_1.2.5/ap/gpl/openssl/crypto/ec/ |
H A D | ecp_nistp521.c | 121 * Each of the nine values is called a 'limb'. Since the limbs are spaced only 123 * bits of each limb overlap with the least significant bits of the next. 130 typedef uint64_t limb; typedef 131 typedef limb felem[NLIMBS]; 134 static const limb bottom57bits = 0x1ffffffffffffff; 135 static const limb bottom58bits = 0x3ffffffffffffff; 143 out[0] = (*((limb *) & in[0])) & bottom58bits; 144 out[1] = (*((limb *) & in[7]) >> 2) & bottom58bits; 145 out[2] = (*((limb *) & in[14]) >> 4) & bottom58bits; 146 out[3] = (*((limb *) [all...] |
H A D | ecp_nistp256.c | 102 * apart, but are 128-bits wide, the most significant bits of each limb overlap 114 typedef uint128_t limb; typedef 115 typedef limb felem[NLIMBS]; 116 typedef limb longfelem[NLIMBS * 2]; 258 # define two105m41m9 (((limb)1) << 105) - (((limb)1) << 41) - (((limb)1) << 9) 259 # define two105 (((limb)1) << 105) 260 # define two105m41p9 (((limb)1) << 105) - (((limb) [all...] |
H A D | ecp_nistp224.c | 62 * A 4-limb representation is an 'felem'; 73 typedef uint64_t limb; typedef 76 typedef limb felem[4]; 419 static const limb two58p2 = (((limb) 1) << 58) + (((limb) 1) << 2); 420 static const limb two58m2 = (((limb) 1) << 58) - (((limb) 1) << 2); 421 static const limb two58m42m [all...] |
/netgear-R7000-V1.0.7.12_1.2.5/ap/gpl/openssl-1.0.2h/crypto/ec/ |
H A D | ecp_nistp521.c | 121 * Each of the nine values is called a 'limb'. Since the limbs are spaced only 123 * bits of each limb overlap with the least significant bits of the next. 130 typedef uint64_t limb; typedef 131 typedef limb felem[NLIMBS]; 134 static const limb bottom57bits = 0x1ffffffffffffff; 135 static const limb bottom58bits = 0x3ffffffffffffff; 143 out[0] = (*((limb *) & in[0])) & bottom58bits; 144 out[1] = (*((limb *) & in[7]) >> 2) & bottom58bits; 145 out[2] = (*((limb *) & in[14]) >> 4) & bottom58bits; 146 out[3] = (*((limb *) [all...] |
H A D | ecp_nistp256.c | 102 * apart, but are 128-bits wide, the most significant bits of each limb overlap 114 typedef uint128_t limb; typedef 115 typedef limb felem[NLIMBS]; 116 typedef limb longfelem[NLIMBS * 2]; 258 # define two105m41m9 (((limb)1) << 105) - (((limb)1) << 41) - (((limb)1) << 9) 259 # define two105 (((limb)1) << 105) 260 # define two105m41p9 (((limb)1) << 105) - (((limb) [all...] |
H A D | ecp_nistp224.c | 62 * A 4-limb representation is an 'felem'; 73 typedef uint64_t limb; typedef 76 typedef limb felem[4]; 419 static const limb two58p2 = (((limb) 1) << 58) + (((limb) 1) << 2); 420 static const limb two58m2 = (((limb) 1) << 58) - (((limb) 1) << 2); 421 static const limb two58m42m [all...] |
/netgear-R7000-V1.0.7.12_1.2.5/ap/gpl/iserver/libgcrypt-1.5.0/mpi/hppa/ |
H A D | mpih-rshift.S | 44 vshd %r22,%r0,%r28 ; compute carry out limb
|
H A D | mpih-lshift.S | 46 vshd %r0,%r22,%r28 ; compute carry out limb
|
/netgear-R7000-V1.0.7.12_1.2.5/ap/gpl/timemachine/libgcrypt-1.5.0/mpi/hppa/ |
H A D | mpih-rshift.S | 44 vshd %r22,%r0,%r28 ; compute carry out limb
|
H A D | mpih-lshift.S | 46 vshd %r0,%r22,%r28 ; compute carry out limb
|
/netgear-R7000-V1.0.7.12_1.2.5/ap/gpl/iserver/libgcrypt-1.5.0/mpi/alpha/ |
H A D | mpih-lshift.S | 39 * This code runs at 4.8 cycles/limb on the 21064. With infinite unrolling, 40 * it would take 4 cycles/limb. It should be possible to get down to 3 41 * cycles/limb since both ldq and stq can be paired with the other used 43 * makes it hard, if not impossible, to get down to 3 cycles/limb: 60 ldq $4,-8($17) # load first limb
|
H A D | mpih-rshift.S | 38 * This code runs at 4.8 cycles/limb on the 21064. With infinite unrolling, 39 * it would take 4 cycles/limb. It should be possible to get down to 3 40 * cycles/limb since both ldq and stq can be paired with the other used 42 * makes it hard, if not impossible, to get down to 3 cycles/limb: 58 ldq $4,0($17) # load first limb
|
/netgear-R7000-V1.0.7.12_1.2.5/ap/gpl/iserver/libgcrypt-1.5.0/mpi/mips3/ |
H A D | mpih-lshift.S | 41 ld $10,-8($5) # load first limb
|
H A D | mpih-rshift.S | 39 ld $10,0($5) # load first limb
|
/netgear-R7000-V1.0.7.12_1.2.5/ap/gpl/timemachine/libgcrypt-1.5.0/mpi/alpha/ |
H A D | mpih-lshift.S | 39 * This code runs at 4.8 cycles/limb on the 21064. With infinite unrolling, 40 * it would take 4 cycles/limb. It should be possible to get down to 3 41 * cycles/limb since both ldq and stq can be paired with the other used 43 * makes it hard, if not impossible, to get down to 3 cycles/limb: 60 ldq $4,-8($17) # load first limb
|
H A D | mpih-rshift.S | 38 * This code runs at 4.8 cycles/limb on the 21064. With infinite unrolling, 39 * it would take 4 cycles/limb. It should be possible to get down to 3 40 * cycles/limb since both ldq and stq can be paired with the other used 42 * makes it hard, if not impossible, to get down to 3 cycles/limb: 58 ldq $4,0($17) # load first limb
|
/netgear-R7000-V1.0.7.12_1.2.5/ap/gpl/timemachine/libgcrypt-1.5.0/mpi/mips3/ |
H A D | mpih-lshift.S | 41 ld $10,-8($5) # load first limb
|
H A D | mpih-rshift.S | 39 ld $10,0($5) # load first limb
|
/netgear-R7000-V1.0.7.12_1.2.5/ap/gpl/iserver/libgcrypt-1.5.0/mpi/hppa1.1/ |
H A D | mpih-mul1.S | 1 /* hppa1.1 mul_1 -- Multiply a limb vector with a limb and store 2 * the result in a second limb vector. 40 * This runs at 9 cycles/limb on a PA7000. With the used instructions, it can 42 * PA7100 it runs at 7 cycles/limb, and that can not be improved either, since 47 * could bring down the times to 8.5 and 6.5 cycles/limb for the PA7000 and 77 ldw -12(%r30),%r19 ; least significant limb in product
|