Ecosystem/xmrig
Ecosystem/xmrig
1
0
Fork 0
mirror of https://github.com/xmrig/xmrig.git synced 2024-11-18 18:11:05 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Fix typo.

Browse source
This commit is contained in:
XMRig 2019-02-03 20:34:24 +07:00
parent f4170ec7bc
commit 5f113a47cf
2 changed files with 30 additions and 30 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

22
src/crypto/cn_gpu_avx.cpp
View file

@ -88,7 +88,7 @@ inline void round_compute(const __m256& n0, const __m256& n1, const __m256& n2,
// 112×4 = 448 // 112×4 = 448
template <bool add> template <bool add>
inline __m256i double_comupte(const __m256& n0, const __m256& n1, const __m256& n2, const __m256& n3, inline __m256i double_compute(const __m256& n0, const __m256& n1, const __m256& n2, const __m256& n3,
float lcnt, float hcnt, const __m256& rnd_c, __m256& sum) float lcnt, float hcnt, const __m256& rnd_c, __m256& sum)
{ {
__m256 c = _mm256_insertf128_ps(_mm256_castps128_ps256(_mm_set1_ps(lcnt)), _mm_set1_ps(hcnt), 1); __m256 c = _mm256_insertf128_ps(_mm256_castps128_ps256(_mm_set1_ps(lcnt)), _mm_set1_ps(hcnt), 1);
@ -113,10 +113,10 @@ inline __m256i double_comupte(const __m256& n0, const __m256& n1, const __m256&
} }
template <size_t rot> template <size_t rot>
inline void double_comupte_wrap(const __m256& n0, const __m256& n1, const __m256& n2, const __m256& n3, inline void double_compute_wrap(const __m256& n0, const __m256& n1, const __m256& n2, const __m256& n3,
float lcnt, float hcnt, const __m256& rnd_c, __m256& sum, __m256i& out) float lcnt, float hcnt, const __m256& rnd_c, __m256& sum, __m256i& out)
{ {
__m256i r = double_comupte<rot % 2 != 0>(n0, n1, n2, n3, lcnt, hcnt, rnd_c, sum); __m256i r = double_compute<rot % 2 != 0>(n0, n1, n2, n3, lcnt, hcnt, rnd_c, sum);
if(rot != 0) if(rot != 0)
r = _mm256_or_si256(_mm256_bslli_epi128(r, 16 - rot), _mm256_bsrli_epi128(r, rot)); r = _mm256_or_si256(_mm256_bslli_epi128(r, 16 - rot), _mm256_bsrli_epi128(r, rot));
@ -151,10 +151,10 @@ void cn_gpu_inner_avx(const uint8_t* spad, uint8_t* lpad)
n33 = _mm256_permute2f128_ps(n23, n23, 0x11); n33 = _mm256_permute2f128_ps(n23, n23, 0x11);
out = _mm256_setzero_si256(); out = _mm256_setzero_si256();
double_comupte_wrap<0>(n01, n10, n22, n33, 1.3437500f, 1.4296875f, rc, suma, out); double_compute_wrap<0>(n01, n10, n22, n33, 1.3437500f, 1.4296875f, rc, suma, out);
double_comupte_wrap<1>(n01, n22, n33, n10, 1.2812500f, 1.3984375f, rc, suma, out); double_compute_wrap<1>(n01, n22, n33, n10, 1.2812500f, 1.3984375f, rc, suma, out);
double_comupte_wrap<2>(n01, n33, n10, n22, 1.3593750f, 1.3828125f, rc, sumb, out); double_compute_wrap<2>(n01, n33, n10, n22, 1.3593750f, 1.3828125f, rc, sumb, out);
double_comupte_wrap<3>(n01, n33, n22, n10, 1.3671875f, 1.3046875f, rc, sumb, out); double_compute_wrap<3>(n01, n33, n22, n10, 1.3671875f, 1.3046875f, rc, sumb, out);
_mm256_store_si256(idx0, _mm256_xor_si256(v01, out)); _mm256_store_si256(idx0, _mm256_xor_si256(v01, out));
sum0 = _mm256_add_ps(suma, sumb); sum0 = _mm256_add_ps(suma, sumb);
out2 = out; out2 = out;
@ -165,10 +165,10 @@ void cn_gpu_inner_avx(const uint8_t* spad, uint8_t* lpad)
n30 = _mm256_permute2f128_ps(n01, n23, 0x03); n30 = _mm256_permute2f128_ps(n01, n23, 0x03);
out = _mm256_setzero_si256(); out = _mm256_setzero_si256();
double_comupte_wrap<0>(n23, n11, n02, n30, 1.4140625f, 1.3203125f, rc, suma, out); double_compute_wrap<0>(n23, n11, n02, n30, 1.4140625f, 1.3203125f, rc, suma, out);
double_comupte_wrap<1>(n23, n02, n30, n11, 1.2734375f, 1.3515625f, rc, suma, out); double_compute_wrap<1>(n23, n02, n30, n11, 1.2734375f, 1.3515625f, rc, suma, out);
double_comupte_wrap<2>(n23, n30, n11, n02, 1.2578125f, 1.3359375f, rc, sumb, out); double_compute_wrap<2>(n23, n30, n11, n02, 1.2578125f, 1.3359375f, rc, sumb, out);
double_comupte_wrap<3>(n23, n30, n02, n11, 1.2890625f, 1.4609375f, rc, sumb, out); double_compute_wrap<3>(n23, n30, n02, n11, 1.2890625f, 1.4609375f, rc, sumb, out);
_mm256_store_si256(idx2, _mm256_xor_si256(v23, out)); _mm256_store_si256(idx2, _mm256_xor_si256(v23, out));
sum1 = _mm256_add_ps(suma, sumb); sum1 = _mm256_add_ps(suma, sumb);

38
src/crypto/cn_gpu_ssse3.cpp
View file

@ -90,7 +90,7 @@ inline void round_compute(__m128 n0, __m128 n1, __m128 n2, __m128 n3, __m128 rnd
// 112×4 = 448 // 112×4 = 448
template<bool add> template<bool add>
inline __m128i single_comupte(__m128 n0, __m128 n1, __m128 n2, __m128 n3, float cnt, __m128 rnd_c, __m128& sum) inline __m128i single_compute(__m128 n0, __m128 n1, __m128 n2, __m128 n3, float cnt, __m128 rnd_c, __m128& sum)
{ {
__m128 c = _mm_set1_ps(cnt); __m128 c = _mm_set1_ps(cnt);
__m128 r = _mm_setzero_ps(); __m128 r = _mm_setzero_ps();
@ -114,9 +114,9 @@ inline __m128i single_comupte(__m128 n0, __m128 n1, __m128 n2, __m128 n3, floa
} }
template<size_t rot> template<size_t rot>
inline void single_comupte_wrap(__m128 n0, __m128 n1, __m128 n2, __m128 n3, float cnt, __m128 rnd_c, __m128& sum, __m128i& out) inline void single_compute_wrap(__m128 n0, __m128 n1, __m128 n2, __m128 n3, float cnt, __m128 rnd_c, __m128& sum, __m128i& out)
{ {
__m128i r = single_comupte<rot % 2 != 0>(n0, n1, n2, n3, cnt, rnd_c, sum); __m128i r = single_compute<rot % 2 != 0>(n0, n1, n2, n3, cnt, rnd_c, sum);
if(rot != 0) if(rot != 0)
r = _mm_or_si128(_mm_slli_si128(r, 16 - rot), _mm_srli_si128(r, rot)); r = _mm_or_si128(_mm_slli_si128(r, 16 - rot), _mm_srli_si128(r, rot));
out = _mm_xor_si128(out, r); out = _mm_xor_si128(out, r);
@ -149,37 +149,37 @@ void cn_gpu_inner_ssse3(const uint8_t* spad, uint8_t* lpad)
__m128i out, out2; __m128i out, out2;
out = _mm_setzero_si128(); out = _mm_setzero_si128();
single_comupte_wrap<0>(n0, n1, n2, n3, 1.3437500f, rc, suma, out); single_compute_wrap<0>(n0, n1, n2, n3, 1.3437500f, rc, suma, out);
single_comupte_wrap<1>(n0, n2, n3, n1, 1.2812500f, rc, suma, out); single_compute_wrap<1>(n0, n2, n3, n1, 1.2812500f, rc, suma, out);
single_comupte_wrap<2>(n0, n3, n1, n2, 1.3593750f, rc, sumb, out); single_compute_wrap<2>(n0, n3, n1, n2, 1.3593750f, rc, sumb, out);
single_comupte_wrap<3>(n0, n3, n2, n1, 1.3671875f, rc, sumb, out); single_compute_wrap<3>(n0, n3, n2, n1, 1.3671875f, rc, sumb, out);
sum0 = _mm_add_ps(suma, sumb); sum0 = _mm_add_ps(suma, sumb);
_mm_store_si128(idx0, _mm_xor_si128(v0, out)); _mm_store_si128(idx0, _mm_xor_si128(v0, out));
out2 = out; out2 = out;
out = _mm_setzero_si128(); out = _mm_setzero_si128();
single_comupte_wrap<0>(n1, n0, n2, n3, 1.4296875f, rc, suma, out); single_compute_wrap<0>(n1, n0, n2, n3, 1.4296875f, rc, suma, out);
single_comupte_wrap<1>(n1, n2, n3, n0, 1.3984375f, rc, suma, out); single_compute_wrap<1>(n1, n2, n3, n0, 1.3984375f, rc, suma, out);
single_comupte_wrap<2>(n1, n3, n0, n2, 1.3828125f, rc, sumb, out); single_compute_wrap<2>(n1, n3, n0, n2, 1.3828125f, rc, sumb, out);
single_comupte_wrap<3>(n1, n3, n2, n0, 1.3046875f, rc, sumb, out); single_compute_wrap<3>(n1, n3, n2, n0, 1.3046875f, rc, sumb, out);
sum1 = _mm_add_ps(suma, sumb); sum1 = _mm_add_ps(suma, sumb);
_mm_store_si128(idx1, _mm_xor_si128(v1, out)); _mm_store_si128(idx1, _mm_xor_si128(v1, out));
out2 = _mm_xor_si128(out2, out); out2 = _mm_xor_si128(out2, out);
out = _mm_setzero_si128(); out = _mm_setzero_si128();
single_comupte_wrap<0>(n2, n1, n0, n3, 1.4140625f, rc, suma, out); single_compute_wrap<0>(n2, n1, n0, n3, 1.4140625f, rc, suma, out);
single_comupte_wrap<1>(n2, n0, n3, n1, 1.2734375f, rc, suma, out); single_compute_wrap<1>(n2, n0, n3, n1, 1.2734375f, rc, suma, out);
single_comupte_wrap<2>(n2, n3, n1, n0, 1.2578125f, rc, sumb, out); single_compute_wrap<2>(n2, n3, n1, n0, 1.2578125f, rc, sumb, out);
single_comupte_wrap<3>(n2, n3, n0, n1, 1.2890625f, rc, sumb, out); single_compute_wrap<3>(n2, n3, n0, n1, 1.2890625f, rc, sumb, out);
sum2 = _mm_add_ps(suma, sumb); sum2 = _mm_add_ps(suma, sumb);
_mm_store_si128(idx2, _mm_xor_si128(v2, out)); _mm_store_si128(idx2, _mm_xor_si128(v2, out));
out2 = _mm_xor_si128(out2, out); out2 = _mm_xor_si128(out2, out);
out = _mm_setzero_si128(); out = _mm_setzero_si128();
single_comupte_wrap<0>(n3, n1, n2, n0, 1.3203125f, rc, suma, out); single_compute_wrap<0>(n3, n1, n2, n0, 1.3203125f, rc, suma, out);
single_comupte_wrap<1>(n3, n2, n0, n1, 1.3515625f, rc, suma, out); single_compute_wrap<1>(n3, n2, n0, n1, 1.3515625f, rc, suma, out);
single_comupte_wrap<2>(n3, n0, n1, n2, 1.3359375f, rc, sumb, out); single_compute_wrap<2>(n3, n0, n1, n2, 1.3359375f, rc, sumb, out);
single_comupte_wrap<3>(n3, n0, n2, n1, 1.4609375f, rc, sumb, out); single_compute_wrap<3>(n3, n0, n2, n1, 1.4609375f, rc, sumb, out);
sum3 = _mm_add_ps(suma, sumb); sum3 = _mm_add_ps(suma, sumb);
_mm_store_si128(idx3, _mm_xor_si128(v3, out)); _mm_store_si128(idx3, _mm_xor_si128(v3, out));
out2 = _mm_xor_si128(out2, out); out2 = _mm_xor_si128(out2, out);