popov/cpuminer-opt-gpu
1
0
Fork 0
mirror of https://github.com/JayDDee/cpuminer-opt.git synced 2026-02-22 16:33:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

v24.2

Browse Source
This commit is contained in:
Jay D Dee
2024-05-20 23:08:50 -04:00
parent 4f930574cc
commit 042d13d1e1
129 changed files with 835 additions and 538 deletions
Download Patch File Download Diff File Expand all files Collapse all files

51
cpu-miner.c
View File

@@ -2852,12 +2852,14 @@ static bool cpu_capability( bool display_only )
bool cpu_has_avx10 = has_avx10();
bool cpu_has_aes = has_aes_ni(); // x86_64 or AArch64 AES
bool cpu_has_vaes = has_vaes();
bool cpu_has_sha = has_sha(); // x86_64 or AArch64
bool cpu_has_sha256 = has_sha(); // x86_64 or AArch64
bool cpu_has_sha512 = has_sha512();
bool sw_has_x86_64 = false;
bool sw_has_aarch64 = false;
int sw_arm_arch = 0;
bool sw_has_neon = false;
int sw_arm_arch = 0; // AArch64
bool sw_has_neon = false; // AArch64
// bool sw_has_sve = false; // AArch64
// bool sw_has_sve2 = false; // AArch64
bool sw_has_sse2 = false; // x86_64
bool sw_has_ssse3 = false; // x86_64
bool sw_has_sse41 = false; // x86_64
@@ -2865,9 +2867,11 @@ static bool cpu_capability( bool display_only )
bool sw_has_avx = false;
bool sw_has_avx2 = false;
bool sw_has_avx512 = false;
bool sw_has_avx10_256 = false;
bool sw_has_avx10_512 = false;
bool sw_has_aes = false;
bool sw_has_vaes = false;
bool sw_has_sha = false; // x86_64 or AArch64 SHA2
bool sw_has_sha256 = false; // x86_64 or AArch64 SHA2
bool sw_has_sha512 = false; // x86_64 or AArch64 SHA3
set_t algo_features = algo_gate.optimizations;
bool algo_has_sse2 = set_incl( SSE2_OPT, algo_features );
@@ -2877,7 +2881,7 @@ static bool cpu_capability( bool display_only )
bool algo_has_avx512 = set_incl( AVX512_OPT, algo_features );
bool algo_has_aes = set_incl( AES_OPT, algo_features );
bool algo_has_vaes = set_incl( VAES_OPT, algo_features );
bool algo_has_sha = set_incl( SHA_OPT, algo_features );
bool algo_has_sha256 = set_incl( SHA_OPT, algo_features );
bool algo_has_sha512 = set_incl( SHA512_OPT, algo_features );
bool algo_has_neon = set_incl( NEON_OPT, algo_features );
bool use_sse2;
@@ -2887,7 +2891,7 @@ static bool cpu_capability( bool display_only )
bool use_avx512;
bool use_aes;
bool use_vaes;
bool use_sha;
bool use_sha256;
bool use_sha512;
bool use_neon;
bool use_none;
@@ -2925,6 +2929,13 @@ static bool cpu_capability( bool display_only )
#if (defined(__AVX512F__) && defined(__AVX512DQ__) && defined(__AVX512BW__) && defined(__AVX512VL__))
sw_has_avx512 = true;
#endif
#if defined(__AVX10_1_256__)
sw_has_avx10_256 = true;
#endif
#if defined(__AVX10_1_512__)
sw_has_avx10_512 = true;
#endif
#if defined(__AES__) || defined(__ARM_FEATURE_AES)
sw_has_aes = true;
#endif
@@ -2932,16 +2943,21 @@ static bool cpu_capability( bool display_only )
sw_has_vaes = true;
#endif
#if defined(__SHA__) || defined(__ARM_FEATURE_SHA2)
sw_has_sha = true;
sw_has_sha256 = true;
#endif
#if defined(__SHA512__) || defined(____ARM_FEATURE_SHA3)
#if defined(__SHA512__) || defined(__ARM_FEATURE_SHA3)
sw_has_sha512 = true;
#endif
#if defined(__ARM_NEON)
sw_has_neon = true;
#endif
// #if defined(__ARM_FEATURE_SVE)
// sw_has_sve = true;
// #endif
// #if defined(__ARM_FEATURE_SVE2)
// sw_has_sve2 = true;
// #endif
cpu_brand_string( cpu_brand );
printf( "CPU: %s\n", cpu_brand );
@@ -2983,7 +2999,7 @@ static bool cpu_capability( bool display_only )
if ( cpu_has_vaes ) printf( " VAES" );
else if ( cpu_has_aes ) printf( " AES" );
if ( cpu_has_sha512 ) printf( " SHA512" );
else if ( cpu_has_sha ) printf( " SHA256" );
else if ( cpu_has_sha256 ) printf( " SHA256" );
if ( cpu_has_avx10 ) printf( " AVX10.%d-%d",
avx10_version(), avx10_vector_length() );
@@ -2998,17 +3014,22 @@ static bool cpu_capability( bool display_only )
else if ( sw_has_sse41 ) printf( " SSE4.1" );
else if ( sw_has_ssse3 ) printf( " SSSE3 " );
else if ( sw_has_sse2 ) printf( " SSE2 " );
if ( sw_has_avx10_512 ) printf( " AVX10-512" );
else if ( sw_has_avx10_256 ) printf( " AVX10-256" );
}
else if ( sw_has_aarch64 )
{
printf( " AArch64" );
if ( sw_arm_arch ) printf( " armv%d", sw_arm_arch );
if ( sw_has_neon ) printf( " NEON" );
// if ( sw_has_sve ) printf( " SVE" );
// else if ( sw_has_sve2 ) printf( " SVE2" );
}
if ( sw_has_vaes ) printf( " VAES" );
else if ( sw_has_aes ) printf( " AES" );
if ( sw_has_sha512 ) printf( " SHA512" );
else if ( sw_has_sha ) printf( " SHA256" );
else if ( sw_has_sha256 ) printf( " SHA256" );
if ( !display_only )
{
@@ -3024,7 +3045,7 @@ static bool cpu_capability( bool display_only )
if ( algo_has_vaes ) printf( " VAES" );
else if ( algo_has_aes ) printf( " AES" );
if ( algo_has_sha512 ) printf( " SHA512" );
else if ( algo_has_sha ) printf( " SHA256" );
else if ( algo_has_sha256 ) printf( " SHA256" );
}
}
printf("\n");
@@ -3068,11 +3089,11 @@ static bool cpu_capability( bool display_only )
use_avx512 = cpu_has_avx512 && sw_has_avx512 && algo_has_avx512;
use_aes = cpu_has_aes && sw_has_aes && algo_has_aes;
use_vaes = cpu_has_vaes && sw_has_vaes && algo_has_vaes;
use_sha = cpu_has_sha && sw_has_sha && algo_has_sha;
use_sha256 = cpu_has_sha256 && sw_has_sha256 && algo_has_sha256;
use_sha512 = cpu_has_sha512 && sw_has_sha512 && algo_has_sha512;
use_neon = sw_has_aarch64 && sw_has_neon && algo_has_neon;
use_none = !( use_sse2 || use_sse42 || use_avx || use_aes || use_avx512
|| use_avx2 || use_sha || use_vaes || use_sha512 || use_neon );
|| use_avx2 || use_sha256 || use_vaes || use_sha512 || use_neon );
// Display best options
applog_nl( "Enabled optimizations:" );
@@ -3090,7 +3111,7 @@ static bool cpu_capability( bool display_only )
if ( use_vaes ) printf( " VAES" );
else if ( use_aes ) printf( " AES" );
if ( use_sha512 ) printf( " SHA512" );
else if ( use_sha ) printf( " SHA256" );
else if ( use_sha256 ) printf( " SHA256" );
if ( use_neon ) printf( " NEON" );
}
printf( "\n" );