DragonX/xmrig-hac
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 1 Wiki Activity
3,152 Commits 2 Branches 1 Tag
3cdfe6b30e1fde5c92dcb6a7f4272d35abb26fe9
Commit Graph

619 Commits

Author SHA1 Message Date
SChernykh
4fbff43988 RandomX ARMv8: optimized dataset read 
Break dependency from readReg2 and readReg3. It should run faster on superscalar and out-of-order CPUs i.e. Apple M1.
 2021-05-20 21:24:28 +02:00
SChernykh
5898413858 RandomX: rewrote dataset read code 
Unified code for AMD and Intel
1% faster on Intel
0.15% faster on AMD Ryzen
 2021-05-20 12:45:42 +02:00
SChernykh
5bd0c4be25 RandomX: added BMI2 version for scratchpad prefetch 
Saves 1 instruction and 1 byte in the main loop.
 2021-05-19 17:52:16 +02:00
SChernykh
ac1e88882c Enabled IMUL_RCP optimization for light mode mining 
Better fix for #2377
 2021-05-17 11:26:40 +02:00
SChernykh
3c9bb68880 Fixed broken light mode mining on x86 2021-05-15 21:41:39 +02:00
SChernykh
112405390a Update sse2neon.h 2021-05-03 18:08:59 +02:00
SChernykh
48d777c0f1 Fixed Zen3 asm for cn/upx2 
- Invalid rounding mode was used which caused rejected shares sometimes
- Also optimized CN implode/explode functions a bit.
 2021-04-21 13:22:25 +02:00
SChernykh
dcbd0bf8c4 RandomX: optimized IMUL_RCP instruction 
+0.4% on AMD Zen2
+0.3% on AMD Zen3
+0.1% on Intel SandyBridge
+0.3% on rx/wow on Intel SandyBridge
 2021-04-19 17:43:58 +02:00
SChernykh
775afaeaa1 Optimized cn/upx for Zen3 
0.9% faster
 2021-04-19 12:29:44 +02:00
SChernykh
a2ece91122 Fix dvision by zero check in percent() 2021-04-19 12:05:07 +02:00
SChernykh
e3ca348fe5 Fixed rounding mode after running cn/upx 2021-04-19 12:02:57 +02:00
SChernykh
c4845d15c2 Optimized cn/upx2 for Ryzen CPUs 2021-04-17 18:18:26 +02:00
SChernykh
41969df701 Update CnHash.cpp 2021-04-17 15:06:31 +02:00
SChernykh
b3d14f51b9 Added support for Uplexa (cn/upx2 algorithm) 2021-04-17 14:53:42 +02:00
SChernykh
97525616d6 Fixed cn-heavy for GCC-8 2021-04-04 10:18:27 +02:00
SChernykh
0b8f89890c Fixed compilation for ARM 2021-04-03 17:50:52 +02:00
xmrig
28b9f07a3c Merge pull request #2228 from esrrhs/dev 
remove useless v4_random_math_init if algo is not cn/r
 2021-04-02 15:49:53 +07:00
SChernykh
2ef064da60 Don't use RandomX JIT if WITH_ASM=OFF 
Because RandomX JIT use asm code
 2021-04-02 10:05:46 +02:00
esrrhs
43a586ce7a remove useless v4_random_math_init if algo is not cn/r 2021-04-02 14:59:09 +08:00
SChernykh
f6232d0753 Optimize cn-heavy in GCC builds 
+0.7% in GCC builds, but GCC is still slower than MSVC on cn-heavy.
 2021-03-28 16:12:09 +02:00
SChernykh
f4a43bb4b0 Optimized cn-heavy 
- Remove unnecessary type conversion when doing `idx0 = d ^ q;`
- Saves 1 CPU cycle in the main loop
- 0.2% speedup on Ryzen 5 5600X, results on other CPUs may vary
 2021-03-27 22:21:01 +01:00
SChernykh
4b270047fe Fix vld1q_u8_x4 compilation error with GCC 10.2 2021-03-12 16:26:02 +01:00
TheGreatMcPain
97908e081f Update sse2neon.h to the latest master. Fixes build on armv7. 
A few days after this header was introduced. Upstream updated it with
armv7 versions of `_mm_aesenc_si128` which allows xmrig to build
on armv7.
 2021-03-02 01:33:25 -06:00
SChernykh
83ccef753a Fixed crash when GPU mining cn-heavy on Zen3 system 2021-02-18 14:49:37 +01:00
SChernykh
f36b4e36b1 Cryptonight: add prefetching to interleaved mode 2021-02-07 23:29:54 +01:00
SChernykh
3e9755c8c8 Optimized cn-heavy for Zen3 
- Uses scratchpad interleaving to access only the closest L3 slice from each CPU core.
- Also activates MSR mod for cn-heavy because CPU prefetchers get confused with interleaving
- 7-8% speedup on Zen3
 2021-02-07 22:05:11 +01:00
SChernykh
3f8732e593 Fix compile error in Termux 2021-02-03 19:05:05 +01:00
XMRig
465dc7edad Fixed macOS build and compile warning. 2021-02-02 13:53:45 +07:00
SChernykh
771fb40eca Update jit_compiler_a64.cpp 2021-02-01 22:52:02 +01:00
SChernykh
e92767eed5 ARM JIT: added missing cache flush 2021-02-01 22:42:35 +01:00
SChernykh
826e540f15 Fixed code allocation for ARM 2021-02-01 22:36:11 +01:00
xmrig
8a40921ab9 Merge pull request #2076 from xmrig/feature-flexible-hugepages 
Added support for flexible huge page sizes on Linux.
 2021-02-02 04:07:41 +07:00
SChernykh
e3828a13f9 Fixed crashes on ARM 2021-02-01 17:07:45 +01:00
XMRig
c92f9df8ab Added support for flexible huge page sizes on Linux. 2021-01-31 23:38:57 +07:00
xmrig
0da777bfe3 Merge pull request #2058 from SChernykh/dev 
RandomX JIT x86: remove unnecessary instructions
 2021-01-24 13:59:56 +07:00
SChernykh
787aad0d62 RandomX JIT x86: remove unnecessary instructions 
Adopted from https://github.com/tevador/RandomX/pull/201
 2021-01-23 22:28:50 +01:00
XMRig
0444103fba Fixed Cache QoS restore on exit where it not supported. 2021-01-24 02:23:27 +07:00
XMRig
61d5f1aacb Added RxMsr class. 2021-01-23 23:23:39 +07:00
XMRig
8a64a827f7 Move Ryzen related fixes to RxFix class. 2021-01-23 00:27:56 +07:00
XMRig
0273ea7140 Added new class Msr. 2021-01-22 23:50:25 +07:00
XMRig
5c7d9447ed Merge branch 'master' into feature-msr2 2021-01-22 16:55:57 +07:00
XMRig
647af97d75 #2043 Fix compile warning. 2021-01-17 17:48:35 +07:00
Richard Mitsuk Lavitt
a8cbc72fd1 fixed grammar in a couple of awkward error messages 2021-01-15 14:33:38 -06:00
SChernykh
fac3902ae1 RandomX x86 JIT: remove redundant CFROUND 2021-01-07 16:20:00 +01:00
SChernykh
90760db1c9 Fix GCC warning 2020-12-19 19:50:52 +01:00
SChernykh
7d92d57059 Another dataset AVX2 init speedup (+3.8% faster on Zen3) 2020-12-19 19:46:31 +01:00
SChernykh
0319b5b622 Added config option for AVX2 dataset init 
-1 = Auto detect
0 = Always disabled
1 = Enabled if AVX2 is supported
 2020-12-19 16:18:49 +01:00
SChernykh
eae3a62345 Auto-detect the fastest code for dataset init 2020-12-19 13:59:28 +01:00
SChernykh
aaf1ed1c99 Dataset initialization with AVX2 (WIP) 2020-12-18 14:53:54 +01:00
XMRig
f169227ebd Huge pages not supported by macOS ARM. 2020-12-16 01:59:20 +07:00