Commit Graph

281 Commits

Author SHA1 Message Date
SChernykh
e766e11dfa RandomX (Arqma variant) support 2019-10-08 19:00:19 +02:00
XMRig
19ff69eb35 Fixed crash. 2019-10-08 08:50:33 +07:00
XMRig
e115db1fea Added "memory-pool" option. 2019-10-07 23:38:01 +07:00
XMRig
d7ae457325 Added class NUMAMemoryPool. 2019-10-07 20:39:04 +07:00
XMRig
78d87138d5 Fixed Linux build. 2019-10-07 13:37:12 +07:00
XMRig
dd0f924bff Added initial memory pool support. 2019-10-07 12:36:40 +07:00
XMRig
f23d206762 VirtualMemory class refactoring. 2019-10-06 17:49:15 +07:00
XMRig
9e65fac905 Use hwloc for set thread affinity. 2019-10-06 14:40:42 +07:00
XMRig
8b8c9e50c9 Merge branch 'optimize-numa' into evo 2019-10-06 11:21:31 +07:00
XMRig
b7398ad9fa Fixed exit condition for RxQueue. 2019-10-06 11:19:32 +07:00
XMRig
3d262d82a0 Added class RxQueue, class Rx now thin static wrapper on top of RxQueue. 2019-10-06 07:47:41 +07:00
SChernykh
9314205de2 Refactored JIT compiler for x86, small RandomX speedup 2019-10-05 21:40:21 +02:00
XMRig
5058da3f29 Fixed exit. 2019-10-05 11:24:22 +07:00
XMRig
717e13818f Implemented RxNUMAStorage. 2019-10-05 08:24:28 +07:00
XMRig
851fe58e25 Simplified VirtualMemory::bindToNUMANode. 2019-10-04 19:52:15 +07:00
XMRig
93a70f4bbe Added RxNUMAStorage stub. 2019-10-04 18:43:03 +07:00
XMRig
473b98ae68 Extended "numa" option for RandomX. 2019-10-04 10:49:55 +07:00
XMRig
f9f5b8a716 Added interface IRxStorage and RxBasicStorage class. 2019-10-04 08:45:13 +07:00
XMRig
8b14b414a4 More cleanup. 2019-10-03 07:45:25 +07:00
XMRig
4a80dfa349 Added class RxSeed. 2019-10-03 04:48:36 +07:00
XMRig
c452846a6a #1212 Fixed RandomX dataset re-initialization. 2019-10-02 06:35:49 +07:00
XMRig
dea1f983f0 Merge branch 'dev' into evo 2019-10-01 07:26:18 +07:00
XMRig
aaddfbe779 Fixed support for systems where total count of NUMA nodes not equal usable count. 2019-10-01 07:19:13 +07:00
XMRig
0f444b747e Merge branch 'dev' into evo 2019-09-28 03:25:59 +07:00
XMRig
b1188210e2 Set "rx/0" as user visible RandomX name. 2019-09-28 03:25:03 +07:00
XMRig
9694164318 Merge branch 'dev' into evo 2019-09-27 05:54:33 +07:00
XMRig
335092f70f Added coin option. 2019-09-27 05:41:45 +07:00
SChernykh
1b33b073f3 Set scratchpad pointer to null by default
To avoid freeing random blocks of memory in some cases.
2019-09-24 08:53:00 +02:00
SChernykh
d346994fbd Workaround for a bug in binutils-2.32-1 on ARM
ldr/madd instruction sequence makes compiled binary crash, so separate them.
2019-09-23 23:12:40 +02:00
XMRig
7303be4c46 Revert instructions_portable.cpp to avoid warning on gcc compilers. 2019-09-22 00:59:53 +07:00
SChernykh
46ef04a4cf Added JIT compiler for RandomX on ARMv8 2019-09-21 10:10:52 +02:00
XMRig
1aee003e4a Backport fixes from v4. 2019-09-20 14:54:18 +07:00
XMRig
3d754cc369 #1180 Fixed race condition in nonce reset. 2019-09-20 00:30:20 +07:00
XMRig
26e3d541b2 Use RxAlgo:base instead of switch. 2019-09-14 12:00:33 +07:00
XMRig
a3e66e94cf Fix compile warning, mostly struct/class inconsistency. 2019-09-13 18:21:05 +07:00
XMRig
1699443520 Merge branch 'dev' into feature-opencl 2019-09-12 13:16:42 +07:00
XMRig
9428659e72 Implemented OpenCL JIT mode. 2019-09-12 13:10:50 +07:00
XMRig
5dfe4ec100 Fixed invalid shares. 2019-09-12 10:00:12 +07:00
SChernykh
5fffeed646 RandomX: optimized loading from scratchpad
Prefetches scratchpad data as soon as possible to calculate data address for the next load.

Up to ~1.4% speedup on Ryzen 7 3700X @ 4.1 GHz, RAM 3200 MHz 14-14-14-28 with optimized sub-timings:
Variant|Before H/S|After H/S
-------|----------|---------
rx/0|8663|8777
rx/wow|9867|10009
rx/loki|8652|8731
2019-09-11 19:10:01 +02:00
XMRig
e9b20fff81 Implemented VM mode for OpenCL RandomX. 2019-09-12 00:01:03 +07:00
XMRig
c4b49d4213 OpenCL RandomX WIP 2019-09-11 15:48:02 +07:00
SChernykh
0e34221026 Fix for 272 initialization threads 2019-09-09 08:15:46 +02:00
XMRig
9fb3f9e01c Initial RandomX source code prepare and compile. 2019-09-08 21:56:18 +07:00
XMRig
121bd87491 Split config generator to separated files. 2019-09-06 19:38:22 +07:00
XMRig
a74c04007f Print detailed information about OpenCL threads. 2019-09-05 13:00:40 +07:00
SChernykh
e2e4fe5f8c Optimized CFROUND
One less micro-op
2019-09-04 20:47:47 +02:00
SChernykh
fffe3c2e10 RandomX optimizations
- Optimized soft AES code, up to +30% hashrate on CPU without AES support
- Added prefetch for the first dataset access, up to +0.1% hashrate
2019-09-04 19:24:12 +02:00
XMRig
c91dcc6837 Relax cn algorithm constants definition. 2019-09-04 13:22:20 +07:00
XMRig
d948d8d06b cn/r part 2 of 2. 2019-09-04 11:23:04 +07:00
XMRig
df5fd65a45 Merge branch 'dev' into feature-opencl 2019-08-30 16:47:51 +07:00