SChernykh
c8c0abdb00
Separate MSR mod for Zen/Zen2 and Zen3
...
Another +0.5% speedup for Zen2
2020-11-08 19:40:44 +01:00
XMRig
99e9073993
Fixed build without HTTP.
2020-11-03 00:21:22 +07:00
XMRig
51690ebad6
#1918 Fixed check for 1GB huge pages on ARM Linux.
2020-11-02 21:26:35 +07:00
XMRig
3910cf9e69
Added vendor to ARM CPUs name and added "arch" field.
2020-11-01 12:06:35 +07:00
XMRig
5f314edb2f
Print benchmark url
2020-10-30 23:24:01 +07:00
XMRig
4914fefb1f
Added "msr" field for CPU backend.
2020-10-25 16:36:37 +07:00
XMRig
03cd56ed73
Implemented PATCH requests.
2020-10-25 13:41:39 +07:00
XMRig
79c96418c7
Implemented BenchClient.
2020-10-24 19:43:23 +07:00
XMRig
36c1cb23e0
Implemented static benchmark verification (--bench --seed --hash)
2020-10-24 13:53:49 +07:00
XMRig
6860450147
Removed legacy CUDA plugin API.
2020-10-20 18:04:56 +07:00
XMRig
328f985e07
Added Benchmark class.
2020-10-17 04:03:42 +07:00
XMRig
7fc7b976bf
Removed m_bench field from Job class.
2020-10-16 20:29:11 +07:00
XMRig
36b1523194
Code cleanup.
2020-10-16 19:35:36 +07:00
XMRig
5155139e9a
Improve some benchmark log messages.
2020-10-16 04:11:50 +07:00
XMRig
a152d6be42
Added CMake option WITH_BENCHMARK.
2020-10-16 02:18:07 +07:00
cohcho
5b4648339a
WorkerJob: fix calls of add/nextRound
...
WorkerJob::nextRound() doesn't require roundSize to be power of 2
Use CudaWorker::intensity(), OclWorker::intensity() to get device batch size
Sync nonce and device iteration
2020-10-15 14:56:37 +00:00
SChernykh
8c45e3226d
Fixed roundSize() usage in OCL/CUDA
2020-10-15 14:39:54 +02:00
SChernykh
c4db1435b2
Make round size power of 2
2020-10-15 14:29:15 +02:00
SChernykh
722e468bd9
Benchmark: added more check hashes and a progress indicator
2020-10-15 08:23:47 +02:00
SChernykh
144f9c4409
Fixed compile errors in Linux
2020-10-14 21:03:21 +02:00
SChernykh
2ecece7b3d
Added benchmark and stress test
...
Easy to use and zero configuration embedded benchmark/stress test.
2020-10-14 19:45:05 +02:00
xmrig
581d004568
Merge pull request #1890 from SChernykh/dev
...
Added argon2/chukwav2 algorithm
2020-10-13 14:54:34 +07:00
SChernykh
4f7186cb0e
Added argon2/chukwav2 algorithm
...
New Turtlecoin algorithm. Source: https://github.com/turtlecoin/turtlecoin/blob/development/src/crypto/hash.h#L57
2020-10-12 08:26:57 +02:00
xmrig
f85efd163c
Merge pull request #1887 from SChernykh/dev
...
Fixed total hashrate update
2020-10-10 22:07:37 +07:00
SChernykh
793a2454ad
Fixed total hashrate update
...
Don't add data points where one of the threads doesn't have hashrate data yet.
2020-10-10 17:00:30 +02:00
cohcho
4a74ce3242
CPU: use raw counter
2020-10-10 13:28:14 +00:00
SChernykh
22a69f70da
Fix HashrateInterpolator::addDataPoint
2020-10-10 11:22:19 +02:00
SChernykh
3fbf2ac3d4
More precise hashrate calculation
...
- Use only steady timestamp counters to guarantee correctness
- CPU backend: directly measure total hashrate using raw hash counters from each thread; update data more often on ARM CPUs because they're slower
- GPU backends: directly measure total hashrate too, but use interpolator with 4 second lag to fix variance from batches of hashes
Total hashrate is now measured directly (realtime for CPU, 4 seconds lag for GPU), so it might differ a bit from the sum of all thread hashrates because data points are taken at different moments in time.
Overhead is reduced a lot since it doesn't have to go through all threads to calculate max total hashrate on every timer tick (2 times a second).
2020-10-10 11:18:01 +02:00
cohcho
17795e3d7b
Worker: specify default value
2020-10-09 14:28:36 +00:00
SChernykh
858463ceba
Change to fetch_xor to make code simpler
2020-10-09 11:50:11 +02:00
SChernykh
a4550f55ea
Fix possible race condition in hashrate counting code
...
Use single atomic operation to switch between data points.
2020-10-09 10:29:18 +02:00
xmrig
1289942567
Merge pull request #1876 from SChernykh/dev
...
RandomX: added `huge-pages-jit` config parameter
2020-10-07 22:48:57 +07:00
SChernykh
44dcded866
RandomX: added huge-pages-jit config parameter
...
Set to false by default, gives 0.2% boost on Ryzen 7 3700X with 16 threads, but hashrate might be unstable on Ryzen between launches. Use with caution.
2020-10-07 17:42:55 +02:00
cohcho
a705ab775b
RandomX: align args
...
tempHash/output must be 16-byte aligned for randomx_calculate_hash{,_first,_next}
2020-10-07 14:47:18 +00:00
SChernykh
ebf259fa7c
RandomX: removed rx/loki
...
Loki forks to PoS on October 9th.
2020-10-02 17:02:52 +02:00
SChernykh
bef9031b03
KawPow: fixed OpenCL memory leak
2020-09-25 10:53:24 +02:00
SChernykh
768a4581e0
Fixed Cryptonight OpenCL for AMD 20.7.2 drivers
...
Vega 64 + Windows 10 + AMD 20.7.2 drivers were broken on Cryptonight algorithms.
2020-09-19 23:12:05 +02:00
SChernykh
c7476e076b
RandomX refactoring, moved more stuff to compile time
...
Small x86 JIT compiler speedup.
2020-09-18 20:51:25 +02:00
cohcho
30be1cd102
reserve at most 1 bit for wrapping detection
2020-09-13 18:42:16 +00:00
xmrig
adf833b60a
Merge pull request #1827 from cohcho/nonce_iteration_without_tests
...
nonce iteration optimization
2020-09-10 19:33:23 +07:00
SChernykh
4a9db89527
RandomX: added SSE4.1-optimized Blake2b
...
+0.15% on `rx/0`
+0.3% on `rx/wow`
2020-09-10 14:28:40 +02:00
cohcho
060c1af4c4
fix nonce mask
2020-09-09 19:39:52 +00:00
cohcho
b826985d05
nonce iteration optimization
...
efficient and correct nonce iteration without duplicates
2020-09-09 10:03:37 +00:00
XMRig
ff13675d31
Improved CUDA loader error reporting and fixed plugin load on Linux.
2020-08-23 21:30:12 +07:00
XMRig
4b682b6633
Better scripts/build.*.sh compatibility.
2020-08-23 15:46:56 +07:00
XMRig
a354e9d217
Fixed tag in OclLib.
2020-08-21 08:21:51 +07:00
XMRig
950b5fa75e
Disable GPU backends with static build on Linux.
2020-08-20 15:02:31 +07:00
XMRig
206b675892
Always use all available threads on ARM.
2020-08-16 17:36:38 +07:00
XMRig
cac48cdd27
Added ARM CPU name detection based on lscpu code.
2020-08-16 15:47:29 +07:00
SChernykh
abb78302b8
Try to allocate scratchpad from dataset's 1 GB huge pages, if normal huge pages are not available
2020-07-31 13:37:22 +02:00
