Topic: SILENTARMY v5: Zcash miner, 115 sol/s on R9 Nano, 70 sol/s on GTX 1070 - page 78. (Read 209309 times)
November 08, 2016, 06:47:36 PM
--use 1?
November 08, 2016, 06:44:39 PM
Tried without Xorg(no overclock), gpu0 slower, don't know why. Cards overclocked by memory(+1000 = +500 on win), getting 41-42, no overclock 35-38. Shitty nvidia-setting can't run without X ;(
November 08, 2016, 06:34:26 PM
with 3 instances getting 240-250 s/s on 6 x 1070 (eXtremal's input.cl and my cpu load workaround)
Could you try with only one gpu? And 1,2,3,4 instances?
hmm.. I really don't know why my gpu0 card runs slower o_O
Code:
$ LD_PRELOAD="./libtime.so" ./silentarmy --instances=3 --use=0
Connecting to us1-zcash.flypool.org:3333
Stratum server sent us the first job
Mining on 1 device
Total 0.0 sol/s [dev0 0.0] 0 shares
Total 39.6 sol/s [dev0 39.6] 0 shares
Total 38.2 sol/s [dev0 38.2] 0 shares
Total 32.4 sol/s [dev0 32.4] 0 shares
Total 33.2 sol/s [dev0 33.2] 0 shares
Total 37.1 sol/s [dev0 37.1] 0 shares
Total 37.0 sol/s [dev0 37.0] 0 shares
Total 34.7 sol/s [dev0 34.7] 0 shares
Total 34.6 sol/s [dev0 34.6] 0 shares
Total 35.6 sol/s [dev0 35.6] 0 shares
but gpu1 and others
Code:
LD_PRELOAD="./libtime.so" ./silentarmy --instances=3 --use=1
Connecting to us1-zcash.flypool.org:3333
Stratum server sent us the first job
Mining on 1 device
Total 0.0 sol/s [dev1 0.0] 0 shares
Total 45.6 sol/s [dev1 45.6] 0 shares
Total 53.0 sol/s [dev1 53.0] 0 shares
Total 47.9 sol/s [dev1 47.9] 0 shares
Total 45.8 sol/s [dev1 45.8] 0 shares
Total 42.6 sol/s [dev1 42.6] 0 shares
Total 43.5 sol/s [dev1 43.5] 0 shares
Total 42.5 sol/s [dev1 42.5] 0 shares
Total 42.6 sol/s [dev1 42.6] 0 shares
Total 43.3 sol/s [dev1 43.3] 0 shares
Maybe because Xorg is running on it ? On windows 10 with Claymore's miner and rx470 cards, I have same issue - card, connected to monitor, runs slower
Code:
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 367.27 Driver Version: 367.27 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 1070 On | 0000:01:00.0 On | N/A |
|100% 32C P0 40W / 195W | 15MiB / 8113MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 1 GeForce GTX 1070 On | 0000:03:00.0 Off | N/A |
|100% 32C P0 42W / 195W | 6MiB / 8113MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 2 GeForce GTX 1070 On | 0000:04:00.0 Off | N/A |
|100% 30C P0 41W / 195W | 6MiB / 8113MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 3 GeForce GTX 1070 On | 0000:05:00.0 Off | N/A |
|100% 31C P0 41W / 195W | 6MiB / 8113MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 4 GeForce GTX 1070 On | 0000:06:00.0 Off | N/A |
|100% 28C P0 41W / 195W | 6MiB / 8113MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 5 GeForce GTX 1070 On | 0000:07:00.0 Off | N/A |
|100% 28C P0 41W / 195W | 6MiB / 8113MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 1751 G /usr/lib/xorg/Xorg 13MiB |
| 1 1751 G /usr/lib/xorg/Xorg 6MiB |
| 2 1751 G /usr/lib/xorg/Xorg 6MiB |
| 3 1751 G /usr/lib/xorg/Xorg 6MiB |
| 4 1751 G /usr/lib/xorg/Xorg 6MiB |
| 5 1751 G /usr/lib/xorg/Xorg 6MiB |
+-----------------------------------------------------------------------------+
November 08, 2016, 06:26:53 PM
with 3 instances getting 240-250 s/s on 6 x 1070 (eXtremal's input.cl and my cpu load workaround)
Could you try with only one gpu? And 1,2,3,4 instances?
November 08, 2016, 05:57:06 PM
with 3 instances getting 240-250 s/s on 6 x 1070 (eXtremal's input.cl and my cpu load workaround)
Code:
Total 243.2 sol/s [dev0 32.1, dev1 40.7, dev2 42.4, dev3 41.3, dev4 47.3, dev5 43.1] 20 shares
Total 244.0 sol/s [dev0 33.9, dev1 42.6, dev2 43.1, dev3 37.9, dev4 44.9, dev5 42.5] 21 shares
Total 247.0 sol/s [dev0 34.2, dev1 42.7, dev2 43.1, dev3 37.2, dev4 44.9, dev5 43.5] 21 shares
Total 245.4 sol/s [dev0 35.1, dev1 43.3, dev2 41.0, dev3 37.3, dev4 43.7, dev5 43.7] 21 shares
Total 244.8 sol/s [dev0 34.4, dev1 44.6, dev2 39.9, dev3 37.6, dev4 43.5, dev5 44.5] 21 shares
Total 243.8 sol/s [dev0 34.0, dev1 43.7, dev2 38.0, dev3 36.6, dev4 42.5, dev5 43.0] 21 shares
Total 244.1 sol/s [dev0 35.1, dev1 45.1, dev2 38.9, dev3 37.2, dev4 42.9, dev5 44.0] 21 shares
Total 244.5 sol/s [dev0 36.9, dev1 44.3, dev2 41.0, dev3 38.7, dev4 41.6, dev5 46.1] 21 shares
Total 244.8 sol/s [dev0 37.2, dev1 43.9, dev2 39.9, dev3 38.0, dev4 42.8, dev5 43.8] 22 shares
Total 243.4 sol/s [dev0 36.0, dev1 43.8, dev2 39.1, dev3 37.3, dev4 42.6, dev5 42.3] 23 shares
Total 243.5 sol/s [dev0 35.8, dev1 43.9, dev2 38.0, dev3 38.2, dev4 41.7, dev5 42.5] 24 shares
Total 243.4 sol/s [dev0 34.4, dev1 41.8, dev2 38.5, dev3 41.0, dev4 42.1, dev5 41.4] 26 shares
Total 243.0 sol/s [dev0 34.5, dev1 42.3, dev2 38.9, dev3 42.0, dev4 41.0, dev5 40.5] 26 shares
Total 243.0 sol/s [dev0 33.9, dev1 42.2, dev2 39.4, dev3 43.1, dev4 40.2, dev5 40.7] 26 shares
Total 243.6 sol/s [dev0 34.5, dev1 42.7, dev2 41.8, dev3 43.6, dev4 40.7, dev5 40.8] 26 shares
Total 243.5 sol/s [dev0 35.1, dev1 45.2, dev2 41.7, dev3 44.7, dev4 42.0, dev5 41.7] 28 shares
Total 244.4 sol/s [dev0 34.7, dev1 45.2, dev2 41.8, dev3 44.5, dev4 42.0, dev5 39.8] 30 shares
Total 243.0 sol/s [dev0 33.9, dev1 44.5, dev2 41.9, dev3 45.2, dev4 40.9, dev5 38.1] 30 shares
Total 244.3 sol/s [dev0 34.5, dev1 43.4, dev2 42.5, dev3 45.9, dev4 40.0, dev5 40.6] 30 shares
Total 245.6 sol/s [dev0 37.2, dev1 46.0, dev2 43.7, dev3 46.0, dev4 38.6, dev5 41.8] 30 shares
Total 247.3 sol/s [dev0 36.9, dev1 48.5, dev2 44.1, dev3 45.5, dev4 40.0, dev5 39.8] 31 shares
Total 245.2 sol/s [dev0 36.2, dev1 47.4, dev2 42.4, dev3 43.6, dev4 40.4, dev5 41.6] 31 shares
Total 245.3 sol/s [dev0 36.4, dev1 45.6, dev2 43.2, dev3 44.3, dev4 40.0, dev5 41.5] 33 shares
Total 244.9 sol/s [dev0 37.7, dev1 45.8, dev2 42.7, dev3 41.7, dev4 40.5, dev5 42.6] 35 shares
Total 246.4 sol/s [dev0 37.4, dev1 46.1, dev2 42.1, dev3 40.8, dev4 41.4, dev5 43.1] 37 shares
November 08, 2016, 05:20:53 PM
Any news for windows release supporting nvidia? 
November 08, 2016, 05:17:38 PM
Finally fixed cpu load.
libtime.c:
6 x 1070
CPU load on celeron 1840
You can adjust sleep time by changing the useconds_t sleep_time value
libtime.c:
Code:
#include
#include
#include
#include
#include
#include
/*
Temprorary fix for silentarmy - nvidia
The MIT License (MIT) Copyright (c) 2016 krnlx, kernelx at me.com
*/
int inited=0;
void *libc = NULL;
int (*libc_clock_gettime)(clockid_t clk_id, struct timespec *tp) = NULL;
static void __attribute__ ((constructor)) lib_init(void) {
if(inited)
return;
libc = dlopen("libc.so.6", RTLD_LAZY);
assert(libc);
libc_clock_gettime = dlsym(libc, "clock_gettime");
assert(libc_clock_gettime);
inited++;
}
useconds_t sleep_time = 1200;
int clock_gettime(clockid_t clk_id, struct timespec *tp){
lib_init();
//printf(".");
usleep(sleep_time);
int r = (*libc_clock_gettime)(clk_id, tp);
return r;
}
Code:
gcc -O2 -fPIC -shared -Wl,-soname,libtime.so -o libtime.so libtime.c
Code:
LD_PRELOAD="./libtime.so" ./silentarmy --instances=2 --use=0,1,2,3,4,5
Connecting to us1-zcash.flypool.org:3333
Stratum server sent us the first job
Mining on 6 devices
Total 0.0 sol/s [dev0 0.0, dev1 0.0, dev2 0.0, dev3 0.0, dev4 0.0, dev5 0.0] 0 shares
Total 243.5 sol/s [dev0 36.8, dev1 39.7, dev2 38.8, dev3 43.7, dev4 50.7, dev5 33.8] 1 share
Total 249.5 sol/s [dev0 39.6, dev1 38.1, dev2 39.6, dev3 40.6, dev4 51.5, dev5 40.1] 1 share
Total 254.8 sol/s [dev0 35.7, dev1 41.6, dev2 43.3, dev3 41.0, dev4 50.2, dev5 43.0] 1 share
Total 253.3 sol/s [dev0 31.2, dev1 41.1, dev2 44.6, dev3 42.6, dev4 51.0, dev5 42.8] 1 share
Total 247.6 sol/s [dev0 31.9, dev1 38.5, dev2 44.8, dev3 40.8, dev4 51.5, dev5 40.0] 1 share
Total 239.8 sol/s [dev0 31.2, dev1 38.0, dev2 43.6, dev3 39.3, dev4 47.6, dev5 40.0] 2 shares
Total 244.4 sol/s [dev0 33.0, dev1 39.7, dev2 42.4, dev3 40.4, dev4 46.9, dev5 42.1] 2 shares
Total 242.6 sol/s [dev0 32.2, dev1 40.1, dev2 42.9, dev3 38.1, dev4 47.1, dev5 42.2] 3 shares
Total 241.7 sol/s [dev0 32.2, dev1 39.6, dev2 42.8, dev3 39.7, dev4 46.6, dev5 40.9] 5 shares
Total 238.2 sol/s [dev0 32.9, dev1 39.6, dev2 42.8, dev3 37.7, dev4 45.8, dev5 39.4] 5 shares
Total 238.4 sol/s [dev0 32.8, dev1 39.7, dev2 42.1, dev3 38.2, dev4 45.0, dev5 40.1] 6 shares
Total 238.1 sol/s [dev0 29.4, dev1 40.2, dev2 43.9, dev3 39.2, dev4 44.1, dev5 39.0] 7 shares
Total 235.0 sol/s [dev0 29.0, dev1 38.0, dev2 44.1, dev3 38.3, dev4 41.7, dev5 38.0] 8 shares
Total 236.0 sol/s [dev0 32.1, dev1 38.1, dev2 43.4, dev3 37.7, dev4 40.6, dev5 37.2] 8 shares
Total 238.3 sol/s [dev0 31.5, dev1 40.5, dev2 44.4, dev3 39.0, dev4 40.4, dev5 37.9] 8 shares
Total 238.5 sol/s [dev0 33.1, dev1 41.8, dev2 43.1, dev3 38.8, dev4 42.5, dev5 38.5] 8 shares
Total 239.7 sol/s [dev0 33.6, dev1 40.5, dev2 44.0, dev3 37.8, dev4 42.5, dev5 38.1] 8 shares
Total 240.3 sol/s [dev0 33.7, dev1 40.9, dev2 45.6, dev3 39.3, dev4 42.3, dev5 36.6] 8 shares
Total 238.7 sol/s [dev0 32.9, dev1 41.7, dev2 44.7, dev3 38.0, dev4 41.0, dev5 37.8] 8 shares
Total 238.3 sol/s [dev0 32.5, dev1 41.7, dev2 44.8, dev3 39.6, dev4 40.1, dev5 39.7] 8 shares
Total 237.5 sol/s [dev0 32.5, dev1 41.0, dev2 45.2, dev3 39.1, dev4 38.9, dev5 39.8] 8 shares
Total 239.0 sol/s [dev0 34.9, dev1 41.5, dev2 43.7, dev3 39.2, dev4 39.7, dev5 41.1] 11 shares
Total 238.7 sol/s [dev0 35.5, dev1 43.1, dev2 42.5, dev3 39.1, dev4 41.7, dev5 41.5] 11 shares
Total 237.0 sol/s [dev0 32.6, dev1 44.1, dev2 41.8, dev3 38.3, dev4 40.0, dev5 41.5] 11 shares
Total 237.2 sol/s [dev0 32.6, dev1 43.7, dev2 39.9, dev3 38.6, dev4 39.9, dev5 40.7] 12 shares
Total 239.3 sol/s [dev0 33.0, dev1 43.3, dev2 41.6, dev3 40.2, dev4 41.4, dev5 41.0] 13 shares
6 x 1070
CPU load on celeron 1840
Code:
top - 01:20:37 up 18:38, 3 users, load average: 3.66, 1.75, 0.91
Tasks: 151 total, 10 running, 141 sleeping, 0 stopped, 0 zombie
%Cpu0 : 62.5 us, 1.0 sy, 0.0 ni, 36.5 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
%Cpu1 : 59.2 us, 4.4 sy, 0.0 ni, 36.4 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
KiB Mem : 3988068 total, 1317788 free, 1465128 used, 1205152 buff/cache
KiB Swap: 0 total, 0 free, 0 used. 2204660 avail Mem
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
15266 krnl 20 0 29.619g 126804 94508 R 10.9 3.2 0:09.40 sa-solver
15273 krnl 20 0 29.614g 149356 111812 R 10.9 3.7 0:10.22 sa-solver
15269 krnl 20 0 29.599g 109488 90176 R 10.6 2.7 0:09.09 sa-solver
15271 krnl 20 0 29.599g 112464 90116 R 10.6 2.8 0:09.23 sa-solver
15275 krnl 20 0 29.610g 116104 94132 S 10.6 2.9 0:09.25 sa-solver
15268 krnl 20 0 29.599g 111812 90404 R 10.3 2.8 0:09.27 sa-solver
15265 krnl 20 0 29.609g 116228 94232 R 9.9 2.9 0:09.28 sa-solver
15272 krnl 20 0 29.610g 113848 94216 R 9.9 2.9 0:09.37 sa-solver
15263 krnl 20 0 29.599g 110692 90236 R 9.6 2.8 0:08.42 sa-solver
15264 krnl 20 0 29.599g 110464 90004 S 9.6 2.8 0:08.18 sa-solver
15274 krnl 20 0 29.599g 109500 90116 S 9.6 2.7 0:09.14 sa-solver
15267 krnl 20 0 29.599g 111504 90152 R 9.3 2.8 0:09.37 sa-solver
You can adjust sleep time by changing the useconds_t sleep_time value
November 08, 2016, 04:52:37 PM
But also +4% is something it just shows that optimizations can still be in cooperated
Absolutely. I even care about +1%, so I'll incorporate the changes!
November 08, 2016, 04:44:25 PM
But also +4% is something it just shows that optimizations can still be in cooperated
Need tests on R9 2xx/3xx cards with new drivers. May be this patch takes effect only on NVidia and old flgrx drivers. November 08, 2016, 03:33:19 PM
I have a few rigs with 8 x R7 265. Starts mining successfully, but in mining screen I see devX up to 6, meaning giving display of hashrate up to the 6-th card. Possible to make it 8 at least?
November 08, 2016, 03:32:56 PM
But also +4% is something it just shows that optimizations can still be in cooperated
November 08, 2016, 03:23:18 PM
My latest kernel results (http://coinsforall.io/distr/input.cl), first row - original SA kernel, second - patched.
Quote
Ubuntu 13.10, Catalyst 14.4, Radeon R9 290 900/1250 (downclocked)
Total 29.1 sol/s [dev0 30.2] 4 shares
Total 41.1 sol/s [dev0 42.0] 2 shares +40%
Ubuntu 16.04, NVidia 367, GeForce GTX1070
Total 196 solutions in 6588.2 ms (29.8 Sol/s)
Total 196 solutions in 5334.1 ms (36.7 Sol/s) +20%
Ubuntu 16.04, amdgpu-pro 16.30, Radeon RX480
Total 50.4 sol/s [dev0 51.0] 4 shares
Total 53.1 sol/s [dev0 53.2] 14 shares +4%
Total 29.1 sol/s [dev0 30.2] 4 shares
Total 41.1 sol/s [dev0 42.0] 2 shares +40%
Ubuntu 16.04, NVidia 367, GeForce GTX1070
Total 196 solutions in 6588.2 ms (29.8 Sol/s)
Total 196 solutions in 5334.1 ms (36.7 Sol/s) +20%
Ubuntu 16.04, amdgpu-pro 16.30, Radeon RX480
Total 50.4 sol/s [dev0 51.0] 4 shares
Total 53.1 sol/s [dev0 53.2] 14 shares +4%
FWIW, exact same silentarmy code running on the same machine on an R9 Nano, dual booting into 2 OSes:
* 33.2 sol/s with fglrx 2:15.201-0ubuntu0.14.04.1 on Ubuntu 14.04
* 47.4 sol/s with amdgpu-pro 16.40 on Ubuntu 16.04
So yeah, a +40% difference just by changing drivers... No wonder you found a way to rework the OpenCL code to get a +40% on fglrx, and it gets you only +4% on amdgpu-pro.
amdgpu-pro compiles the OpenCL code just really, really well on its own, with almost no needs for manual tweaks.
November 08, 2016, 03:18:17 PM
Is there a Windows release or not?!
Genoil's link gives a source only ?
Genoil's link gives a source only ?
It is a Python script + exe.
1. Download as ZIP on https://github.com/Genoil/silentarmy/tree/windows and unzip
2. Install Python 3.5+ for Win x64: https://www.python.org/downloads/windows/ (allow installer add python to PATH)
3. Launch .bat file (don't edit
)Something strange happens with the last build ..the build before was working fine.
November 08, 2016, 02:52:09 PM
v4 tested and running in ubuntu 16.04 with amd-gpu-pro 16.40! ... seems a bit more solutions per thread!
Good work @mrb!
Good work @mrb!
November 08, 2016, 02:50:05 PM
My latest kernel results (http://coinsforall.io/distr/input.cl), first row - original SA kernel, second - patched.
Quote
Ubuntu 13.10, Catalyst 14.4, Radeon R9 290 900/1250 (downclocked)
Total 29.1 sol/s [dev0 30.2] 4 shares
Total 41.1 sol/s [dev0 42.0] 2 shares +40%
Ubuntu 16.04, NVidia 367, GeForce GTX1070
Total 196 solutions in 6588.2 ms (29.8 Sol/s)
Total 196 solutions in 5334.1 ms (36.7 Sol/s) +20%
Ubuntu 16.04, amdgpu-pro 16.30, Radeon RX480
Total 50.4 sol/s [dev0 51.0] 4 shares
Total 53.1 sol/s [dev0 53.2] 14 shares +4%
Total 29.1 sol/s [dev0 30.2] 4 shares
Total 41.1 sol/s [dev0 42.0] 2 shares +40%
Ubuntu 16.04, NVidia 367, GeForce GTX1070
Total 196 solutions in 6588.2 ms (29.8 Sol/s)
Total 196 solutions in 5334.1 ms (36.7 Sol/s) +20%
Ubuntu 16.04, amdgpu-pro 16.30, Radeon RX480
Total 50.4 sol/s [dev0 51.0] 4 shares
Total 53.1 sol/s [dev0 53.2] 14 shares +4%
November 08, 2016, 02:40:27 PM
Just tested out v4 on Ubuntu/fglrx and a couple R9 380 cards. No material change.
Reducing MAX_SOLS from 2000 to 500 reduces the amount of data xfer over the PCI-E bus, and gives me about a 3-5% speed improvement running sa-solver.
In my fork I have OPTIM_SIMPLIFY_ROUND turned on, and the MAX_SOLS reduced.
https://github.com/nerdralph/sa-nr
Reducing MAX_SOLS from 2000 to 500 reduces the amount of data xfer over the PCI-E bus, and gives me about a 3-5% speed improvement running sa-solver.
In my fork I have OPTIM_SIMPLIFY_ROUND turned on, and the MAX_SOLS reduced.
https://github.com/nerdralph/sa-nr
Watch out you are missing solutions if you reduce MAX_SOLS. It might be ok if it correspondingly reduces the Equihash runtime, but you have to look at how many solutions you lose on a good sample >1k, such as "./sa-solver --nonces 1000"
that good or bad, as I have just updated this miner.
seems like its good.
November 08, 2016, 02:28:51 PM
Just tested out v4 on Ubuntu/fglrx and a couple R9 380 cards. No material change.
Reducing MAX_SOLS from 2000 to 500 reduces the amount of data xfer over the PCI-E bus, and gives me about a 3-5% speed improvement running sa-solver.
In my fork I have OPTIM_SIMPLIFY_ROUND turned on, and the MAX_SOLS reduced.
https://github.com/nerdralph/sa-nr
Reducing MAX_SOLS from 2000 to 500 reduces the amount of data xfer over the PCI-E bus, and gives me about a 3-5% speed improvement running sa-solver.
In my fork I have OPTIM_SIMPLIFY_ROUND turned on, and the MAX_SOLS reduced.
https://github.com/nerdralph/sa-nr
Watch out you are missing solutions if you reduce MAX_SOLS. It might be ok if it correspondingly reduces the Equihash runtime, but you have to look at how many solutions you lose on a good sample >1k, such as "./sa-solver --nonces 1000"
November 08, 2016, 02:20:43 PM
Just tested out v4 on Ubuntu/fglrx and a couple R9 380 cards. No material change.
Reducing MAX_SOLS from 2000 to 500 reduces the amount of data xfer over the PCI-E bus, and gives me about a 3-5% speed improvement running sa-solver.
In my fork I have OPTIM_SIMPLIFY_ROUND turned on, and the MAX_SOLS reduced.
https://github.com/nerdralph/sa-nr
Reducing MAX_SOLS from 2000 to 500 reduces the amount of data xfer over the PCI-E bus, and gives me about a 3-5% speed improvement running sa-solver.
In my fork I have OPTIM_SIMPLIFY_ROUND turned on, and the MAX_SOLS reduced.
https://github.com/nerdralph/sa-nr
November 08, 2016, 01:58:18 PM
Can u add values to your considerations ?
It would be even better
Thank you
Sure, here you go:
A. 1 x R9 380
Ubuntu 14.04.4 desktop, fglrx
Was 18-21 sol/s, now is 25-26 sol/s
I was getting over 30 on v3 using tighter memory timings and a 980/1550 clock.
http://nerdralph.blogspot.ca/2016/09/advanced-tonga-bios-editing.html
Still have to merge/build v4 to test it.
November 08, 2016, 12:46:25 PM
C. 5 x RX470 + 1 x RX480
Ubuntu 16.04.1 server, amdgpu-pro
Was 135-147 sol/s, now is 145-148 sol/s
Ubuntu 16.04.1 server, amdgpu-pro
Was 135-147 sol/s, now is 145-148 sol/s
Just wanted to ask about CPU model and usage. It seems that the performance highly depends on the CPU.
Running 470x2 on old single core gave me 7 per thread on 470.
Slightly better mobo + CPU gave 14/thread (28 per 470) - almost your result, see below.
Claymore's gives 45 Sol/s per 470 on Windows - on the 1st single core machine with 1GB of RAM (45 vs 14 Sols/s of this miner on the same mobo under Linux).
The results with single card and few cards on the same machine are also quite different. I use G3240 dual core CPUs on my rigs and it is not enough to run this at full speed.
So as for me, moving some filtering from CPU/bus into cards should be considered as a priority task if we want to outperform Claymore's miner. I've read that it is ALREADY faster. But trying on my cards (7950, 7970, 390, 390X, 470) - all with G3240 CPUs - I was not able to get the same results. Was close to with a single RX470 (Elpida RAM, 1500 straps, 2000 mem/1260 GPU - 46-47 Sol/s, Ubuntu 16.04 + amdgpu-pro 16.40) but without latest eXtremal patch.
So I do not understand how do you get a half of that only (148/6=25 Sol/s) on you rigs, but almost sure that the CPU usage is the reason. Have no other ideas.
Jump to: