Pages:
Author

Topic: WildRig Multi 0.35.1 beta 2 multi-algo miner for AMD & NVIDIA - page 62. (Read 91598 times)

newbie
Activity: 45
Merit: 0
your intensity maybe a bit high also you could specify your worksize (--opencl-launch=22)  0,64,128,256

I also undervolt and down clock Core speeds for ETH. But sometimes I forget and start to mine another algo. Phi|Hex|X16r etc and my undervolting causes miners to crash. -- just a FYI kind of thing Wink

majika, Kind Sir, Thank you so much...

I should have known that my undervolts were the first place to start. I literally had them so carefully tuned down for cryptonight that even 25mv lower on any one card and the system was unstable. So I should have know I would have to work them down again for a new algo.

I bumped them all back up to 1v and I can get ~10Mh/s from the Vega56's and 11 from the 64's. I am still playing around.

Things I have found so far:

- 3 threads is a no go, no matter how low the intensity is set. System crashes immediately.
- I can run intensity at one setting higher for single thread. Hash rates are similar single or dual at highes stable intensity.
- lowering the memory clock a bunch from my CN settings didn't hurt performance at all.

I'm still experimenting but I at least can hash stable now.

Thanks again
full member
Activity: 176
Merit: 100
lol.. now im seeing the kernels in the OP.
Cheer's melloyellow for heads up & Andrucrypt for the fine work Wink

Quote
"Download kernels for Hawaii(mirror)"


-- Yeah most def. shall do.. I've not mined on the 290Xs rig for quite some time. I was doing some XMR mining ages ago with them. I will bring them out of retirement to run a few tests across the algo's supported with wildRig miner & report back (asap).


Here's what I have so far with a non-reference 290x with a failing water block on it.  These were done using the parameters for rx570/rx580 so they are not optimal at all.  

bcd not done yet
c11   6.46 MH/s
geek not done yet
hex not done yet
hmq1725   2.56 MH/s
phi   5.41 MH/s
renesis not done yet
sonoa   700.18 KH/s
timetravel   12.58 MH/s
tribus   23.64 MH/s
x16r   6.22 MH/s
x16s   4.43 MH/s
x17   4.41 MH/s
jr. member
Activity: 64
Merit: 4
lol.. now im seeing the kernels in the OP.
Cheer's melloyellow for heads up & Andrucrypt for the fine work Wink

Quote
"Download kernels for Hawaii(mirror)"


-- Yeah most def. shall do.. I've not mined on the 290Xs rig for quite some time. I was doing some XMR mining ages ago with them. I will bring them out of retirement to run a few tests across the algo's supported with wildRig miner & report back (asap). 



full member
Activity: 176
Merit: 100
Does anything special need to be done in order for Hawaii cards to run besides downloading the optional bin.  I'm benchmarking WildRig Multi in RainbowMiner and can't get any shares?
It might be because --opencl-threads  --opencl-launch are not set by default in RainbowMiner.
Does anybody have any good thread, intensity, and worksize settings for a 290x?

I am just about to blow the dust off my old Sapphire R9 290X TRIXX GPU's and run Wildrig-multi miner on these algo's (Test first if its even possible to run.) if so I will try help.

I have a good feeling about the 290x's mining HEX. I think they will smash that algo. idk need to test. Wink

FYI names of my old generated kernels from back in the day using my gpus/older rigs. "Hawaiig2tc12404w256l4" & "Pitcairnglg2tc12404w256l4" FWIT Wink

Awesome.  Let me know what you works for you.  I'll add those parameters in when I get a chance to see what happens and report back.

Those generated names are likely from sgminer compiling bins for scrypt or scrypt-n or darkcoin or something along those lines.  Andrucrypt gave us the kernels in the optional download so make sure you grab that.

Your 290x's work great on cryptonight algos, by the way, if you're not using them at all.
jr. member
Activity: 64
Merit: 4
Does anything special need to be done in order for Hawaii cards to run besides downloading the optional bin.  I'm benchmarking WildRig Multi in RainbowMiner and can't get any shares?
It might be because --opencl-threads  --opencl-launch are not set by default in RainbowMiner.
Does anybody have any good thread, intensity, and worksize settings for a 290x?

I am just about to blow the dust off my old Sapphire R9 290X TRIXX GPU's and run Wildrig-multi miner on these algo's (Test first if its even possible to run.) if so I will try help.

I have a good feeling about the 290x's mining HEX. I think they will smash that algo. idk need to test. Wink

FYI names of my old generated kernels from back in the day using my gpus/older rigs. "Hawaiig2tc12404w256l4" & "Pitcairnglg2tc12404w256l4" FWIT Wink
jr. member
Activity: 64
Merit: 4
Test RIG x10 Sapphire RX580 Special Editions 1430Mhz/2250Mhz BIOS modded from 1750 timing straps

AMD TEST RIG - Platform & Drivers.
Radeon Software Version - 17.11.2
Radeon Software Edition - Crimson ReLive
Windows Version - Windows 10 (64 bit)
System Memory - 8 GB
CPU Type - Intel(R) Core(TM) i7-7700K CPU @ 4.20GHz
GPU #0-9 Radeon RX 580 Series
kernel: Ellesmere, intensity: 20, worksize: 64/256, cu: 36


 * VERSIONS:     WildRig/0.12.0 beta libuv/1.21.0 OpenCL/2.0 MSVC/2015
 * CPU:          Intel(R) Core(TM) i7-7700K CPU @ 4.20GHz x64 AES-NI
 * ALGO:         x16r + HEX
[15:29:33] donation: 2%
[15:29:33] strategy: 1
[15:29:33] send-stale: no
[15:29:33] watchdog: enabled
[15:29:33] good-risers: default

Hashrate as reported on pool. 3.0909% 110 Mh/s   <===


Algo: x16s
GPU: 10 RX580 8GB Special Editions NITRO+ @ 1430mhz/2100 @ BIOS Mod to 1750 mhz timing straps.

[15:29:33] speed 10s/60s/15m 99989 83536 74286 kH/s max: 101575 kH/s
Code:
wildrig.exe --algo x16s --opencl-platform=1 --opencl-threads 2 --opencl-launch 20x64 --strategy=1 --opencl-affinity=4 --multiple-instance  --url=stratum+tcp://quar.pickaxe.pro:3033 --user=.RX580RIG2 --pass=c=QUAR --print-full --watchdog

Code:
[15:29:33] GPU #0: 10s/60s/15m 10173 8293 7475 kH/s
[15:29:33] GPU #1: 10s/60s/15m 9806 8325 7381 kH/s
[15:29:33] GPU #2: 10s/60s/15m 10004 8324 7434 kH/s
[15:29:33] GPU #3: 10s/60s/15m 10018 8293 7405 kH/s
[15:29:33] GPU #4: 10s/60s/15m 9760 8319 7421 kH/s
[15:29:33] GPU #5: 10s/60s/15m 10018 8322 7352 kH/s
[15:29:33] GPU #6: 10s/60s/15m 10065 8436 7511 kH/s
[15:29:33] GPU #7: 10s/60s/15m 10109 8440 7518 kH/s
[15:29:33] GPU #8: 10s/60s/15m 10025 8450 7445 kH/s
[15:29:33] GPU #9: 10s/60s/15m 10006 8343 7340 kH/s
[15:29:33] speed 10s/60s/15m 99989 83536 74286 kH/s max: 101575 kH/s
[15:29:34] accepted(416/0) diff 2.01G avg. effort 102.79% (100 ms)

Notes.
I have am able to achieve much higher Hash Rates However, The increase to a higher Thread count to 3/4
Increase in Intensity (above 21,22,23) makes my rig become unstable and crashes after 20-30mins of mining.

The above is stable over 12hrs.
--opencl-threads 3 --opencl-launch 21x256 --strategy=0 --opencl-affinity=4 ~ 147MH/s



Algo: HEX

Code:
[18:40:52] GPU #0: 10s/60s/15m 8245 8310 n/a kH/s
[18:40:52] GPU #1: 10s/60s/15m 8172 8192 8112 kH/s
[18:40:52] GPU #2: 10s/60s/15m 8141 8187 8111 kH/s
[18:40:52] GPU #3: 10s/60s/15m 8153 8213 8127 kH/s
[18:40:52] GPU #4: 10s/60s/15m 8263 8344 n/a kH/s
[18:40:52] GPU #5: 10s/60s/15m 8132 8196 8115 kH/s
[18:40:52] GPU #6: 10s/60s/15m 8255 8314 8233 kH/s
[18:40:52] GPU #7: 10s/60s/15m 8274 8321 8237 kH/s
[18:40:52] GPU #8: 10s/60s/15m 8302 8332 8261 kH/s
[18:40:52] GPU #9: 10s/60s/15m 8125 8199 8122 kH/s
[18:40:52] speed 10s/60s/15m 82067 82610 73563 kH/s max: 82807 kH/s
[18:40:54] accepted(347/2) diff 67.11M avg. effort 105.29% (102 ms)

Code:
[18:40:22] speed 10s/60s/15m 82832 82634 73297 kH/s max: 82807 kH/s HEX
wildrig.exe --algo hex --opencl-platform=1 --opencl-threads 3 --opencl-launch 22x256 --opencl-affinity=4 --strategy=0 --multiple-instance  --url=stratum+tcp://xdna.pickaxe.pro:5133 --user=
.RX580RIG1 --pass=c=XDNA --print-full --watchdog

POOLSIDE STATS
Summary   Miners   Shares        Hashrate*         Reject*
hex            1           50.2399%   81.4 Mh/s         0%

Worker                Details                  Extra             Algo      Diff     Hashrate*        Reject*
RX580RIG1         WildRig/0.12.0    c=XDNA       hex      24      81.4 Mh/s        0



NB

:: speed 10s/60s/15m 120582 114144 90425 kH/s max: 147196 kH/s  <=== Wink
147MH/s is easily achievable but its hard to mine longer than 30mins stable/without crashing. (Will need to keep tweaking things) perhaps mess with powertune 0~+10%.
increase with a touch of memory 'or' core voltage (-cvddc -mvddc) to see if this will help with stability.  idk.. ??
full member
Activity: 176
Merit: 100
Does anything special need to be done in order for Hawaii cards to run besides downloading the optional bin.  I'm benchmarking WildRig Multi in RainbowMiner and can't get any shares?
It might be because --opencl-threads  --opencl-launch are not set by default in RainbowMiner.
Does anybody have any good thread, intensity, and worksize settings for a 290x?
member
Activity: 719
Merit: 49
Hi,

I tried the v12 beta version on Hive OS but although I had selected the algo name from the known algos (e.g. bcd, x17 etc.), it consistently takes the "aergo" as the default algorithm. I also tried to add "--algo bcd" switch into the extra parameters field, but this time gives incorrect parameter error. It works fine with aergo but as you may know the pool is different, so it gets rejected by the pool.

I wrote to Hive OS support but they wrote that it was due to the miner's limited support to Hive OS and being a beta version etc.

What am I doing wrong? Any suggestion would be really aprreciated.

Hive OS Custom Miner Settings:
Hive OS Miner Log:

Thx Smiley
Hi,

currently my scripts don't support "algo" parameter from HiveOS GUI(will fix this in next release), so it should work using "--algo bcd" string in extra-parameters. What error miner shows exactly? Any screen?
jr. member
Activity: 73
Merit: 2
Code:
[21:47:15] donation: 2%
[21:47:15] strategy: 0
[21:47:15] send-stale: no
[21:47:15] watchdog: disabled
[21:47:15] good-risers: default
[21:47:15] compiling code and initializing GPUs. This will take a while...
[21:47:15] GPU #0 [BusID: #4] Vega [Radeon RX Vega]
[21:47:15] kernel: gfx900, intensity: 1024, worksize: 0/256, cu: 56
[21:47:15] Error CL_BUILD_PROGRAM_FAILURE when calling clBuildProgram
[21:47:15] Failed to start OpenCL threads
AMD vega 56 with the rocm 1.9.0 driver, ubuntu 18.04
As written in OP, miner supports only amdgpu-pro drivers right now. I can build kernels for rocm, but I have no possibility to test them at all and guarantee same speed.


Still not considering rocm? I have another miners on rocm and can't really be switching to amdgpu-pro just to try this on on vega rig.
newbie
Activity: 23
Merit: 18
Hi,

I tried the v12 beta version on Hive OS but although I had selected the algo name from the known algos (e.g. bcd, x17 etc.), it consistently takes the "aergo" as the default algorithm. I also tried to add "--algo bcd" switch into the extra parameters field, but this time gives incorrect parameter error. It works fine with aergo but as you may know the pool is different, so it gets rejected by the pool.

I wrote to Hive OS support but they wrote that it was due to the miner's limited support to Hive OS and being a beta version etc.

What am I doing wrong? Any suggestion would be really aprreciated.

Hive OS Custom Miner Settings: https://ibb.co/fOB9op
Hive OS Miner Log: https://ibb.co/eV3Eop

Thx Smiley
jr. member
Activity: 64
Merit: 4
your intensity maybe a bit high also you could specify your worksize (--opencl-launch=22)  0,64,128,256

Give this a try.

> wildrig.exe --algo hex --opencl-platform=1 --opencl-threads 2 --opencl-launch 21x128 --strategy=1

Next try

> wildrig.exe --algo hex --opencl-platform=1 --opencl-threads 3 --opencl-launch 21x256 --strategy=0

.. etc..
Take speed readings of hashrate faster 300 shares or whatever. leave it run to get a good avg. (not just peek)
Rinse & repeat.

I haven't done much reading up on either HEX algo nor, Phi algo so I need to find out what exactly either of the algo's prefer.

- Are HEX | Phi memory intensive ?
- Does the Engine speeds affect the hashing of the algo's above?
- Can GPU engine speeds be reduced to save on power (as an example) Drop Core from max to 80% of max without it affecting hashrates.

These are all questions which I must learn more about. its always good to understand deeper as opposed to blindly setting mining parameters/options.

The name of the game is to mine the fastest speeds in the most efficient manner.


The only thing I can go on to help is to watch the hashrates on your miner if you see them bouncing around a lot after a couple of hundred shares have gone past lower your intensity.

Your Vega GPU's run HBM2 memory as opposed to RX470/80 570/80 GDDR5 so the settings I found work best will obviously be different to yours. However, i understand you are seeking baselines. As per your initial question.

Perhaps you can figure things out best by working backwards.

Start with lowest settings of everything
1 Thread
0 worksize
18 intensity
0 stratagy

Then ramp your settings up.
pick one setting to focus your benchmark on then increase. once you found fastest. sweetspot pick the next setting then ramp than up and so on.

Once that is done you can test with the Strategy=0|1 option.
--Independent GPU's over multi-gpu as one device.
Quote
so every GPU(or thread of it, if use --opencl-threads) work in its own range. Using strategy 1 all GPU's will work as one device

As was said previously by someone its had to actually benchmark as the algos change.
An option could be to benchmark in sgminer-kl or whatever other miner to find exact best speeds using the "--benchmark BBBBBBBB" <<= (change to match hex algo)

Find a miner where you are able to issue the --benchmark command.

Finding optimal settings in that miner and then port/copy your findings /config over into wildrig (for as close as possible match) [Just a suggestion ;]

Good Luck.

I will post my fastest settings for all the algos when I have a finished. Wink
FYI:

I also undervolt and down clock Core speeds for ETH. But sometimes I forget and start to mine another algo. Phi|Hex|X16r etc and my undervolting causes miners to crash. -- just a FYI kind of thing Wink
newbie
Activity: 45
Merit: 0
Hi,

 I am new to this software, also new to any other algo than CN. I've been mining with my 9 card RX Vega rig on XMR for a loooong time and wanted to give something new a try.

I have been playing with WildRig for a bit today trying to get it stable on my Vega rig.

I would greatly appriciate a baseline for settings to try.  Intensity? Threads? Worksize? Etc.

I was able to get it hashing for almost 10 minutes with this config from the post by majika


wildrig.exe --algo hex --opencl-platform=1 --opencl-threads 2 --opencl-launch 23x256 --strategy=0 --multiple-instance  --url= --user=.RX580RIG1 --pass=x --print-full --watchdog

Thanks,
Biker

OK, I have it running stable now for 15 minutes.
Using the following bat file
Code:
@echo off

:loop
wildrig.exe --algo hex --opencl-platform=1 --opencl-threads 2 --opencl-launch=22x0 --strategy=0 --multiple-instance --url=stratum+tcp://eu.gos.cx:5000 --user=XJ8FntKNSPnMB6k11WSHrNXG5NwrYJGLea.Biker --pass=c=XDNA
if ERRORLEVEL 1000 goto custom
timeout /t 5
goto loop

:custom
echo Custom command here
timeout /t 5
goto lo

 GPUs seems to be running at about max load, for hex it obviously fluctuates but hangs in the 90+ % range most of the time and wattage per GPU is reading ~100-120w.
Hashrate is ~10.2Mh/s per GPU.   My undervolt and clock speeds are still set where they were optimal for Cryptonight algo.


ANY further help here will be greatly appreciated  Grin
newbie
Activity: 45
Merit: 0
Hi,

 I am new to this software, also new to any other algo than CN. I've been mining with my 9 card RX Vega rig on XMR for a loooong time and wanted to give something new a try.

I have been playing with WildRig for a bit today trying to get it stable on my Vega rig.

I would greatly appriciate a baseline for settings to try.  Intensity? Threads? Worksize? Etc.

I was able to get it hashing for almost 10 minutes with this config from the post by majika


wildrig.exe --algo hex --opencl-platform=1 --opencl-threads 2 --opencl-launch 23x256 --strategy=0 --multiple-instance  --url= --user=.RX580RIG1 --pass=x --print-full --watchdog

Thanks,
Biker
member
Activity: 232
Merit: 24




Thank you Majika, I took the liberty to link your post about the HEX hashrate to our XDNA Ann thread! Smiley
https://bitcointalksearch.org/topic/m.46554102
member
Activity: 719
Merit: 49
my test with something hex
 
GPU0 ASUS ROGSTRIX RX560 4GB
GPU1 MSI ARMOR RX570 8GB
GPU2 ASUS ROGSTRIX RX570 4GB
GPU3 MSI ARMOR RX480 4GB
GPU4 MSI GAMING RX580 4GB
GPU5 MSI ARMOR RX470 4GB

wildrig.exe --algo hex --opencl-threads 3 --opencl-launch 23x256 --strategy=0 --multiple-instance --print-full --url=stratum+tcp://xdna.suprnova.cc:4919 --user=USER.WORKER --pass=x


Intensity 23 is too high for RX560 card, and that's why you get n/a on it. And worksize 256 is not optimal as I know. What about hashrate - seems you get worst hashorder. Today I will make some benchmarks with worst and best situations.
jr. member
Activity: 69
Merit: 1
my test with something hex
 
GPU0 ASUS ROGSTRIX RX560 4GB
GPU1 MSI ARMOR RX570 8GB
GPU2 ASUS ROGSTRIX RX570 4GB
GPU3 MSI ARMOR RX480 4GB
GPU4 MSI GAMING RX580 4GB
GPU5 MSI ARMOR RX470 4GB

wildrig.exe --algo hex --opencl-threads 3 --opencl-launch 23x256 --strategy=0 --multiple-instance --print-full --url=stratum+tcp://xdna.suprnova.cc:4919 --user=USER.WORKER --pass=x

jr. member
Activity: 64
Merit: 4
About 2-3 shares per GPU - that depends on difficulty set by pool, lower difficulty and higher intensity - higher chance to get more than 1 share to be solved Smiley What about x16r algo - this is hard to find faster parameters because of random hashorder right now, without bench mode.
True. It all depends on the pool.
Can the above member say which pool he is using for conducting his test using the miner on HEX?


Test pool(s)
supranova

stratum+tcp://xdna.suprnova.cc:4919     0% fee atm Wink [Primary]
stratum+tcp://rvn.suprnova.cc:6667       1.0 % fee [Primary]
stratum+tcp://eu.ravenminer.com:1111  0.9% fee [backup]
jr. member
Activity: 64
Merit: 4
yeah, as I start mining my shares fly up the mining console screen but as the diff adapts these occurrences slow down and then stops as pool diff is normalized.

i'm going through every setting no matter how silly it looks I want to be methodical as possible and go through every switch/option to find the sweet spot for each algo. tbh, whats already on the OP example is pretty much spot on. but on the HEX algo. Its the fastest miner I've used period!.

Regards the X16r I need to finish going over (repeating) tests with the --strategy=1 option enabled.

The Strategy experimental setting is an interesting concept to make the (*multi-gpu*) mining rig act as a single device. I think that's cool.
I have never came across something like this before.  Wink
member
Activity: 340
Merit: 15
About 2-3 shares per GPU - that depends on difficulty set by pool, lower difficulty and higher intensity - higher chance to get more than 1 share to be solved Smiley What about x16r algo - this is hard to find faster parameters because of random hashorder right now, without bench mode.
True. It all depends on the pool.
Can the above member say which pool he is using for conducting his test using the miner on HEX?
member
Activity: 719
Merit: 49
About 2-3 shares per GPU - that depends on difficulty set by pool, lower difficulty and higher intensity - higher chance to get more than 1 share to be solved Smiley What about x16r algo - this is hard to find faster parameters because of random hashorder right now, without bench mode.
Pages:
Jump to: