SRBMiner Cryptonight AMD GPU Miner V1.9.3 - native algo switching

[NCarter84]

Still looking into why my 8 GB card running slow than my 4 GB cards on CN-Lite V7.

4 GB runs around 2 khs
8 GB runs about 1.7 khs

I"ve tried alot of different combos with threads, worksizes and intensity... I went all the way up to 200 intensity with dual threads work size 8... Runs slower... I tried worksize 16, slow.... Seems like I top out at intensity of 130/8/2 at 1700.

[doktor83]

Hi
Problem upgrading from 1.6.9 to 1.7.2 : impossible to get v8 working !
No problem under heavy or haven.
But v8 give me no share and lots of memory errors after few minutes.

Any idea ?

EDIT 3 : more tests

5 RX 570 4G
core 1100,900mV
mem 1980,900mV
Drivers 18.9.2


What's wrong ?

Maybe the 18.9.2 is the problem.
Have you tried with some of the recommended drivers?

[ipe4enko]

Hi, doktor83

I have Rx550 card gigabyte and Asus. On 1.6.8 version of miner i'm using "fragments" : 338033747 and about 520 H/s on v7 or stellite.
But on version 1.7.2 if i set "fragments" : 33803374 then miner set 1. And perfomance about 500 H/s.
If "fragments" : 9997 then speed 480 H/s. If "fragments" : 9998 then speed 500 H/s.

Is it possible to return old behaviour, and i can set "fragments" : 33803374 ?

[doktor83]

Hi, doktor83

I have Rx550 card gigabyte and Asus. On 1.6.8 version of miner i'm using "fragments" : 338033747 and about 520 H/s on v7 or stellite.
But on version 1.7.2 if i set "fragments" : 33803374 then miner set 1. And perfomance about 500 H/s.
If "fragments" : 9997 then speed 480 H/s. If "fragments" : 9998 then speed 500 H/s.

Is it possible to return old behaviour, and i can set "fragments" : 33803374 ?

no but you can set fragments to 9999 and it will use the 'old fashioned' way

[ipe4enko]

Hi, doktor83

I have Rx550 card gigabyte and Asus. On 1.6.8 version of miner i'm using "fragments" : 338033747 and about 520 H/s on v7 or stellite.
But on version 1.7.2 if i set "fragments" : 33803374 then miner set 1. And perfomance about 500 H/s.
If "fragments" : 9997 then speed 480 H/s. If "fragments" : 9998 then speed 500 H/s.

Is it possible to return old behaviour, and i can set "fragments" : 33803374 ?

no but you can set fragments to 9999 and it will use the 'old fashioned' way

With fragments =9999 speed about 500 h/s 

[livada]

Hi again doktor,

I use srbminer from months with no problem in my 9 vega 64 rig.

I use v1.6x - v1.7.0 weeks with no restart or hang.

But version 1.7.2, after 8 hours or so hangs.

In log i got this:

https://justpaste.it/4rctv

Any ideia what could be?

Regards

Hi Mate!

Please try my Vega64 settings: Algo: BittubeV2 (Bittube Coin)

Settings: { "id" : 0, "intensity" : 61, "worksize" : 9, "threads" : 2},

[2018-12-04 22:01:02] GPU0 :    2058 H/s [T: 31c, RPM: 2674, CC: 1408 MHz, MC: 1100 MHz][BUS:28]
[2018-12-04 22:01:02] GPU1 :    2062 H/s [T: 30c, RPM: 2507, CC: 1408 MHz, MC: 1100 MHz][BUS:19]
[2018-12-04 22:01:02] GPU2 :    2064 H/s [T: 30c, RPM: 2598, CC: 1408 MHz, MC: 1100 MHz][BUS:37]
[2018-12-04 22:01:02] GPU3 :    2058 H/s [T: 37c, RPM: 3687, CC: 1408 MHz, MC: 1100 MHz][BUS:34]
[2018-12-04 22:01:02] GPU4 :    2056 H/s [T: 38c, RPM: 3791, CC: 1408 MHz, MC: 1100 MHz][BUS:9]
[2018-12-04 22:01:02] GPU5 :    2061 H/s [T: 25c, RPM: 2574, CC: 1408 MHz, MC: 1100 MHz][BUS:13]
[2018-12-04 22:01:02] GPU6 :    2057 H/s [T: 34c, RPM: 3693, CC: 1408 MHz, MC: 1100 MHz][BUS:31]
[2018-12-04 22:01:02] GPU7 :    2064 H/s [T: 32c, RPM: 2598, CC: 1408 MHz, MC: 1100 MHz][BUS:16]
[2018-12-04 22:01:02] GPU8 :    2055 H/s [T: 39c, RPM: 3707, CC: 1408 MHz, MC: 1100 MHz][BUS:25]
[2018-12-04 22:01:02] GPU9 :    2060 H/s [T: 33c, RPM: 2560, CC: 1408 MHz, MC: 1100 MHz][BUS:40]
[2018-12-04 22:01:02] GPU10:    2057 H/s [T: 32c, RPM: 2621, CC: 1408 MHz, MC: 1100 MHz][BUS:22]
[2018-12-04 22:01:02] GPU11:    2061 H/s [T: 27c, RPM: 2570, CC: 1408 MHz, MC: 1100 MHz][BUS:5]

maybe that helps, but maybe CHIP´s from your cards are to hot in about 8 hours? check that too for sure.

greetings
Patrick  

This is not true. Your setup and  HR is FAKE.  You not send any  SS for this. i have 3000HR (125W) with vega 56 on tube algo  you only 2050? your setup is bad

[2018-12-08 20:11:026] GPU0 :    3143 H/s [T: 21c, RPM: 2387, CC: 1380 MHz, MC: 1090 MHz][BUS:7]
Settings: { "id" : 0, "intensity" : 63, "worksize" : 8, "threads" : 2},

for all other gys. This is not true.

Vega 64 LE not have 3700 RPM(my have 3300RPM max) - all other card not have temp 30C
i have vega  64LE and this temp have only or NORTH POLE
normal TEMP for 64LE is 40C on 2000RPM

LOL Mate!
WTF you´r taking about??? Why should i post any "fake" for??? look previous Pages i posted the same, AND iv´e WATERCOOLED Vega64 and cool it from my window directly with 7-10 degrees and i can draw my fans up to 4900 RPM, so dont blame me please, if you have no brain or informations about other series of card´s! THX!  

So you have 3143 H/s on "TUBEV2" So please share me your FULL settings! I am ready to learn something new!  

greetings
Patrick

Send 1 SS statistic 30min on tube. You not send 1 SS and your statistic is FAKE. I have Vega 64 LE and max RPM is 3300.
4900 rpm have air vega not LE.

2000 + on TUBE? i have 4000 HR now
tomorow i have 5600HR on tube with 1 vega 56 - this is true (and your statistic is true )

wona see SS for 5600 on 1 vega 56? first send your SS with 2050H on tube.

for other gys- This HR is FAKE.

setupo and HR for lsHaKeR on tube:
[2018-12-09 12:31:22] GPU0 : 4356 H/s [T: 36c, RPM: 2174, CC: 1498 MHz, MC: 1110 MHz][BUS:7]

http://imgbox.com/EKKivdDp

[evlo]

11 gpus are nitro+ rx570 8g samsung
1 gpu is xfx rx580 8g samsung
is this optimal config?
cnv8
[2018-12-08 11:56:20] GPU0 :    986 H/s [T: 62c, RPM: 1902, CC: 1300 MHz, MC: 2040 MHz][BUS:7]
[2018-12-08 11:56:20] GPU1 :    983 H/s [T: 57c, RPM: 1788, CC: 1300 MHz, MC: 2040 MHz][BUS:12]
[2018-12-08 11:56:20] GPU2 :    976 H/s [T: 59c, RPM: 1809, CC: 1300 MHz, MC: 1970 MHz][BUS:10]
[2018-12-08 11:56:20] GPU3 :    987 H/s [T: 57c, RPM: 1784, CC: 1300 MHz, MC: 2040 MHz][BUS:2]
[2018-12-08 11:56:20] GPU4 :    978 H/s [T: 61c, RPM: 1898, CC: 1300 MHz, MC: 1970 MHz][BUS:8]
[2018-12-08 11:56:20] GPU5 :    987 H/s [T: 54c, RPM: 1518, CC: 1300 MHz, MC: 2040 MHz][BUS:5]
[2018-12-08 11:56:20] GPU6 :    987 H/s [T: 58c, RPM: 1855, CC: 1300 MHz, MC: 2040 MHz][BUS:1]
[2018-12-08 11:56:20] GPU7 :    988 H/s [T: 60c, RPM: 1885, CC: 1300 MHz, MC: 2040 MHz][BUS:13]
[2018-12-08 11:56:20] GPU8 :    1024 H/s [T: 66c, RPM: 1707, CC: 1300 MHz, MC: 2040 MHz][BUS:4]
[2018-12-08 11:56:20] GPU9 :    987 H/s [T: 55c, RPM: 1780, CC: 1300 MHz, MC: 2040 MHz][BUS:3]
[2018-12-08 11:56:20] GPU10:    986 H/s [T: 55c, RPM: 1776, CC: 1300 MHz, MC: 2040 MHz][BUS:9]
[2018-12-08 11:56:20] GPU11:    978 H/s [T: 62c, RPM: 1852, CC: 1300 MHz, MC: 1970 MHz][BUS:11]
[2018-12-08 11:56:20] Total:    11847 H/s
~1600w for whole rig

finally now, when profitability is low, i'm starting sw benchmarks - i mine with each miner sw for 48h and check consumption and real hasheshes on pool. Obviously clean boot before every sw switch. But unfortunately nothing to publish, as with srb miner i noticed i run eth OC (1200mhz) instead of 1300mhz on most cards. I still don't know if 1300mhz is optimal for XMR performance/consumption wise.

I hope at least now, I can get reasonable replies from people with experience and no bullshit about 70w consumption of 570 card etc. like there were people claiming in January.

I guess I have this on all the cards (i extracted from two different and verified all cards are samsung K4G80325FB )
777000000000000022CC1C00CEE55C46C0590E1532CD66090060070014051420FA8900A00300000 012123442C3353C19

voltages obviously lowest stable without some extreme trying for each mV.

SRBbios editor setting is:
777000000000000022CC1C00EF695D49D0571016360D460B0060870014051420FA8900A00300000 01B11333DB4303A17

Mind sharing your config on SRB miner?

this card seems ok

[2018-12-08 11:56:20] GPU8 :    1024 H/s [T: 66c, RPM: 1707, CC: 1300 MHz, MC: 2040 MHz][BUS:4]

if its samsung you can go higher (v8 is a bit to memory intensive), try 2100

MY core is 1270 on 4gb nitro+ getting 1025 on all cards

That one is 580 i think, no config, i test all mining sw with default or auto tune settings. 2100 would be for the slower memory timings. If you get 1025 with rx570s with slower timings and 2100 memory it might be worth a try for me.
Currntly this is pretty much maximum I can run without errors..

[korm]

777000000000000022AA1C00315A5436A0550F15B68C1506004082007C041420CA8980A9020004C 01712262B612B3715
 

Its Elpida timings good for micron ??

why elpida? its my timings for yeston rx460 4G micron. And it's rock https://bitcointalk.org/index.php?topic=2090765.msg28547172#msg28547172

[zeef]

Hi again doktor,

I use srbminer from months with no problem in my 9 vega 64 rig.

I use v1.6x - v1.7.0 weeks with no restart or hang.

But version 1.7.2, after 8 hours or so hangs.

In log i got this:

https://justpaste.it/4rctv

Any ideia what could be?

Regards

[2018-12-08 03:21:02] watchdog: GPU1 [BUS: 32] hashing speed is 0 H/S
[2018-12-08 03:21:13] watchdog: GPU1 [BUS: 32] hashing speed is 0 H/S
[2018-12-08 03:21:24] watchdog: GPU crash detected, miner will try to restart

[2018-12-08 03:50:02] Startup monitor detected that GPU's could not be initialised in 120 seconds

Looks like this gpu crashed so badly that even after miner restart it won't start , and it can't be initialised on miner startup.
Set a watchdog script to restart whole machine on crash, then on windows start set miner to start auto.

Hi doktor

Thanks for help,

I use the reboot_script_gpu_watchdog pointing to your reboot-windows batch file,  its ok like this ? : 

"reboot_script_gpu_watchdog"  : "reboot-windows.bat",

What im reporting is only crashes that wasn t before in vegas rigs .

All my 3 rigs crashed badly with 1.7.2 version. Even my standalone vega 64 and 56 on my two desktop pc had one appcarsh ->  https://justpaste.it/7jrik

In my rx580 rig all ok with a crazy performance boost  

Thanks

[wgestickley]

Hey Doc,

i have a quick Question about the Devfee.

I am currently testing the Cryptonight Profit Switcher, Version 6.0.0. The Profit Switching closes the Miner complete and restart it when a Algo/Coin Switch accure.

Question: What is is this case with the DevFee? Do i have to "pay" it with every new Miner Start or do you have done some Modifications for this case that it is stored that this Miner has already paid the Fee?

THX
Oliver

[doktor83]

Hey Doc,

i have a quick Question about the Devfee.

I am currently testing the Cryptonight Profit Switcher, Version 6.0.0. The Profit Switching closes the Miner complete and restart it when a Algo/Coin Switch accure.

Question: What is is this case with the DevFee? Do i have to "pay" it with every new Miner Start or do you have done some Modifications for this case that it is stored that this Miner has already paid the Fee?

THX
Oliver

If the app kills the miner, and starts it again then the devfee timer is reset.
If you use a smart profit switcher which uses the stratum protocol to signal the miner there is an algo switch, then the devfee timer isn't reset, even if the miner restarts itself.

[Skylptor]

Srb  мoжeт дoбывaть aлгopитм  MTP?  Ecли нeт тo пoжaлyйcтa дoбaвьтe в Maйнep

[abg00]

Srb  мoжeт дoбывaть aлгopитм  MTP?  Ecли нeт тo пoжaлyйcтa дoбaвьтe в Maйнep

mtp etherum coin. srbminer is cryptonight miner

этo aнглoязычнaя вeткa и aвтop нe читaeт пo-pycки

[justoiurii]

Hi Dok) Could you add HYC support (https://hycon.io, Cryptonight V7)? Thanks i advance)

[columbiamining]

11 gpus are nitro+ rx570 8g samsung
1 gpu is xfx rx580 8g samsung
is this optimal config?
cnv8
[2018-12-08 11:56:20] GPU0 :    986 H/s [T: 62c, RPM: 1902, CC: 1300 MHz, MC: 2040 MHz][BUS:7]
[2018-12-08 11:56:20] Total:    11847 H/s
~1600w for whole rig

finally now, when profitability is low, i'm starting sw benchmarks - i mine with each miner sw for 48h and check consumption and real hasheshes on pool. Obviously clean boot before every sw switch. But unfortunately nothing to publish, as with srb miner i noticed i run eth OC (1200mhz) instead of 1300mhz on most cards. I still don't know if 1300mhz is optimal for XMR performance/consumption wise.

I hope at least now, I can get reasonable replies from people with experience and no bullshit about 70w consumption of 570 card etc. like there were people claiming in January.

Assuming you're using 1.7.2? I've noticed only minimal gains going above 1200mhz core. You might gain an extra 50h/s per GPU by going up to 1350-1400mhz, but you're using considerably more wattage to do so. It's not worth it in my opinion. I have free electricity and I'm only setting my core clocks at 1200.

Here are the stats from one of my 580 8gb rigs:

https://i.imgur.com/M8P72pH.png

https://i.imgur.com/7VWcvdr.png

This is the strap I use for my Samsung cards, it's the Ubermix 3.2 strap. Looks like you're using the same one.

777000000000000022CC1C00CEE55C46C0590E1532CD66090060070014051420FA8900A00300000 012123442C3353C19

Normally with this strap I can't get much above 2000mhz memory clock. Although that's normal.

[stsocrat]

1. SRBMiner-CN.exe --config Config\config-normalv8.txt --pools pools.txt
OK!

2. SRBMiner-CN.exe --ccryptonighttype normalv8 --cgpuintensity 0 --cgputhreads 2 --cpool pool --cwallet wallet --cpassword password

Initializing OpenCL devices, please wait..

Can't open config.txt

Why?

[gregshawdj]

Hi Guys

I recently upgraded to v1.7.2 so that I could mine FreeHaven. I was using AMD drivers 17.11.4 and it was running really well for a day until it crashed. I could start it up and it would run again for a few hours. SRB miner was telling me I should use newer drivers so I tried 18.5.1 and 18.6.1 but with these the miner won’t even start! I’m also running windows 10 1709 build.

I can get this version of the miner to mine other CN algos.

I have also noticed that even if I set in the config double thread false it still applies double threading? I never used double threads with the other CN coins I mined as the rig was more stable. I just can’t seem to get this version of the miner to not double thread!

This happens with all the algos, does anyone know what it’s doing this?

[NCarter84]

Hi Guys

I recently upgraded to v1.7.2 so that I could mine FreeHaven. I was using AMD drivers 17.11.4 and it was running really well for a day until it crashed. I could start it up and it would run again for a few hours. SRB miner was telling me I should use newer drivers so I tried 18.5.1 and 18.6.1 but with these the miner won’t even start! I’m also running windows 10 1709 build.

I can get this version of the miner to mine other CN algos.

I have also noticed that even if I set in the config double thread false it still applies double threading? I never used double threads with the other CN coins I mined as the rig was more stable. I just can’t seem to get this version of the miner to not double thread!

This happens with all the algos, does anyone know what it’s doing this?

Did you clear the cache?

[NCarter84]

Still looking into why my 8 GB card running slow than my 4 GB cards on CN-Lite V7.

4 GB runs around 2 khs
8 GB runs about 1.7 khs

I"ve tried alot of different combos with threads, worksizes and intensity... I went all the way up to 200 intensity with dual threads work size 8... Runs slower... I tried worksize 16, slow.... Seems like I top out at intensity of 130/8/2 at 1700.

Hey dok, got any input on this one?

[justoiurii]

HYC dev prepared new xmrig miner with hyc_cryptonight algo. Doc could you create the same?