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

[MaxMidnite]

For RX570/RX580
If 4gb cards.  59 intensity, wksize 8, dual threads
If 8gb cards   114-120, wksize 8, dual threads

Hi man + others

Thanks for the info, these are my new set of results; (double threads = true)

I cannot push this over 86 with 80 being most stable (5200+ hash)

ALL CARDS ARE 8GB

"gpu_conf" :
[
   { "id" : 0, "intensity" : 86, "worksize" : 8, "threads" : 2, "kernel" : 0},
   { "id" : 1, "intensity" : 86, "worksize" : 8, "threads" : 2, "kernel" : 0},
   { "id" : 2, "intensity" : 86, "worksize" : 8, "threads" : 2, "kernel" : 0},
   { "id" : 3, "intensity" : 86, "worksize" : 8, "threads" : 2, "kernel" : 0},
   { "id" : 4, "intensity" : 86, "worksize" : 8, "threads" : 2, "kernel" : 0},
   { "id" : 5, "intensity" : 86, "worksize" : 8, "threads" : 2, "kernel" : 0}
]
}

[2018-06-08 15:09:52] GPU0:    896 H/s [T: 57c, RPM: 2187, CC: 1150 MHz, MC: 2175 MHz][BUS:5]
[2018-06-08 15:09:52] GPU1:    842 H/s [T: 72c, RPM: 2128, CC: 1150 MHz, MC: 2175 MHz][BUS:1]
[2018-06-08 15:09:52] GPU2:    848 H/s [T: 60c, RPM: 2031, CC: 1150 MHz, MC: 2175 MHz][BUS:4]
[2018-06-08 15:09:52] GPU3:    852 H/s [T: 63c, RPM: 2127, CC: 1150 MHz, MC: 2175 MHz][BUS:3]
[2018-06-08 15:09:52] GPU4:    893 H/s [T: 54c, RPM: 2918, CC: 1150 MHz, MC: 2175 MHz][BUS:2]
[2018-06-08 15:09:52] GPU5:    893 H/s [T: 44c, RPM: 3036, CC: 1150 MHz, MC: 2175 MHz][BUS:6]
[2018-06-08 15:09:52] Total:   5224 H/s


If I do I get this error:
 Error CL_MEM_OBJECT_ALLOCATION_FAILURE when calling clEnqueueNDRangeKernel for kernel 0 for DeviceID 1 (Thread 3)

How much is your virtual memory set to?  Make it at least  sum of VRAM of each GPU.
Also in order to hash higher increase core clocks to 1250 range (voltages 875-925mV core, 900mV mem)

Also I don't have  "kernel" : 0, in my GPU config settings.  Not sure how that changes things.

Hi Man,

Thanks for the reply, you are correct my core clock values where bit low from previous setup to save power.
Ether mining different to v7 mining. I have now increased CC value to 1250 and 1000/mv
I have initial VRAM set to 34000 and 56000 Max - this should be plenty.
I'm getting more hash now with this new setting, see below. I know wonder if can increase intensity further due to changes.

[2018-06-09 08:46:49] GPU0 :    937 H/s [T: 58c, RPM: 2214, CC: 1250 MHz, MC: 2175 MHz][BUS:5]
[2018-06-09 08:46:49] GPU1 :    812 H/s [T: 73c, RPM: 2119, CC: 1250 MHz, MC: 2175 MHz][BUS:1]
[2018-06-09 08:46:49] GPU2 :    880 H/s [T: 63c, RPM: 2047, CC: 1250 MHz, MC: 2175 MHz][BUS:4]
[2018-06-09 08:46:49] GPU3 :    856 H/s [T: 66c, RPM: 2132, CC: 1250 MHz, MC: 2175 MHz][BUS:3]
[2018-06-09 08:46:49] GPU4 :    936 H/s [T: 55c, RPM: 2901, CC: 1250 MHz, MC: 2175 MHz][BUS:2]
[2018-06-09 08:46:49] GPU5 :    938 H/s [T: 42c, RPM: 3090, CC: 1250 MHz, MC: 2175 MHz][BUS:6]
[2018-06-09 08:46:49] Total:    5359 H/s

[1679790954]

1679790954

[brunocrypto]

I have 8x Video Cards but after the last update(v1.5.9) i can only see GPU0 but the hash rate is right showing the value of all 8x cards.
I don't know what is wrong, maybe i messed something, i merged all the files except config.txt and pool.txt.

I'm still testing, ill try to do a fresh config again to solve the issue.

Same problem here ! (but i don't know if it come from the last update, first time i use SRBMiner)

But it's only do it on 1 rig on the 3 i got

[MaxMidnite]

V1.6.0
- Added support for Haven new algo after fork (block 89200)
- Added support for Masari new algo (fast) after fork (block 204000)
- Job timeout default is now 20 minutes
- More logging on miner startup
- Added option 'persistent_memory' in gpu_conf

+ 2 new algos added : 'haven' and 'fast'
So after the fork you can use them like :

"cryptonight_type" : "fast"
"cryptonight_type" : "haven"


+ As some have issues with miner 'just waiting some time' on startup, i added more logging option so we can catch where it hangs, and i can possibly fix that
+ 'persistent_memory' parameter added to gpu_conf, basically what it does is tries to allocate some memory that is normally not available to the GPU. It can increase hashrate in some ocassions, by finding the right intensity and threads setup, but also it can make the gpu crash, so it's up to you to experiment

I am still looking for the cause of random miner crashes (mostly on new 'great' win 1803 update), also for the abnormality that causes higher hashrate on after X runs.

Thanks for the update seems good, sorry for silly question, how does one access the log file?

Also for persistent_memory what kind of test rig you test on and any recommended settings?

[RuMiner]

+ 'persistent_memory' parameter added to gpu_conf, basically what it does is tries to allocate some memory that is normally not available to the GPU. It can increase hashrate in some ocassions, by finding the right intensity and threads setup, but also it can make the gpu crash, so it's up to you to experiment

thanks for update
is this correct usage?
   { "id" : 0, "intensity" : 112, "worksize" : 16, "threads" : 2, "persistent_memory" : true},
didn't crash, but no hashrate changes too

PS you broke JSON API. it shows "cryptonight_type": "\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000" instead of normalv7

[LeonManveli]

Hi all! where is the SRBMiner Cryptonight AMD GPU Miner V1.6.0 link? there are only old versions


http://

[doktor83]

+ 'persistent_memory' parameter added to gpu_conf, basically what it does is tries to allocate some memory that is normally not available to the GPU. It can increase hashrate in some ocassions, by finding the right intensity and threads setup, but also it can make the gpu crash, so it's up to you to experiment

thanks for update
is this correct usage?
   { "id" : 0, "intensity" : 112, "worksize" : 16, "threads" : 2, "persistent_memory" : true},
didn't crash, but no hashrate changes too

PS you broke JSON API. it shows "cryptonight_type": "\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000" instead of normalv7

FIXED, thanks for reporting

[LeonManveli]

Thank you doktor83

[Iamtutut]

Coin Bixbite - on the algorithm cryptonight heavy
Mining Pools
https://pool.bixbite.pro Official pool

Wallet (CLI)
https://github.com/BixBite-project/bixbite/releases

https://t.me/bixbitecoin

"•BXB-miner with CPU/NVIDIA support" for a cryptonight based coin...AMD cards are much more efficient to mine CN, even heavy.

[dingdongtobias]

Thanks for the update seems good, sorry for silly question, how does one access the log file?

You must turn on log first, put in start.bat :

SRBMiner-CN.exe --config config.txt --pools pools.txt --logfile log.txt

[RuMiner]

Thanks for the update seems good, sorry for silly question, how does one access the log file?

You must turn on log first, put in start.bat :
SRBMiner-CN.exe --config config.txt --pools pools.txt --logfile log.txt

N.B. if you point a subdir, you shoud create it first, miner cannot make it itself
I use "--logfile logs\%date%-%random%.log", so every miner start is a different log file

[YukiSG]

For those using 4GB 570/580.

1.6.0 has some major update (IMO) that benefits quite alot.

Dual Thread now Hashes higher than single thread. (roughly 20-40h/s increase per card)

Currently im back to using

29 intensity with dual thread of worksize 8.

as compared to 59 intensity single thread worksize 8.

I gained 20h-40h/s per card.

my Elpida card is hashing at 900/hs while the rest of the hynix are hashing at 810 - 860h/s

Like my work/Discovery?

Donate me to help me clear my debt.

ETH: 0x052898045e73a56691eca5210EE6a2Aa15397295

[popow_sergei]

[2018-06-08 23:07:33] Error CL_OUT_OF_HOST_MEMORY when creating clCreateCommandQueue for DeviceID 0 (Thread 0)
[2018-06-08 23:07:33] Error initing GPU's. Stopping miner process

[Kgonla]

Here is what is slowing down the startup:

[2018-06-09 13:52:22] CryptonightV7 mode enabled
[2018-06-09 13:53:46] DevFee pool SET
[2018-06-09 13:53:46] DevFee address SET
[2018-06-09 13:53:46] Starting init of mining threads

The difficulty monitor is slow responding & varies alot, the fastest I had after 4 times have been 12 seconds:

[2018-06-09 13:56:38] pool_have_job: Pool difficulty: 60000
[2018-06-09 13:56:38] pool_have_job: Pool sent a new job (ID: 487651957152411)
[2018-06-09 13:56:50] difficulty_monitor: Difficulty (60000) is higher than the defined max difficulty for this pool (40000)!

And the worst 42 seconds

Also, after running the program for a few minutes it closed without advice

[hesido]

Kill windows 1803 update with fire: Here's what we have so far about win 1803 update effect on me, after applying all known fixes:

I can't get all 13 gpus to work above 900 h/s each, a very low intensity seems to work. I used to do stable 1040 h/s.
Hash rate fluctuates badly.
GPU-Z trick does NOT work.

tldr, Windows 1803 update is utter bullshit. I failed to prevent it from installing, so it's my fault. I'm thinking of leaving to a miner Os, I hope SRBMiner adds linux compatibility.

[Kgonla]

I think I can confirm...

After the difficulty monitor is executed for 4th time the program closes itself

[2018-06-09 15:01:33] Connection to pool lost. Reconnecting in 10 seconds.

And the program is closed.

[lebuawu2]

Kill windows 1803 update with fire: Here's what we have so far about win 1803 update effect on me, after applying all known fixes:

I can't get all 13 gpus to work above 900 h/s each, a very low intensity seems to work. I used to do stable 1040 h/s.
Hash rate fluctuates badly.
GPU-Z trick does NOT work.

tldr, Windows 1803 update is utter bullshit. I failed to prevent it from installing, so it's my fault. I'm thinking of leaving to a miner Os, I hope SRBMiner adds linux compatibility.

I think it's more easy to just go back to an earlier build of Windows 10, that solved my problem.

http://www.thewindowsclub.com/go-back-to-an-earlier-build-of-windows-10

[doktor83]

I think I can confirm...

After the difficulty monitor is executed for 4th time the program closes itself

[2018-06-09 15:01:33] Connection to pool lost. Reconnecting in 10 seconds.

And the program is closed.

i will check it out.
 Also it's not triggerring momentally when the set difficulty is reached. Check is made every 45 seconds.
No point setting it for ex 1 sec because it just eats up cpu resources, and i think its not THAT important to disconnect momentally.
But correct me if i am wrong.

[SeriousOne]

@ dok

I am so sorry, i saw you have done changes on DevFee.
I just wanted to report it was really never caused by DevFee, it was just a coincidence.

I don't use Windows 10 for personal use, i have used only Windows 7 before very well.
So i never looked in to the new Settings Panel of the "metro" gui, i just prefer to use old windows settings windows.
Seems to that i have to adopt the new Settings panels.

Sorry again, best miner we have ever seen, happy to know that you get your well deserved fees  

[hesido]

Kill windows 1803 update with fire: Here's what we have so far about win 1803 update effect on me, after applying all known fixes:

I can't get all 13 gpus to work above 900 h/s each, a very low intensity seems to work. I used to do stable 1040 h/s.
Hash rate fluctuates badly.
GPU-Z trick does NOT work.

tldr, Windows 1803 update is utter bullshit. I failed to prevent it from installing, so it's my fault. I'm thinking of leaving to a miner Os, I hope SRBMiner adds linux compatibility.

I think it's more easy to just go back to an earlier build of Windows 10, that solved my problem.

http://www.thewindowsclub.com/go-back-to-an-earlier-build-of-windows-10

Doing this as we speak, great tip, I honestly thought it was only possible with windows restore turned on, which I had turned off.

Awesome..

[JuanHungLo]

dok,
I have an interesting problem.  Using v1.6.0, with 5 rx470-8gb, stellite4 algo:  All 5 GPUs initialize and are identified correctly on your miner, then after pool connection, display shows only FOUR GPUs, but the first one, GPU0, has DOUBLE the output of all others, I tried restarting, but with same output.  Now this isn't a hashrate problem at the pool, just on the miner display.  I think that the GPU bus identification is wrong on one of the cards by the miner software.