Topic: Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux) - page 256. (Read 6590565 times)

Nice going, thanks for posting a followup.
Happy mining.

BTW: I have also see Radeon settings revert some GPU to graphics after a hard crash.
It's a good plan to map your rig, (physically identify each GPU, and it's logical number in claymore), note what hash rate you expect from each gpu.
Monitoring in Eth manager is a good way to spot check if each GPU is performing.

One of the nastier things about the AMD drivers, is the hit your hash rate takes, if it reverts to graphics, (usually more than 30%), and if that wasn't enough, while you burn the same power!

If you swap a card to another slot, it will get a new pci bus address, (and this means, radeon setting will be back to graphics), but claymore is only counting from 0.

Suppose you have 3 GPU. (GPU numbers below, as per claymore)

GPU0 is on pci node2
GPU1 is on pci node6
GPU2 is on pci node7

You set all 3 to compute in radeon settings, and away you go.

Later, for whatever reason, (new card, bios change, swap out riser, cable etc), you move GPU2

GPU2 is now moved to pci node1, in claymore you will still see 3x GPU, but the addressing will be.

GPU0 is on pci node1
GPU1 is on pci node2
GPU2 is on pci node6

You only changed/moved one GPU, but this reorders the addressing, and while this won't impact claymore, (other than to confuse you perhaps), Radeon settings are ONLY using addressing on the pci bus, so for sure you will have to set compute mode for GPU0, (it will be graphics by default, as radeon detects it as a new gpu found, as it's on a newly used bus address).

This gets SUPER painful as you expand your rig.

I've been toying with the idea on my next build to, only add 1 single GPU, let radeon detect it, I can set to compute, shutdown, and move that single gpu to the next slot. Repeat, move, repeat.
That way, assuming the GPU make/model is the same, all bus nodes will have a "remembered" gpu/bus address, which is set to compute.

BUT all this takes a LOT of time, and with radeon settings as they are, this time increases exponentially with each GPU added.

What is SO LAME, is AMD do not, have not, added a SIMPLE install switch, to clean install drivers in compute mode, (graphics disabled), Hell, even Mr Claymore has managed that, what's AMDs excuse?
 
What bugs me, is how often this lame method to set compute mode is actually needed.

I really hope AMD make some improvements here, and my suggestion would be a installer switch

setup.exe /compute

Which tells the installer to install in only compute mode, rendering radeon settings, and graphics to compute mode redundant, and save LOADS of time and hassle for their customers.

I should add, that all the rendering farm customers would certainly appreciate that as well.

Good luck

Nice going, and thanks for taking the time to feedback results.
I was wondering about that nanopool address, it's possible they used eu1.nanopool.org:9999 previously, but have changed the servers/names and have various aliases. The problem with that can be, the routers in between, DNS cache and arp tables will have duplicates. This stuff is self healing, but can actually take days to settle down, and in the mean time, things, (like name resolution, routing etc) can take longer than expected, (expected as in, longer than the time out limits your end, or along the path etc), and as you're essentially on the far end of a long chain, all those "little" delays can add up, and the application will appear to hang your end, while in reality, it's just waiting.

Hopefully, setting the log to debug, you will see a bunch of additional entries, like "waiting... etc etc" if you don't see anything logging, then probably claymore is hung. (Don't forget, if you're using notepad to view the live log, you need to refresh/reopen it, to see new entries.
Hitting f5 in explorer, should show you the log file size increasing, is another method to determine if anything is logging.

(don't forget to disable debug log when you're stable)

Virtual memory. It's far from clear what claymore is doing with that, and although 16GB is the minimum, I'd suggest it's never a good plan to aim for the bottom!!
It's only VM anyway, set it as high as you like, (although to be safe, less than free space on disc, plus some headroom).
Suppose you have 100GB free, setting VM to 60GB should be pretty safe, assuming your rig is dedicated for only mining.

Ok, afterburner.
Less than ideal to use that imho, any application can crash, what damage can happen if afterburner crashes? I'd rather not find that out the hard way.

Did you spot in 11.7, Nvidia temperature control is now supported?
 added temperature management and overclock support for recent Nvidia cards in Windows: "-tt", "-powlim", "-cclock", "-mclock", "-tt", "-fanmax", "-fanmin" options are supported for Nvidia too.

I'd suggest using it. That way, if claymore crashes, (no more work is pushed to GPUs), and at least your GPUs revert to bios control, cool down etc.
Right now, if your afterburner crashes, and claymore carries on, you're going to potentially cook those cards, or force them to their hardware trip point, (probably something scary like 100C plus).

NOT GOOD.

Remember, mining is a bit extreme for GPUs.
Like anything engineered, they are designed for a purpose, and have a duty cycle to match.
It's not expected under typical use, that GPUs are running, flat out, 24/7, this means they will probably have a shorter lifespan than if they were used for gaming.

Render farms, (not so dissimilar from mining), are constantly swapping out/in cards, and usually have failovers and backup nodes to pick up the slack.

So the key here is stability, steady temperature control, depending on the fan type, probably constant rotation, speed regulation, rather than on/off/on is going to be preferable. Think of it like trying to avoid the extremes, and flatten out the peaks and troughs.

Mechanically, we have to consider these GPUs are getting hot! Parts are MUCH hotter than 100C, very reliant on cooling, and different materials expand/contract at different rates. Copper is one of the more extreme examples, but any such mechanical expansion is also going to reduce life, contribute to increase failure. It's the CHANGE, (expansion), that causes mechanical stress. Avoiding that as much as possible, will extend life.

If you're in this for the long run, ROI etc, then it's a good idea to consider these angles, and imho, get cards with a well engineered cooling scheme, active and passive, reduce voltages/current etc, set the upper cut limits lower, (e.g 98C to 88C), set a reasonably LOW target temp, (60C), if it means you work the fans harder, so be it, those are cheap to replace compared to a GPU.

(Find some stable settings, and customise your BIOS to reflect that). That way, if all else fails, the hardware will self protect.

Last advice, wifi.
I was amazed at how many instabilities evaporated when I switched that off.
Part of the problem is, these days, wifi routers are doing all kinds of (for open rigs, nasty stuff), like beam forming, and essentially flooding the space with microwaves, as are of course all the devices trying to connect, like phones etc.
In your typical PC, it's in a metal box, microwaves don't travel well through flat/sheet metal, and wifi antennas are OUTSIDE the PC case, (even in laptops, antennas are as far away from the motherboard  as possible.

Now we have open mining rigs, no casing, (shielding) and worse still, we have a LOAD of wire stands running all over the place. Cables from PCIe cards to risers, (should be shielded, but who knows), but also we have a lot of single strands, power lines to the GPUs, and normally ALL this stuff is further shielded inside a metal computer case. Furthermore, all components in there will share a common shield ground.

While microwaves, (and any "radio" band waves for that matter), well those LOVE single strands of wire, those being brilliant antennas, being it receiving or transmitting.

So, what all this adds up to is TROUBLE, when exposing open rig, no common ground, worse still, multiple floating grounds, multiple power supplies lots and lots of microwave sucking antennas.

As recently as 2 weeks ago, (after months of mining), I finally got around to hard-wiring LAN to all my rigs.
This comes after months of chasing down gremlins, timeouts, dropped sockets at pool, (while still having local internet access), random lockups, crashes, you name it, it's a long list, and a LOT of time.

I'd say, I'd fought hard to tune out 50~60% of events, reduced frequency for sure, but once every 7~10 days, I'd have a succession of dropped connections, resulting in stale shares down the line.

Since eliminating wifi, (and 1 rig on powerline LAN), I'm amazed at the difference, (and kicking myself, as I know better), I guess it had become a bit of a campaign, which with hindsight, was a lost cause :-)

What one needs to consider is, steady, steady, steady is the key to mining.
Eliminate as many of the potential risks to that, and you're set up to go steady :-)

Good  luck.
 

Read this

FAILOVER

Use "epools.txt" and "dpools.txt" files to specify additional pools (you can use "-epoolsfile" and "-dpoolsfile" options to use different filenames).
These files have text format, one pool per line. Every pool has 3 connection attempts.
Miner disconnects automatically if pool does not send new jobs for a long time or if pool rejects too many shares.
If the first character of a line is ";" or "#", this line will be ignored.
Do not change spacing, spaces between parameters and values are required for parsing.
If you need to specify "," character in parameter value, use two commas - ,, will be treated as one comma.
You can reload "epools.txt" and "dpools.txt" files in runtime by pressing "r" key.
Pool specified in the command line is "main" pool, miner will try to return to it every 30 minutes if it has to use some different pool from the list.
If no pool was specified in the command line then first pool in the failover pools list is main pool.
You can change 30 minutes time period to some different value with "-ftime" option, or use "-ftime 0" to disable switching to main pool.
You can also use environment variables in "epools.txt", "dpools.txt" and "config.txt" files. For example, define "WORKER" environment variable and use it as "%WORKER%" in config.txt or in epools.txt.
You can also select current pool in runtime by pressing "e" or "d" key.

You are correct, first time I turned on computing only for the first card. thanks a lot. all works fine now.

The timeout is just so msi afterburner will spin up the fans before the mining starts. I have the rig on wifi sometimes and it need some more seconds to reconnect to
It was in a single line, just automatic line break in notepad when i copied it. Same on both machines too. Virtual memory was 16gb but after a while it finally said that dag creation failed. I need 29 GB virtual memory to fill DAG on 6x gtx 1070 and 1x gtx 1060 card. I have put 32-40GB virtual memory now instead. The machine was reinstalled to with 10 pro instead of 10 pro n. Shouldn´t do anything i guess.
You are right, the server is wrong but eu1-nanopool.org actually works too
It seems that the problem was solved with virtual memory, i used 5x 1070 before and added two cards today.

OK, here's the deal with blockchain drivers.
Beta release, driver specific for mining, but now kind of dated, and only support 8 GPU max.
Later drivers, (general release), include a new feature to switch between graphics and compute in the Radeon Settings gui.
(Check out the Radeon  release notes!)

Radeon Settings / gaming /Global settings, (select each GPU in turn, LAME AMD, utterly LAME!), select compute, confirm OK for restart.
After setting all GPU to compute, shutdown PC, reboot etc.
(I've seen on every radeon release, failures to mode change, and reboot will fix that).

Suggest you use 18.3.4

Been running that for some weeks, it's stable so far, and supports up to 13 GPU per rig.

Good luck.

I would check pool setting again, last time when I got this message it had to do with my worker name and password setup

Hello, need some advice. Using V11.7 with AMD Blockchain drivers and getting 29.6 from my RX580 8Gb but when tried to use latest Adrenalin driver hash rate goes down to 19.9
This rig has 2 cards both RX 580s 8 Gb the one that gets to 32.2 has Hynix memory the one I have a problem with - Samsung.  Overclocking is the same with both drivers, I did not even touch it. Computing mode was turned on in new drivers. All patches installed in both cases.
Switched back to blockchain driver - the second card went back to 29.6 any advice what am I doing wrong?

setx GPU_FORCE_64BIT_PTR 0
setx GPU_MAX_HEAP_SIZE 100
setx GPU_USE_SYNC_OBJECTS 1
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_SINGLE_ALLOC_PERCENT 100
EthDcrMiner64.exe -epool us-east.ethash-hub.miningpoolhub.com:20535 -ewal walladdress.backroom -esm 2 -epsw x -dwal -dpsw x -nofee 1 -mode 1


Trying this pool as I'm getting a decent ping from it.  Would not work with the SSL option before the address.  Took that out and am getting:


ETH: Authorization failed
: {"id":2,"result":false,"error":null}
ETH: Connection lost, retry in 20 sec...

Try to use gnu screen.
Create screen session, then run miner and then detach from the screen.

Correct, 1050ti's.

Interesting, why the timeout 15?
Note, the exe string needs to be single line, no c returns, or line breaks etc.
Also, 2 things you could try on the problem rig.
-mode 1
-dbg 1

I'd try stripping that to the minimum, (also make sure you set your virtual memory to at least 16GB!!!)

setx GPU_FORCE_64BIT_PTR 1
setx GPU_MAX_HEAP_SIZE 100
setx GPU_USE_SYNC_OBJECTS 1
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_SINGLE_ALLOC_PERCENT 100
EthDcrMiner64.exe -epool eu1.nanopool.org:9999 -ewal wallet.rig/email -epsw x -mode 1


BUT.....I just checked, and it looks like nonopool servers are different to what you have above.

Should be

eth-eu1.nanopool.org

Are you 100% sure, the server you are using is correct?

Cheers

As explained in the README.txt, you need to add the ETH backup pools to epools.txt and secondary coin backup pools to dpools.txt. If the main pool goes down the miner will automatically cycle to the backup pools and return to the main pool once it's back online. You can also adjust the amount of time before the miner will switch back to the main pool with -ftime (default is 30 minutes).

Good luck

These are 1050ti's right?  Not regular 1050's?

I run the same config on another computer with claymore 11.7 and windows 1803. It will start without any problems. The difference is windows 10 pro on the working and windows 10 pro n on the non working. I don´t have any older logs left from the new rig. The config file is copied from the working computer.

Log from the working computer:
16:11:22:156   3504   
16:11:22:159   3504   ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ»
16:11:22:161   3504   º                Claymore's Dual GPU Miner - v11.7               º
16:11:22:165   3504   º              ETH + DCR/SIA/LBC/PASC/BLAKE2S/KECCAK             º
16:11:22:167   3504   ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ
16:11:22:169   3504   
16:11:22:171   3504   b583
16:11:22:378   3504   ETH: 3 pools are specified
16:11:22:380   3504   Main Ethereum pool is eu1.nanopool.org:9999
16:11:22:382   3504   DCR: 0 pool is specified
16:11:22:452   3504   OpenCL platform: NVIDIA CUDA
16:11:22:455   3504   AMD OpenCL platform not found
16:11:22:699   3504   CUDA initializing...

16:11:22:702   3504   NVIDIA Cards available: 1
16:11:22:704   3504   CUDA Driver Version/Runtime Version: 9.1/8.0
16:11:22:707   3504   GPU #0: GeForce GTX 1070, 8192 MB available, 15 compute units, capability: 6.1  (pci bus 3:0:0)
16:11:22:710   3504   Total cards: 1
16:11:26:724   3504   NVML version: 9.391.01
16:11:27:276   3504   SSL: Imported 54 certificates from local storage
16:11:27:320   2a04   ETH: Stratum - connecting to 'eu1.nanopool.org' <198.251.88.37> port 9999 (unsecure)


Log from the non working computer:
16:31:13:690   22a4   
16:31:13:690   22a4   ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ»
16:31:13:690   22a4   º                Claymore's Dual GPU Miner - v11.7               º
16:31:13:690   22a4   º              ETH + DCR/SIA/LBC/PASC/BLAKE2S/KECCAK             º
16:31:13:690   22a4   ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ
16:31:13:690   22a4   
16:31:13:690   22a4   b583
16:31:13:905   22a4   ETH: 3 pools are specified
16:31:13:906   22a4   Main Ethereum pool is eu1.nanopool.org:9999


And the config itself:
setx GPU_FORCE_64BIT_PTR 0
setx GPU_MAX_HEAP_SIZE 100
setx GPU_USE_SYNC_OBJECTS 1
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_SINGLE_ALLOC_PERCENT 100
TIMEOUT 15
EthDcrMiner64.exe -epool eu1.nanopool.org:9999 -ewal wallet.rig/email
-epsw x -mode 1 -ftime 10 -wd 1 -logsmaxsize 5 -r 60 -minspeed 80 -mport 0

I would suggest using the ethermine ssl port instead of the standard stratum port:

-epool ssl://us1.ethermine.org:5555

You will have MUCH less stale shares and it's a secure connection. Also make sure to setup backup pools in the epools.txt file in case the main pool goes down. If you want to get paid more frequently, you can adjust the minimum payout amount in the pool settings. Dual mining DCR, SIA, PASC or LBC is no longer a profitable option for GPU's since ASIC's were released for those algorithms.

Hi there.
Make sure you unblock EthDcrMiner64.exe
(right click, general tab, check unblock, /ok)
Turn defender ON, and set an exception for EthDcrMiner64.exe
It can be useful to relax security while debugging, but it's bad form to ignore it forthwith.

I'd suggest you take a look at your log file.
I'm assuming you mean, claymore 11.7 was working before you updated windows, and you changed NOTHING else, drivers etc, cmiiw.
That being so, look at a previous log, (from when it was working), that will show your full init, through to mining, and then compare with your current log.
At some point, your current log will halt, if you line old/new logs up, you will see what SHOULD be happening next, and that is where you need to investigate.

Good luck.

Hi there,
First off, there should be no line/break, CR, after last variable, and your exe string.

setx GPU_FORCE_64BIT_PTR 0
setx GPU_MAX_HEAP_SIZE 100
setx GPU_USE_SYNC_OBJECTS 1
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_SINGLE_ALLOC_PERCENT 100
EthDcrMiner64.exe -epool us1.ethermine.org:4444 -ewal 0xMyEtherWalletAddress.backroom -esm 2 -epsw x -dpool stratum+tcp://dcr.suprnova.cc:3252 -dwal Redhex.my -dpsw x -nofee 0 -mode 1


Question: How is this computer connecting to the internet?
If you are using wifi, DON'T, connect via CAT6, ideally direct into the router/modem link to internet.

There are multiple postings with this issue, I also encountered it in my early days mining, where I was feeling my way in, working off my desk, in isolation, and wifi was convenient.

Also, try ETH only mining, (-mode 1), if you're PSUs are weak, not only will cards fail to init, but you can pull down other things, (like wifi).
I also had some funky system crashes, eventually traced that to openrig concept, zero shielding from microwaves, and the wifi as kicking it offline. This was very repeatable, if I placed myself between the rig and the antenna, it was ok, I'd get up from my desk, and 1~2 times every other day, it would crash.

Since I switched to cable connection, I've not had a single socket time out error.

How often are you seeing that?

If it's occasional, (1~2 times a day), that could be normal, depends on the quality of your internet connection.

You could also try this in the console

tracert us1.ethermine.org

Less is better, I just tried that from the UK, it's rubbish, I've hit 30 hops, and it's timing out on many.
You could also try

ping /t us1.ethermine.org

Keep that console open, and next time you see the socket closed, check you still have a connection, (not timing out).

There are basically 2 possibilities here, the socket is closed your end, (as us1.ethermine.org became unreachable), or ethermine closed it their end, (unlikely) but possible if under attack for example.


Saving the best to last,
Lastly, what adjustments do I need to make to get paid as quickly as possible?

I can certainly help you here, it's simple, mine faster :-)

Anyone got issues after updating to Windows 1803?

Miner just stops without any errors
15:23:19:655   2624   Check and remove old log files...
15:23:19:655   2624   args: -epool eth-eu1.nanopool.org:9999 -ewal MYWALLET.MYRIG/MYEMAIL -epsw x -mode 1 -ftime 10
15:23:19:655   2624   
15:23:19:655   2624   ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ»
15:23:19:655   2624   º                Claymore's Dual GPU Miner - v11.7               º
15:23:19:655   2624   º              ETH + DCR/SIA/LBC/PASC/BLAKE2S/KECCAK             º
15:23:19:655   2624   ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ
15:23:19:655   2624   
15:23:19:655   2624   b583
15:23:19:874   2624   ETH: 3 pools are specified
15:23:19:874   2624   Main Ethereum pool is eth-eu1.nanopool.org:9999

Windows defender turned off.