Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux)

[MentalNomad]

I think incorrect shares is faulty shares which were detected locally.
Rejected shares are incorrect shares which were send to pool, but pool rejected it.

I think that's right.

Ok. I think rejected and incorrect shares are different. Look at the example below (Claymore 9.:

Code:

GPU #0: Ellesmere, 4096 MB available, 36 compute units
GPU #1: Ellesmere, 4096 MB available, 36 compute units
GPU #2: Ellesmere, 4096 MB available, 36 compute units
GPU #3: Ellesmere, 4096 MB available, 32 compute units
GPU #4: Ellesmere, 8192 MB available, 32 compute units
GPU #5: Ellesmere, 8192 MB available, 32 compute units
ETH - Total Speed: 162.347 Mh/s, Total Shares: 13023(2222+2204+2132+2168+2187+2227), Rejected: 0, Time: 100:05
ETH: GPU0 27.513 Mh/s, GPU1 27.380 Mh/s, GPU2 26.206 Mh/s, GPU3 27.200 Mh/s, GPU4 27.025 Mh/s, GPU5 27.024 Mh/s
Incorrect ETH shares: GPU0 25, GPU1 76, GPU2 10
 1 minute average ETH total speed: 162.193 Mh/s
Pool switches: ETH - 0, DCR - 0
Current ETH share target: 0x00000000ffb34c02 (diff: 4300MH), epoch 34(1.27GB)
GPU0 t=73C fan=60%, GPU1 t=73C fan=50%, GPU2 t=72C fan=40%, GPU3 t=74C fan=49%, GPU4 t=69C fan=37%, GPU5 t=69C fan=38%

zero (0) rejected shares, but multiple incorrect shares: GPU0 25, GPU1 76, GPU2 10

Any ideas?

I can see one incorrect share or two in some of my 24-hour logs, but not most of them. Maybe I've just never seen one, on-screen.

I do notice that you have six cards... and of the four cards over 70C, three are throwing incorrect shares. The two that are below 70C are clean.

I can tell you that I crank my fans way higher at those temps than you do; when the GPU core is that hot, the memory may be even hotter, and many cards can't handle it. I have cards that crap out entirely when they break 70C. Some cards can obviously go way hotter; varies between brands and models, and within a model, can vary by individual card and/or where it is in the rig (because of airflow.)

Also, side note: the power draw for fans goes up more than linearly as you go above 50%; the back pressure makes the fan noisy and less power-efficient. Backing off the OC may even be more cost-effective than pinning the fan at 100%.

[Victorio]

zero (0) rejected shares, but multiple incorrect shares: GPU0 25, GPU1 76, GPU2 10

Any ideas?

The memory does not seem stable. If you get 1-3% of incorrect shares in solo mining, the memory lacks voltage or timings are wrong. Have you flashed your GPUs with modified BIOS?

Code:

I suppose your RX480/RX580 are equipped with Elpida memory and you flashed the timings from RX470/RX570 equipped with Samsung.

First of all, I would recommend checking the memory stability - run the cards at stock frequency and voltage. If you still get incorrect shares - you set wrong timings in the BIOS timings table.
Second, use OverdriveNTool to enforce voltage levels, say 900 mV for both core and memory.
Third, do the fine tuning for each GPU, until incorrect shares do not show up anymore.

I second the idea of MentalNomad that overheating might cause memory instability.

[SnigglysMine]

Need help please!    I am configuring a mixed card rig (by mixed I mean Nvidia and AMD cards on the same rig).  Using Claymore's latest release and I am unable to get the decred portion to work.  It keep saying:

DCR: Authorization failed
: null


The hashrate shows up though (3640.5 Mhs)

Any ideas?

Here's my command line:

EthDcrMiner64.exe -epool us1.ethpool.org:3333 -esm 1 -di 01234 -ewal "walletaddress" -epsw x -dpool dcr.suprnova.cc:3252 -dwal "walletaddress.workername" -dpsw x -allpools 1

[Kumic]

Has anyone tried Claymore's v10.1 Dual Ethereum AMD+NVIDIA GPU Miner?
I had some issues lately with AMD cards, I can say that they are working ok now,
but I would like to know how new miner does his job.

[greatauror28]

confirmed the wind 10 64 bit new drivers work for 10+ cards

I have a 10 GPU rig running now, what I did find is uninstalling afterburner first helps

also with so many GPU it would also sometimes hang on when u run claymore, I fixed this by using -gser 2 option

It still hangs when i try and dual mine so I need to do more tweaking.

ALso AMD settings and MSI after burner do not play nice togther, I decided to roll with wattman since I cannot set the voltages on this mortherboard

the ASUS 16 Slot one with MSI AB

Which GPU that you use on your rig?
pure AMD GPU  or mix with nvidia?

No all Rx570 and 470 mix , no nividia cards

Are you saying actually that you're running 10 AMD cards on WinX with "amd latest driver" - is this correct?
And if so what rates are you getting on ethash with this driver?

12 cards actually, just added two more this AM. and Yes thats correct.
hash rates are the same as block chain driver, the kicker is my systems are way more stable with this new driver

I had one rig of six R9 nano I upgraded to this new driver with modded evilop bios that used to crash one a day, its been two days with the new driver and its rock solid.

I also have not yet seen the issue i had before with the robin hood and beta blockchain driver where I would see hashrate of 8.33 or some crap sometimes, but its still
early.

Only issue with the new drivers is they can be hard to install, sometimes taking multiple DDU cleanings to install, and I HAD to remoVE AB.

I might install it back , becase wattman is crap

that's interesting. is that a win update, or just driver? you should try to use overdriventool for clocks and volts, it's better then claymore - as you can set proper stages as well.
also AB is known to have nasty issues

not sure I follow you...please indicate the rates that you are getting from rx570 on ethereum with latest(version?) driver?

just tested some of the latest "crimson-relive-17.10.2-oct23"
and it doesnt not have the dag fix, so no matter if it supports over 8 cards (it also works better with the volts from overdriventool ...) but 23MH on ETH vs 30MH on 17.7.2 or whatever the ver number of blockchain drv...

Did you change the workload setting in AMD settings?  I updated my 2 GPU rig and didn't lose any speed.  Probably upgrading my 5 and 11 GPU rigs sometime this weekend.

I've been having this stupid OpenCL error for a whole week since last night and i've tried all the troubleshooting and tried-and-tested workarounds but to no avail - until I updated to Win10 Fall Creator's Update and AMD 17.10.3 last night.

MY RIG IS FINALLY STABLE AF! 

It's been running 15+ hrs now without any OpenCL errors nor any hiccups.

I'm so relieved to finally have it stable and working perfectly.

What a nice birthday present 

[Bibi187]

I would like to be honest, if i can escape your dev fee i will do it ...

It seems you made a lot of money and dont keep this one good has he can be with all found you get.

I expect some release with new algo top hashrate, more dual options, more detailed "s" report, like hashrate for 6hours, 12hours, 24hours.

The last update you made about DAG creation is ON YOUR is nice ...

Cause without this is more like 3,5% - 4% we loose ...

So thanks for this and do more good work please!

I would add where is checksum ? You earn some million dollars ... You have time to make a checksum ... Or pay someone for oO

[Gunna01]

I would like to be honest, if i can escape your dev fee i will do it ...

It seems you made a lot of money and dont keep this one good has he can be with all found you get.

I expect some release with new algo top hashrate, more dual options, more detailed "s" report, like hashrate for 6hours, 12hours, 24hours.

The last update you made about DAG creation is ON YOUR is nice ...

Cause without this is more like 3,5% - 4% we loose ...

So thanks for this and do more good work please!

I would add where is checksum ? You earn some million dollars ... You have time to make a checksum ... Or pay someone for oO

Personally I think there should be a timer set for DEV mining. Where there is 2 hours of DEV mining then it stops or reduces to once every 12\24 hours. Counter is reset each time the miner restarts. Each miner then contributes 2 hours of hashing time towards the development of Claymore and then once or twice more daily. I really dont like the idea of ongoing DEV fee especially if my miner has an uptime of weeks\months.

[leonix007]

Hello, Guys. I'm asking for help as I got stuck with a problem. I read through the topic and tried many things. but didn't work out.

I have two farms mining ETH+DCR for 6 and 4 cards. The problem with a second farm.

I have 4 RX570 cards, motherboard ASrock z97. It used to give me 100 mh/s stable. However, then miner started to crash with the message:
"NO AMD OPENCL or NVIDIA CUDA GPUs found, exit".

ok... uninstall drivers, update cards with Crimson ReLive beta Blockchain Driver, patch drivers, restart and launch miner again.

The problem arises exactly in 120 minutes again... not 119 or 121, it is exactly two hours before it crashes again. The monitor blinks a couple of time. speed goes to 0, and the end.

I tried to repatch cards via bios (atiflash), played with graphic cards, nothing works. The only thing I noticed that if I launch 3 cards from 4... they work fine for a relatively long time without crashes. Notable, that I tried each of the card, and they seem to be normal.

However, when i start all 4... it works only 120 minute and "NO AMD OPENCL...".

Please, let me know if there are any ideas, I can provide you with more info, whatever you ask.

I don't have a solution, but if it's really exactly 120 minutes, one workaround is to set the rig to auto-reboot after 119 minutes.

I've used that as a workaround on rigs running Win 10 Ent trial, which is good for 90 days... after 90 days, it expires, and will shut down after 60 minutes. So I set the machine up to reboot in 59 minutes. With a rig on SSD, it actually mines 98% as much as usual despite doing the reboots, so it's a good workaround until you have time to fix it properly.

Try to change the riser of the fourth gpu, if still persist, check the psu power rating, been on that, my psu is not enough powering up 4th card when dag is getting bigger along the way.

[dollos]

All the idiots that are trying to cheat Claymores system please know that you are stealing. the man puts his effort and time into giving us something that is making us money.
he is entitled to make that money. if you dont like that then dont use it fair and simple.

Yes he is a dick most of the time with his attitude of knowing it all but still that does not allow or give us the right to steal from him.


Anyone who is stealing in this form know that you are feeding your families with stolen Money!!!

Wrong is always wrong and if you support wrong in whatever way you need to look deep down in your heart and check where it has all gone wrong.

This is a good and clean community so lets keep it that way.

[DJMorgan]

I recommend the developer - to improve the code for performance, in comparison with version 9.8 it became worse - not important is the hash the number of the shares is important - there is really performance.
Also work on stability, a lot of failures per day. with ping is not associated with loss of packets.
PS: RX 580

[Ursul0]

ok gyus some good news:
crimson-relive-17.10.2-oct23 driver is
1) more stable than the blockchain drv (has the DAG fix that can be enabled with "compute" workload)
2) has proper API (no issues for setting up the volts with overdriven tool)
3) supports more then 8 AMD gpus on the same platform

all good news indeed.

[DJMorgan]

Will there be support for 14-16 cards in Windows? the output of drivers will be under Linux? Right now I use on 14 GPU ethOS.

[Ursul0]

Will there be support for 14-16 cards in Windows? the output of drivers will be under Linux? Right now I use on 14 GPU ethOS.

I guess it's there as long as you have free pcie lanes.
AMD apparently fixed... whatever was their issue with windows

[staman]

ok gyus some good news:
crimson-relive-17.10.2-oct23 driver is
1) more stable than the blockchain drv (has the DAG fix that can be enabled with "compute" workload)
2) has proper API (no issues for setting up the volts with overdriven tool)
3) supports more then 8 AMD gpus on the same platform

all good news indeed.

https://s1.postimg.org/1jpgnkmudb/10x570.png

Well, i strongly disagree. My personal experience (and plenty of others) is completely opposite to yours

1) how can you even say that out loud? that driver is not even 7 days old
2) OverdriveNTool has been working just fine with blockchain drv (23aug)
3) yes, it let you post it in device manager and yes, it let you mine with it.. but do try enable compute workload and see that mess if you are running 12 cards! And loosing 4mh's per card with NO-dag-driver * 12 cards.. hell no !

Please stay away from such a comments, you are fooling others!

[Kukus]

I did a try to install new AMD drivers. The clue is that I am unable to change "GPU workload" setting for one card in a rig. It remains at "Graphics" when I trying change it to "Compute". All other cards (similar ones) switched with no problems.

[greatauror28]

ok gyus some good news:
crimson-relive-17.10.2-oct23 driver is
1) more stable than the blockchain drv (has the DAG fix that can be enabled with "compute" workload)
2) has proper API (no issues for setting up the volts with overdriven tool)
3) supports more then 8 AMD gpus on the same platform

all good news indeed.

Well, i strongly disagree. My personal experience (and plenty of others) is completely opposite to yours

1) how can you even say that out loud? that driver is not even 7 days old
2) OverdriveNTool has been working just fine with blockchain drv (23aug)
3) yes, it let you post it in device manager and yes, it let you mine with it.. but do try enable compute workload and see that mess if you are running 12 cards! And loosing 4mh's per card with NO-dag-driver * 12 cards.. hell no !

Please stay away from such a comments, you are fooling others!

Actually what Ursul0 cited is pretty factual, as I've experienced it myself.
This driver, together with the Win10 FCU fixed my OpenCL issues - and i've tried everything under the sun to fix it. I used to use aug 23 blockchain driver and a recently-installed win10 pro without any updates.

I only run 8 cards but here's some proof that's it's stable.



Previously with the setup above, miner doesn't get past 3 hours without openCL issues.
It has the dag fix just like the blockchain drivers - and this version is fully-supported by AMD and not in Beta status.

Note that you have to enable *compute* settings in each card in the Radeon Settings for this to work.

[perazdera]

So i need to to set bios to factory so that it can pass signature check?

You need to patch the video drivers. Get the bios signature patcher here: https://www.monitortests.com/forum/Thread-AMD-ATI-Pixel-Clock-Patcher

Just to check if I understand it correctly. I need to install driver then patch the driver with signature patcher from that link. Does that patcher work for every driver? Mainly I want to get AMD Blockchain Workloads driver to work.

Works with AMD Blockchain driver as well as with new one.
Procedure:
1) clean you system with DDU and reboot
2) install new driver and don't reboot
3) run utility Thread-AMD-ATI-Pixel-Clock-Patcher
4) and then restart the system

Thank you. I will try this today when I get home. One more question. Because i did not modded cards myself, I bought them like that from some guy, is there some way that I can find out what is factory settings for my cards?

You can find the memory type by reading from the BIOS.  Save the bios using atiflash and read it off from Polaris Bios Editor.  That would indicate the memory you have on the GPU.  You can google for the original BIOS for that brand, particular model and memory.

Thank you guys allot, you are the best Smiley Smiley I hope I will menage to try every suggestion today, if not I will definitely try it over the weekend and get back to you for the results.

I tried it out over the weekend and it worked. Problem is solved. Thank you guys, I really appreciate for the help

[OXUS]

Hello

i have problem with my rig running claymore .... i get low hashrate on eth for some reason ...

windows 10 ... claymore v10.0

[Victorio]

i have problem with my rig running claymore .... i get low hashrate on eth for some reason ...

Post your miner configuration (the content of your .bat file).

[Ursul0]

ok gyus some good news:
crimson-relive-17.10.2-oct23 driver is
1) more stable than the blockchain drv (has the DAG fix that can be enabled with "compute" workload)
2) has proper API (no issues for setting up the volts with overdriven tool)
3) supports more then 8 AMD gpus on the same platform

all good news indeed.

Well, i strongly disagree. My personal experience (and plenty of others) is completely opposite to yours

1) how can you even say that out loud? that driver is not even 7 days old
2) OverdriveNTool has been working just fine with blockchain drv (23aug)
3) yes, it let you post it in device manager and yes, it let you mine with it.. but do try enable compute workload and see that mess if you are running 12 cards! And loosing 4mh's per card with NO-dag-driver * 12 cards.. hell no !

Please stay away from such a comments, you are fooling others!

really? this is where you want to be right? wow. fuck idiots... I guess



see at the bottom: 2*10*570


also on 2) see this: