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

[cryptoillusion1]

Yesterday I tried to undervolt my rig and now I am having some major issues, any help would be greatly appreciated.

Code:

Rig is 6x RX 580s (all with mixed Hynix and Samsung memory) on Asus Mining Expert board running blockchain driver with Polaris one click bios.

The usual OC and voltage setting I use in the config file for Claymore's miner are

-cclock 1150
-cvddc 900
-mvddc 950
-mclock 2200

My usual power usage at the wall is 1070W, HWinfo usually shows gpu chip power of  97w per card.

The settings I tried were 

-cclock 1150
-cvddc 875
-mvddc 925
-mclock 2150

After starting Claymore v 11.4 I went back into the config file and edited the settings back to my usual oc/voltages so that if it crashed it would apply them on reboot.

After about an hour the rig crashed (CCL error gpu 5)it rebooted and it restarted claymore with my usual settings. I noticed that the power usage at the wall had jumped up to around 1285w and HWinfo was showing gpu power usage of around 120w per card. HWinfo and GPU-z were showing that my GPU's were running at the usual clocks/voltages 1150/2200 900/950 so I stopped Claymore and manually restarted my rig. 

When it restarted the MB done the initial single beep before posting then tripple beeped and wouldn't post. I powered off the rig, 'flicked the PSU switches off and let it still for a few mins. Powered the rig back up again and same result (single beep, tripple beep not posting).
I powered off the rig unplugged GPU 0 and GPU 1 at the MB, swapped them over (gpu 0 to pcie 1x slot gpu 1 to pcie 16x slot) and the rig posted then booted to Windows went into device manager and it was only picking up 5 of the 6 GPU's (not the GPU that was previously GPU 0). 

I shut down the rig, unplugged the GPU at the riser and PSU and rebooted (posted and booted to Windows with no issues), I started Claymore (started mining fine) and when I checked HWinfo/GPU-z same result previous clock/voltage settings with higher gpu power usage (120w per card) than usual.

I stopped Claymore, shut down the rig, unplugged the rig at the wall, put GPU's in there original positions, moved the bios switch on the GPU that wasn't being picked up to 'silent mode' and booted the rig up.

It posted and booted into Windows but was running really odd, mouse movements, opening windows etc were slow and sluggish. I then moved the bios switch back to original position then used ATIflash via cmd to 're flash all cards back to original bios, uninstalled driver via DDU and rebooted the rig. Reinstalled blockchain driver, rebooted and mouse movements, opening windows etc were still slow and sluggish. 

Used ATIflash via cmd and flashed all GPU'S with bios I had them running on previously (one click Polaris Bios Editor). Applied atikmdag-patcher and rebooted, things were still running slow and sluggish, ran claymore and it started mining with all 6 GPU'S at original config setting, getting the same Mh/s but all the GPU's are still using more power than they usually would (120w per GPU and whole rig 1285w at wall).

Apologies for how long this is but I wanted to give a good idea on what is happening, what I have tried already etc.

Thoughts? Anyone had a similar issue that the have managed to resolve? Thanks in advance to anyone that can help me.

Ps

How do I put this into one of the 'box thing's' so it doesn't take up so much room on the page?

-----------------------------------
when you click on the # button you will see  "code" and "/code" - insert your text between these two and when you post it will show up the code window

regarding power- try reducing the clocks..

Quote from: big al on Today at 01:06:55 AM
This is what I've found too with dual mining.  11.3 or 11.4 you have to slow down the cards to get stability but still 4% incorrect shares.

With 11.0 I get slightly slower speeds vs the same card speed but I can turn them up more with stability. I dont have any incorrect shares so far a couple hours in.  The effective speed is much higher.

The incorrect shares is due to the speed boost claymore did, claymore increased memory intensity for the boost and now to make it right people need to reduce their memory clock speed.

[1713578908]

1713578908

[1713578908]

1713578908

[1713578908]

1713578908

[Chiko.Chavez]

Thanks cryptoillusion1, have edited it.

Re: lowering the clocks, although this will lower the power consumption it still won't address the issue that at the same clocks it is consuming more power than usual ie lower clocks will be consuming same power as the higher clocks were.

The other issue is that the original GPU 0 will only work in the 16x PCI-e slot if connected to any of the PCI-e 1x slots it fails to get picked up.

[KGV]

Hey guys just wanted to check if anyone is mining with 1080s. I have 2 x 1080s in my set-up and they both average about 21 mh/s on the claymore miner...this seems 4-5 mh/s too slow or is that just me?

Those cards should get 30 to 32 mh/s each, easy.
Obviously you don't have your system or cards configured correctly.

1080 gives 21 mh\s on ETH. TI gives 32 + up to 38 depending on card.

[janding]

Hey guys just wanted to check if anyone is mining with 1080s. I have 2 x 1080s in my set-up and they both average about 21 mh/s on the claymore miner...this seems 4-5 mh/s too slow or is that just me?

Those cards should get 30 to 32 mh/s each, easy.
Obviously you don't have your system or cards configured correctly.

1080 gives 21 mh\s on ETH. TI gives 32 + up to 38 depending on card.

I need to correct my statement. I misread it.
My RX-580's all get 30 to 32 mh/s. I don't know anything about 1080's should get.

[cryptoillusion1]

Thanks cryptoillusion1, have edited it.

Re: lowering the clocks, although this will lower the power consumption it still won't address the issue that at the same clocks it is consuming more power than usual ie lower clocks will be consuming same power as the higher clocks were.

The other issue is that the original GPU 0 will only work in the 16x PCI-e slot if connected to any of the PCI-e 1x slots it fails to get picked up.

I was wondering if the clock and settings have not changed from when you altered them - these are stored in the registry and may not have changed
Why don't you reset all the settings to the bios settings
To do this you could open radeon settings and in global settings 'reset' each card and try

GPU not getting picked up- Try (1) install another gpu on 16x -gpu0 not connected - restart (2) switch computer off- install gpu0 on 1x - start computer
if above does not help- reinstall driver

[romansteel]

Hey guys just wanted to check if anyone is mining with 1080s. I have 2 x 1080s in my set-up and they both average about 21 mh/s on the claymore miner...this seems 4-5 mh/s too slow or is that just me?

Those cards should get 30 to 32 mh/s each, easy.
Obviously you don't have your system or cards configured correctly.

1080 gives 21 mh\s on ETH. TI gives 32 + up to 38 depending on card.

Thankyou

[z0n0]

Here's a question:
If my biggest and main problem with mining is heat, whats the best way to lower the heat from GPUs?
Should I use lower clocks, is there a command line which tells the miner to mine slower/less intense...?
I'm willing to sacrifice the hash speed for lower temps.
Thanks!

I dual mine which generates a lot of heat. I find lowering the power limit has the biggest effect on lowering temperatures and power usage, while affecting hash rate the least. Next is undervolting the core and the least effect on lowering temperatures is adjusting the clock speed, which also affects the hash rate the most compared to adjusting the power limit and undervolting.

Otherwise you can also lower the temperature by increasing the airflow to dissipate the air around the rig. I use a box fan which is cheap and does a good job.

Thanks.
Is there a line in miner to lower power usage since my msi afterburner doesnt work on 6 cards rig?

[Vann]

You can adjust the power limit, undervolt, clocks and fan speed directly in the Claymore config. The README file explains the diferent configuration options. These are the settings I use for my RX 480/580 8GB cards.

https://bitcointalk.org/index.php?topic=1433925.msg32024794#msg32024794

You can also use OverdriveNtool to create profiles for your cards as explained on www.mining.help.

[Ursul0]

I need a little help guys. I have one AMD card (will be adding more) to my 7 card Nidvia rig. I only want to use Claymore for the AMD card. This is my bat file, but it's still running all the cards. Can someone tell me what I need to do to tweak it?

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

-di 0

EthDcrMiner64.exe -epool us1.ethermine.org:14444 -ewal 0xfe943a6dabf68375754d07044812f76a0dd01d64 -epsw x

your bat file contains some weird "-di 0" command that would not be executed in any way. try moving "-di 0" after the name of the application "EthDcrMiner64.exe"
Or better use config.txt to put all your params in there. Also you can use "platform" flag for the same purpose.

[Chiko.Chavez]

Thanks cryptoillusion1, have edited it.

Re: lowering the clocks, although this will lower the power consumption it still won't address the issue that at the same clocks it is consuming more power than usual ie lower clocks will be consuming same power as the higher clocks were.

The other issue is that the original GPU 0 will only work in the 16x PCI-e slot if connected to any of the PCI-e 1x slots it fails to get picked up.

I was wondering if the clock and settings have not changed from when you altered them - these are stored in the registry and may not have changed
Why don't you reset all the settings to the bios settings
To do this you could open radeon settings and in global settings 'reset' each card and try

GPU not getting picked up- Try (1) install another gpu on 16x -gpu0 not connected - restart (2) switch computer off- install gpu0 on 1x - start computer
if above does not help- reinstall driver

When I check the clocks via HWinfo/GPU-z etc it shows the clocks/voltages being what I have set them to in Claymore 900/950 1150/2200. I lowered the clocks to test if they were stuck and the clocks showed up at what I had changed them to 1100/2150. If the clocks were stuck in the registry would GPU-z etc show the stuck clocks or the clocks I had entered into the config file?

I used DDU to remove the driver, reinstalled blockchain driver. Flashed bios back to stock and the power consumption is still a lot higher than it should be. I tried flashing with a Sapphire RX 580 8gb Hynix memory 'tried and tested' bios from Anorak.tech and same result.
Way higher power consumption than usual at same clocks/voltages.

Tried powering rig down, unplugging GPU 0, plugging one of the other GPU's into the 16x slot, booting through to Windows, powering down, connecting 'GPU 0' to any of the other PCI-e slots and booting up. MB and Windows don't 'see the card.


Thanks once again for helping me with this issue, it's frustrating the heck out of me.

As a side note, the rig is running smoothly atm with GPU 0 in the PCI-e slot and claymore runs on sake clocks and voltage as it originally did with same Mh/s but still with way higher than usual power consumption.


Think I may reflash to stock bios, reinstall Windows and see if starting from 'fresh makes any difference.

[frikazoid]

Hello,

Strange behavior in the last version 11.4.
Version 11.1 share accept 50ms - 80ms
Version 11.4 share accept 120ms - 5000ms

And 11.4 show in log 0 shares 2 hours but pool stat shows 20-30 shares per hour. In this version lower amount of accepted shares.

Back to 11.1

Single card rx580.
eth-eu1.nanopool.org

[starcat102]

Highly optimized 13x RX480 rigs, custom per card bios, etc. 0 invalid shares, 24ms accept time, on eth1.ethermine.org
Was running perfeclty stable for weeks with 11.1

Witih 11.2, 11.3 and 11.4 crashes all the time.
Hash rate increases but there is no point in it if I have to lower the memclocks to keep the cards from crashing.

Back to 11.1 and stable again, relatively high stale rate (5-15%). 0 invalid shares. 24ms accept time.
Back to 10.6 and lowered stale rate even more (2%).

For me, I'll stay with 10.6
Or can you think of any cons?

[nars28]

Highly optimized 13x RX480 rigs, custom per card bios, etc. 0 invalid shares, 24ms accept time, on eth1.ethermine.org
Was running perfeclty stable for weeks with 11.1

Witih 11.2, 11.3 and 11.4 crashes all the time.
Hash rate increases but there is no point in it if I have to lower the memclocks to keep the cards from crashing.

Back to 11.1 and stable again, relatively high stale rate (5-15%). 0 invalid shares. 24ms accept time.
Back to 10.6 and lowered stale rate even more (2%).

For me, I'll stay with 10.6
Or can you think of any cons?

Try Phoenix miner, low stales, high hash, stable.

[Popaj]

With the past problem you helped me thank you. The problem was made by the windows defender.

Now I have a new question.  

My rig contains 6 x sapphire rx 580 nitro + 8gb.
They are all modular bios and almost work perfectly.
Four cards are with Hynix memory and two with samsung memory.
Card with Hynix memory give 31.3 Mh / s and  with Samsung's 27.8 Mh/s.

Is it possible to set in the Claymore that these with Samsung memory give 31.3 Mh/s?  

Thank you!

[starcat102]

Is it possible to set in the Claymore that these with Samsung memory give 31.3 Mh/s? 

No, you have to optimize their BIOS.

Look here, Hynix and Samsung covered with 580'ies:
http://1stminingrig.com/best-bios-rom-for-sapphire-rx-580-8gb-special-edition-hynix-memory-31-mhs/

[starcat102]

Try Phoenix miner, low stales, high hash, stable.

Thanks for the hint, Phoenix is only Windows, that's make it for me not an option.
Should probably check ethminer again...

[Claymore]

I see some people have problems with latest versions because they use hard OC and/or custom bioses created for old versions. They work fine with old GPU kernels, but I modified kernels in v11.2 and you need to readjust your settings.
For those who cannot do it (or don't want to do it) I can revert some changes and release a version with "old-style" binaries. So I will release v11.5 within a few hours and it will be the only change in this version, so don't upgrade if v11.4 works fine for you.

[bithead007]

https://i.imgur.com/EL9xHtp.gif

My experience with 11.4 and Stale Shares.
Arrow marks the spot where I upgraded from 11.3 to 11.4
The image is 11.3 to left of arrow and 11.4 to the right.

I would say this is a great success in reducing stale shares from 11.3 to 11.4
Also everything else is stable, running cool and hash rates are equal to or greater than all other Claymore versions.

I would say this is a complete success.

@janding:

How do you generate that nice graph and what do you look for in the logs to generate the numbers?  Thanks for any n00b advice.

CJ

[webrats]

I see some people have problems with latest versions because they use hard OC and/or custom bioses created for old versions. They work fine with old GPU kernels, but I modified kernels in v11.2 and you need to readjust your settings.
For those who cannot do it (or don't want to do it) I can revert some changes and release a version with "old-style" binaries. So I will release v11.5 within a few hours and it will be the only change in this version, so don't upgrade if v11.4 works fine for you.

this would be great thank you.

[RuMiner]

I see some people have problems with latest versions because they use hard OC and/or custom bioses created for old versions. They work fine with old GPU kernels, but I modified kernels in v11.2 and you need to readjust your settings.
For those who cannot do it (or don't want to do it) I can revert some changes and release a version with "old-style" binaries. So I will release v11.5 within a few hours and it will be the only change in this version, so don't upgrade if v11.4 works fine for you.

does it mean 11.5 will have issue with stale shares again? I like 3% stale instead of 8% for 11.1, but a couple of rigs began very rare crashing since updating to 11.2.