Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux) - page 910.

AlainC

member

Activity: 91

Merit: 10

Hi,
I just build a Rig with 2 RX480 8G (Saphire) and an Asrock H81 pro btc v2 (4GB dram, +20GB swap), on ubuntu 16.04 desktop and amdgpu-pro-17.10-414273 (reinstalled from scratch with a clean procedure)
I don't yet have all my risers so GPU 0 is on my 16x PCIE slot.
Rom are stock (I suppose, reconditioned).
I used claymore 9.3 to mine ETH only.
I optimised ethi and dcri,
my config.txt contains

Code:

-mode 1 -asm 2 -etha 0,0 -dcri 12,12 -ethi 16,16
-tt 80 -ttli 85 -tstop 90 -fanmin 80
-etht 200 -ftime 10 -r 1 -retrydelay 5
-gser 1

and for the environment I need 64bits pointers, or else it does not work.

Code:

GPU_FORCE_64BIT_PTR=1
GPU_MAX_HEAP_SIZE=100
GPU_USE_SYNC_OBJECTS=1
GPU_MAX_ALLOC_PERCENT=100
GPU_SINGLE_ALLOC_PERCENT=100

and all seemed fine. never invalid shares, stable throughput at 24.535MH/s*2 (total >49MH/s ...). I consume 395W from the plug (MB consume about 50W).

at the beginning there was no other message than

Quote

WATCHDOG: GPU 0 hangs in OpenCL call, exit

but with time I caught kernel messages

Quote

kernel: gmc_v8_0_process_interrupt: 36 callbacks suppressed
kernel: amdgpu 0000:01:00.0: GPU fault detected: 147 0x00504802
kernel: amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x001E920A
kernel: amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x04048002
kernel: amdgpu 0000:01:00.0: VM fault (0x02, vmid 2) at page 2003466, read from 'TC4' (0x54433400) (72)
...
WATCHDOG: GPU 0 hangs in OpenCL call, exit
...
gmc_v8_0_process_interrupt: 36 callbacks suppressed
amdgpu 0000:01:00.0: GPU fault detected: 146 0x0600480c
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x001032C0
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x0804800C
amdgpu 0000:01:00.0: VM fault (0x0c, vmid 4) at page 1061568, read from 'TC4' (0x54433400) (72)
amdgpu 0000:01:00.0: GPU fault detected: 147 0x06004808
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x001032C0
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x08048008
amdgpu 0000:01:00.0: VM fault (0x08, vmid 4) at page 1061568, read from 'TC4' (0x54433400) (72)
amdgpu 0000:01:00.0: GPU fault detected: 146 0x0678040c
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x001432CF
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x0800400C
amdgpu 0000:01:00.0: VM fault (0x0c, vmid 4) at page 1323727, read from 'TC1' (0x54433100) (4)
...
amdgpu 0000:01:00.0: GPU fault detected: 147 0x06004808
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x001432C0
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x08048008
amdgpu 0000:01:00.0: VM fault (0x08, vmid 4) at page 1323712, read from 'TC4' (0x54433400) (72)
...
WATCHDOG: GPU 0 hangs in OpenCL call, exit
...
kernel: gmc_v8_0_process_interrupt: 36 callbacks suppressed
kernel: amdgpu 0000:01:00.0: GPU fault detected: 146 0x06404404
kernel: amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x001832C8
kernel: amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x04044004
kernel: amdgpu 0000:01:00.0: VM fault (0x04, vmid 2) at page 1585864, read from 'TC5' (0x54433500) (68)
kernel: amdgpu 0000:01:00.0: GPU fault detected: 146 0x0668440c
kernel: amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x001832CD
kernel: amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x0404400C
kernel: amdgpu 0000:01:00.0: VM fault (0x0c, vmid 2) at page 1585869, read from 'TC5' (0x54433500) (68)
...
WATCHDOG: GPU 0 hangs in OpenCL call, exit
...
ethdcrminer64[1166]: ETH - Total Speed: 49.078 Mh/s, Total Shares: 41, Rejected: 0, Time: 01:03
ethdcrminer64[1166]: ETH: GPU0 24.539 Mh/s, GPU1 24.539 Mh/s
ethdcrminer64[1166]: GPU0 t=77C fan=72%, GPU1 t=75C fan=72%
gmc_v8_0_process_interrupt: 36 callbacks suppressed
amdgpu 0000:01:00.0: GPU fault detected: 147 0x06004808
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x001032C0
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x04048008
amdgpu 0000:01:00.0: VM fault (0x08, vmid 2) at page 1061568, read from 'TC4' (0x54433400) (72)
amdgpu 0000:01:00.0: GPU fault detected: 146 0x0678480c
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x001832CF
...
amdgpu 0000:01:00.0: GPU fault detected: 146 0x06780804
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x001432CF
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x04008004
amdgpu 0000:01:00.0: VM fault (0x04, vmid 2) at page 1323727, read from 'TC0' (0x54433000) (8)
...
WATCHDOG: GPU 0 hangs in OpenCL call, exit
...

I have many hypothesis.

One if that it does not work well with 16x slot ? Is that a known problem?
One is that it is related to overheating of the card (I've pushed the fan to test) . Is it s usual symptom?
One is overclocking (I did not change ROM but maybe the card is modded). Is it a usual symptoom
One is my DCRI/ETHI being "abusive" for the card . Is it a typical symptom?
One is software problem. NB: I've migrated to 9.2. maybe some option to tweak? Since GPU1 is OK (maybe it does not have time to worry), I'm not convinced.

The good point is that the watchdog works well...

Claymore

donator

Activity: 1610

Merit: 1325

Miners developer

Quote from: PandaPanda! on May 01, 2017, 04:44:04 AM

@Claymore - your warning system on ethmon is not working properly. When the batch file runs, it always returns "1" for %2 variable, regardless of whether the rig is overheating, low hash, or offline - it will always return "1". Please fix this! Thanks!

Just check it, it works properly. My test:
1. Start miner and EthMan.
2. Add miner to EthMan, set zero delay for BAT file.
3. In miner properties, set target hashrate x2 more than miner produces. BAT is started immediately, sample.bat shows "Problem ID: 3".
4. Return target hashrate to zero.
5. Set target temperature as 10C. BAT is started immediately, sample.bat shows "Problem ID: 2".

If you get some different result, let me know step-by-step guide how to reproduce the issue.

PandaPanda!

newbie

Activity: 35

Merit: 0

@Claymore - your warning system on ethmon is not working properly. When the batch file runs, it always returns "1" for %2 variable, regardless of whether the rig is overheating, low hash, or offline - it will always return "1". Please fix this! Thanks!

PandaPanda!

newbie

Activity: 35

Merit: 0

Quote from: Ursul0 on April 30, 2017, 09:06:55 AM

Quote from: PandaPanda! on April 30, 2017, 06:39:03 AM

I have one rig in my farm which keeps crashing every 3 hours or so.

Usually I can open up the log file and find out which particular GPU is causing issues, and rectify the problem without much headache. But this time, i'm getting this message...

13:18:08:445   5ac   buf: {"id":4,"jsonrpc":"2.0","result":true}
13:18:08:445   5ac   parse packet: 38
13:18:08:445   5ac   ETH: Share accepted (47 ms)!
13:18:08:445   5ac   new buf size: 0
13:18:09:865   eb8   GPU0 t=70C fan=82%, GPU1 t=67C fan=78%, GPU2 t=65C fan=81%, GPU3 t=65C fan=74%, GPU4 t=71C fan=79%
13:18:09:865   eb8   em hbt: 0, fm hbt: 16,
13:18:09:865   eb8   watchdog - thread 0, hb time 250
13:18:09:865   eb8   watchdog - thread 1, hb time 32
13:18:09:865   eb8   watchdog - thread 2, hb time 32
13:18:09:865   eb8   watchdog - thread 3, hb time 19906
13:18:09:865   eb8   watchdog - thread 4, hb time 234
13:18:09:865   eb8   watchdog - thread 5, hb time 32
13:18:09:865   eb8   watchdog - thread 6, hb time 297
13:18:09:865   eb8   watchdog - thread 7, hb time 94
13:18:09:865   eb8   watchdog - thread 8, hb time 125
13:18:09:865   eb8   watchdog - thread 9, hb time 328
13:18:09:865   eb8   WATCHDOG: GPU error, you need to restart miner Sad

13:18:09:865 eb8 Rebooting

Its not specifically telling me which GPU is having issues, although i'm assuming its whatever was working on "thread 3" - how can I find out which GPU is working on thread 3? Claymore, please help! This is driving me nuts. This rig usually hangs after the "Rebooting" message.

2 threads per GPU in their order, "time 19906"

So thread 3 is GPU1?

RobertD

newbie

Activity: 42

Merit: 0

Quote from: tribrzinec on April 30, 2017, 10:39:59 PM

Hello mining community,

I would like to ask for some help. Recently bought a mining rig 6x RX 480

https://www.mindfactory.de/product_info.php/8GB-PowerColor-Radeon-RX-480-Red-Dragon-Aktiv-PCIe-3-0-x16--Retail-_1124443.html

I am using Biostar TB85 motherboard, 12 GB RAM, EVGA 1300w, 120ssd, intel celeron 2x2.8GHz, windows 10 64.

I have this problem. On minute after I start claymore miner hashrate on one card drop to 0 and I get report - GPU hangs in opencl call - Need to restart miner.

I am searching for solution for couple of days already.

- I tried different GPU drivers, if I try to install 17.3 or 17.4 crimson computer crash.

- Virtual memory is on 16GB.

- If I start a miner without this card which lost hashrate then same thing happen to other 5 cards, hashrate drop to 0 on all 5 cards.

- I reinstall windows and all drivers, still same issue

- I switch cables and cards, nothing positive happen.

- tried other miners, pools, old version of claymore miner..

- tried function -gser -eser

Tried probly everything i could find on google, still no solution. Not sure if that is specific problem of motherboard, graphic card? other parts are more or less standard mining parts and probly have nothing to do with that problem.

Hop someone can help or give any proposition what should I do or try.

Thank you

Best regards

Don't just plug in 6 cards and expect it to work. Don't randomly remove cards, windows drivers don't like it.

Test each card and pcie slot individually. That is how everyone else does it.

I'm averaging 1 faulty card per 10 new cards. Most fail within 10 days.

GabryRox

sr. member

Activity: 600

Merit: 261

Quote from: chronek on May 01, 2017, 02:22:58 AM

Nice work with 9.3 , before i could oc rx480 4GB hynix memory only 1920-1950, now without peoblems can to 2080 at least (still samsung memory better with 9.2 and 9.3 to 2150 without problem) Who would want buy 8GB cards when 4GB do the same work...

Yeah, that was my thought as well when buying 4 rigs worth of 470 4GB over the past 6 months. However, I have heard that the potential life is likely longer on the 8GB variety due to ever-expanding difficulties of mining algos, etc. While that in theory may have some validity, I doubt the 4GB versions will be ineffective any time soon. I fully expect them to very productive at least through the end of 2017 and hopefully beyond. Personally, I think that ETH will go POC before the 4GB 400 series becomes ineffective in mining it, so i am happy with my 4gb versions that I got for much cheaper than the 8gb variety.

GabryRox

sr. member

Activity: 600

Merit: 261

Quote from: rawbot on April 30, 2017, 07:00:09 PM

Quote from: Elder III on April 30, 2017, 05:52:57 PM

Quote from: notbatman on April 30, 2017, 01:12:44 PM

An RX460 4GB OC will be arriving soon, I'm betting +20MH/s on this $150 card. We'll see...

Those 1080's are like $1k?

In the USA at least the GTX 1080 can be had for under $500 and the GTX 1080 Ti is $700. I'm sure the 1080 Ti is $1000 or more in some countries that get hit hard with import taxes on PC hardware. Sad

In the UK the outrageous Brexit vote means that used b grade RX470 GPU's are now £30 more expensive than what we paid brand new for them in November. But hey, at least we didn't elect an sociopathic orangutan as POTUS lol.

Hate to break it to you but that price rise in 470's (even used ones) probably has very little (if anything) to do with brexit. In fact, prices of both 470 & 480 have gone through the roof due to 1) phase out preparing for release of 500 series and 2) huge spike in demand for good mining GPUs since the huge spike in alt coin prices about 6 weeks ago. These 2 factors, combined with the fact that the early reviews on the 500 series GPUs are far from glowing (about same hash-rate with more power draw and big issues getting more than 4 GPUs to work per rig), mean that demand for 470/480 is probably near an all-time high, but of course, there is now very limited supply (esp new ones), so as a result, prices of sky-rocketed. Indeed, I tracked a few sales of used 470s on ebay recently and they went for about $20-$30 more than I paid for new ones about 4-5 weeks ago, just before this huge shortage began. The new ones are even tougher to get... many ebay and other 3rd party sellers are at least asking close to $300 already, and it seems likely that will go up given the issues with the 500 series. Even though I got a hold of enough 470s to fully stock my four 6-GPU rigs, I[m beginning to wish I would have scooped up a couple dozen for around the $160-$170 each they were only a couple months ago. Live and learn I guess...

chronek

sr. member

Activity: 273

Merit: 250

BD People Are Legend

Nice work with 9.3 , before i could oc rx480 4GB hynix memory only 1920-1950, now without peoblems can to 2080 at least (still samsung memory better with 9.2 and 9.3 to 2150 without problem) Who would want buy 8GB cards when 4GB do the same work...

dogjunior

sr. member

Activity: 407

Merit: 255

Quote from: tribrzinec on April 30, 2017, 10:39:59 PM

Hello mining community,

I would like to ask for some help. Recently bought a mining rig 6x RX 480

https://www.mindfactory.de/product_info.php/8GB-PowerColor-Radeon-RX-480-Red-Dragon-Aktiv-PCIe-3-0-x16--Retail-_1124443.html

I am using Biostar TB85 motherboard, 12 GB RAM, EVGA 1300w, 120ssd, intel celeron 2x2.8GHz, windows 10 64.

I have this problem. On minute after I start claymore miner hashrate on one card drop to 0 and I get report - GPU hangs in opencl call - Need to restart miner.

I am searching for solution for couple of days already.

- I tried different GPU drivers, if I try to install 17.3 or 17.4 crimson computer crash.

- Virtual memory is on 16GB.

- If I start a miner without this card which lost hashrate then same thing happen to other 5 cards, hashrate drop to 0 on all 5 cards.

- I reinstall windows and all drivers, still same issue

- I switch cables and cards, nothing positive happen.

- tried other miners, pools, old version of claymore miner..

- tried function -gser -eser

Tried probly everything i could find on google, still no solution. Not sure if that is specific problem of motherboard, graphic card? other parts are more or less standard mining parts and probly have nothing to do with that problem.

Hop someone can help or give any proposition what should I do or try.

Thank you

Best regards

Have you tried bring up the cards one at a time in windows. Also depending on the alt coin you use crimson 15.12 for eth or for zcash crimson 16.12.2. Bring up the cards one at a time until you have all 6 going.

FFI2013

hero member

Activity: 906

Merit: 507

Quote from: cryptominer420 on April 30, 2017, 10:29:10 AM

Quote from: FFI2013 on April 30, 2017, 09:13:08 AM

which drivers work well with r9 and rx cards on my mixed card rig I have a 270 that no longer mines eth and was going to replace it with my odd ball rx470

I am using 16.11.5 I have a mixed rig with RX 470's (4GB Cards) and R9 270X (2GB cards) and when setup this way I am still getting 15MH/s out of the 270X's but without 1 470 per 270X the 270X's won't mine. My guess is that the Dag is being shared from the 4GB cards.

Thanks I'll have to try it out I have gigabyte 270s and these have been running solid since LTC

alucard20724

sr. member

Activity: 703

Merit: 272

Quote from: RedRockMining on April 30, 2017, 11:00:09 PM

Quote from: alucard20724 on April 30, 2017, 10:43:02 PM

make sure your RX 580 is running at the correct clocks. It sounds like it's underclocked. this will happen sometimes if you just install a new card without uninstalling and reinstalling the drivers. DDU FTW Wink

OMG, that worked! Thank you! I had MSI Afterburner running but I was only using it to undervolt and never changed the clock speed but it was not at the default speed for the 580. Back up to speed now!

Thanks for the quick assist. Much appreciated!

i only know cause i speak from experience... Grin

RedRockMining

member

Activity: 127

Merit: 10

make sure your RX 580 is running at the correct clocks. It sounds like it's underclocked. this will happen sometimes if you just install a new card without uninstalling and reinstalling the drivers. DDU FTW Wink

[/quote]

OMG, that worked! Thank you! I had MSI Afterburner running but I was only using it to undervolt and never changed the clock speed but it was not at the default speed for the 580. Back up to speed now!

Thanks for the quick assist. Much appreciated!

alucard20724

sr. member

Activity: 703

Merit: 272

Quote from: RedRockMining on April 30, 2017, 10:23:36 PM

Claymore v9.3 + RX 580 + AMD Chipset and Graphics Drivers from 4/27 + latest Asrock firmware + W10 = only 14 Mh/s.

On the same system 390s ran at 28Mh/s and 480s at 24Mh/s.

Why the speed drop? Anyone else seeing this?

make sure your RX 580 is running at the correct clocks. It sounds like it's underclocked. this will happen sometimes if you just install a new card without uninstalling and reinstalling the drivers. DDU FTW Wink

tribrzinec

newbie

Activity: 11

Merit: 0

Hello mining community,

I would like to ask for some help. Recently bought a mining rig 6x RX 480

https://www.mindfactory.de/product_info.php/8GB-PowerColor-Radeon-RX-480-Red-Dragon-Aktiv-PCIe-3-0-x16--Retail-_1124443.html

I am using Biostar TB85 motherboard, 12 GB RAM, EVGA 1300w, 120ssd, intel celeron 2x2.8GHz, windows 10 64.

I have this problem. On minute after I start claymore miner hashrate on one card drop to 0 and I get report - GPU hangs in opencl call - Need to restart miner.

I am searching for solution for couple of days already.

- I tried different GPU drivers, if I try to install 17.3 or 17.4 crimson computer crash.

- Virtual memory is on 16GB.

- If I start a miner without this card which lost hashrate then same thing happen to other 5 cards, hashrate drop to 0 on all 5 cards.

- I reinstall windows and all drivers, still same issue

- I switch cables and cards, nothing positive happen.

- tried other miners, pools, old version of claymore miner..

- tried function -gser -eser

Tried probly everything i could find on google, still no solution. Not sure if that is specific problem of motherboard, graphic card? other parts are more or less standard mining parts and probly have nothing to do with that problem.

Hop someone can help or give any proposition what should I do or try.

Thank you

Best regards

RedRockMining

member

Activity: 127

Merit: 10

Claymore v9.3 + RX 580 + AMD Chipset and Graphics Drivers from 4/27 + latest Asrock firmware + W10 = only 14 Mh/s.

On the same system 390s ran at 28Mh/s and 480s at 24Mh/s.

Why the speed drop? Anyone else seeing this?

Hypernova

newbie

Activity: 3

Merit: 0

Quote from: Junkey on April 30, 2017, 07:04:16 PM

So I just switched from ubuntu to win10. I am getting one card drop down to 4mh. From what I recall reading it's because there is no hdmi connection plugged into it. What is the fix for that on a system you want to remain headless?

Ebay for HDMI dummy plugs.

Junkey

full member

Activity: 164

Merit: 100

So I just switched from ubuntu to win10. I am getting one card drop down to 4mh. From what I recall reading it's because there is no hdmi connection plugged into it. What is the fix for that on a system you want to remain headless?

rawbot

sr. member

Activity: 458

Merit: 250

Quote from: Elder III on April 30, 2017, 05:52:57 PM

Quote from: notbatman on April 30, 2017, 01:12:44 PM

An RX460 4GB OC will be arriving soon, I'm betting +20MH/s on this $150 card. We'll see...

Those 1080's are like $1k?

In the USA at least the GTX 1080 can be had for under $500 and the GTX 1080 Ti is $700. I'm sure the 1080 Ti is $1000 or more in some countries that get hit hard with import taxes on PC hardware. Sad

In the UK the outrageous Brexit vote means that used b grade RX470 GPU's are now £30 more expensive than what we paid brand new for them in November. But hey, at least we didn't elect an sociopathic orangutan as POTUS lol.

Elder III

sr. member

Activity: 1246

Merit: 274

Quote from: notbatman on April 30, 2017, 01:12:44 PM

An RX460 4GB OC will be arriving soon, I'm betting +20MH/s on this $150 card. We'll see...

Those 1080's are like $1k?

In the USA at least the GTX 1080 can be had for under $500 and the GTX 1080 Ti is $700. I'm sure the 1080 Ti is $1000 or more in some countries that get hit hard with import taxes on PC hardware. Sad

Wicard

newbie

Activity: 54

Merit: 0

Quote from: kostickole92 on April 30, 2017, 02:31:38 PM

Quote from: Wicard on April 29, 2017, 09:33:02 PM

@Claymore,

Thanks for the hard work, I did got increase on my miner, 2 x Sapphire RX480 Nitro+ 8Gb Dual mining,
Before ETC+DCR = 60.9 / 1645 MH/s
Now ETC+DCR = 61.140 / 1711

Still fluctuating but I just started mining on 9.3, will check it later, need to go sleep, time now 03:31am South africa

My settings same as 9.2 except -dcri from 27 to 28
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 -ethi 16 -dcri 28 -epool eu1-etc.ethermine.org:14444 -ewal "Wallet.Name" -epsw x -allpools 1 -mode 0 -dpool dcr-eu.coinmine.pl:2222 -dwal "User.Name" -dpsw x -asm 1

Keep the good stuff going!

what is your overcloack settings for this hash ?

Overclock settings
Core Voltage = -48
Power Limit = +25
Core Clock = 1140
Memory Clock = 2220

Temp +-66 degree C

http://funkyimg.com/i/2stpV.png

Topic: Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux) - page 910. (Read 6590565 times)