Bitcoin Forum

Hello,

my rig freezes because of the new clay with 4 pcs rx480, win10 system.
The cards are not overheating.
Everything is on basic settings no tuning (no undervolting) or bios modding.

Settings (GPU_MAX_ALLOC_PERCENT), virtual memory is ok.
I tried with the older driver as well but it freezes after 15 mins again. I decreased the intensity to 4 but i got the same results.
It works perfectly with the version 8.
Do you guys have any idea what should I do?

Thanks!

Is your PSU okay ? I had such problems with an "el-cheapo" PSU which was overloaded once the cards where working optimized and under full load

I use this:  CHIEFTEC A80 750W PSU.
One card consumes 120w.

I'm having the same issues with two seperate SeaSonic X-850 units. I guess just an issue with the new version.

This is likely due to Version 8.0 had a default intensity of 4. 

Version 9.0 now has a default intensity of 6.

It crashes some of my machines so i set intensity to 5 or 4.

Some I leave at 6.

Keep in mind intensity 6 will draw more power as well.

I already tried with the settings below:
-i 4,4,4,4 and it still freezing after 15 mins.
 

I never heard of that brand of PSU.
As for the other fellow, I've had Seasonic pci-e plugs melt a few times on the psu modular connectors and on the video card end.
IMO, they don't have the best wire thickness.


Is your Windows 10 a clean install or is it an upgrade over previous OS?  I've found many errors due to this.  I goto recovery, advanced, and download the clean install tool, then reinstall.
I then only install amd driver from 9/16 - 9/20 without the amd management stuff.
Then i install msi afterburner.
Things are so much simpler than wattman now.

One of my  6 x rx 480 -8gb rigs crashes or freezes if i set intensity to 6.  It uses 2 x EVGA 650W gold psu to power the computer + 6 cards.
If I undervolt it too much it also freezes or crashes.  This does not occur on my other 6 rigs.


Its also possible that your have too much power being used by a single rail on the PSU.

What is your miner config?

Mine is


If you do -i 3, does your system crash?
I don't think you need to repeat the number if you want to set it for all cards.

I use this:  CHIEFTEC A80 750W PSU.
The win10 was a clean install and I use the afterburner too.
At the moment there is two videocard PSU connectors are divided to four so one rail us divided to two way.
I have not tried the "i -3" but I will, thank you for the advice!

My config:
-zpool eu1-zcash.flypool.org:3333
-zwal
-zpsw z
-i 4,4,4,4
-r 1

And settings:
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

Try to mine on 2-3 cards as a test. Perhaps it's PSU, if you see that GPU takes 120W, it takes it in average. Peak wattage can be higher and if several cards have a peak at same moment, it can cause such problems.

Not to say they wont shut down but i have 9 rigs running mostly 380x with a few 290 and 390 scattered around and running (i 8) on windows 10 and latest drivers with no issues as of yet. 380x cards are tuned to 850/1500 -50v running @ 166 sol/s and seem to be fine. 290 and 390 are stock. I run Lepa 1600w power supplies on all my rigs. A bit more expensive but shouldn't ever need more power.

The same problem with one my rig - Asrock H61 BTC Pro MB and four Gigabyte R9 280X. PSU best quality Corsair AX860.
V8 no problem, V9 - Win crashed after aprox. 30-60min. Intensity -i 0

same problem  on all my R9 280x rigs. All rigs freeze after some time in version 9.
GPU Asus, MSI, Gigabyte. MB Asrock H61 and H81 BTC Pro. Win10 Pro
In V8 all ok. All lowest intensity -i 0

I have the exact same problem on some of the 79xx and 280X series cards.

I just finished upgrade, only 2 rigs out of 94 freezed. I was doing it remotely so I have no idea what happened. I use -i 2 on all rigs, and on V8 both freezed rigs were working perfectly. Farm is 30 mins to drive ao I'll have to get there tomorrow, untill then I'm just guessing whats wrong..

try with -nofee 1 as your first command line argument.

it leaves out some optimizations that draw extra power and workload from the GPU.

you may well get to keep a higher intensity to make up for lack of features that crash your card anyways.

V9 is simply a little bit harder on GPUs. You need to find which GPU(s) is crashing and increase it's voltage slightly, or decrease it's clock speed. I had to adjust about 10 card's voltages across my 100+ GPU farm to get V9 stable. V8 had been stable as a rock since it's release. Now V9 is too with slightly increased voltages.

Not sure if it is psu CPU ram or voltage settings.

I am now making sure  all my rigs have the same cards no mix and match.

Most are working well. But some of my four card rigs now freeze.

If I pull the fourth card the rig works.

I,can get around this with three cards.

But tomorrow I will take my four,card msi rx 470 rigs and  play with voltage settings

Sharing experiences. The system turns off after the run 15m. Remove Antivirus, and after the card overclocking closed window of afterburner. Launched miner and closed the window. After 15 minutes, there is a window of unknown origin, and immediately closes. Mining after there in the normal mode.

Morning,
alrady 6 out of 94 crashed, getting my self ready to get to the farm da seen whats wrong

You will have spike far above the average consumption. Your psu is very good and have some limitation on max current available as a security. With the spikes and an highly optimized miner, you will probably go over the limitation and the psu stop.

You need to have a good margin on the psu side, like 30% above the average consumption to be OK. Good psu have some overhead margin too (like 10%): the whole combined should prevent any problems.

Look there : http://www.tomshardware.com/reviews/amd-radeon-rx-480-polaris-10,4616-9.html
During a gaming session you may have spikes at nearly 300W for a single card...

Just follow the main claymore thread and you will know why.

You need to reduce the clock frequency for the crashed ones. That is what I did, they are working now, and hash more.

I'll try that as soon as get there, but monitoring would harb because most of them crashed after ~6 hrs of work.
What cards are you using and on what clock speed ?

Check not average, but MAX power spending
You will be surprised when you will see 250W

seems like if x amount of cards submit a share at the same time, the power spike caused can make your psu protection kick in and shut off your rig.

What will be the sollution for that ? I also wonder why only on this last version, on v8 they were working fine

It's already been hinted at. As the mining software for Zcash nears its peak optimization the GPUs are being tasked harder and harder. At first with the initial Zcash releases the GPUs would spend very little time doing computations, that's why you could push them hard and they would still use very little power as it was like asking asking a math major to do a 2+2 calculation once every 10 minutes. Now as the optimizations roll in the GPUs are actually being taxed and the power requirements are increasing to meet this need, as well as they might start crashing when pushed to limits they previously could handle. Again asking a math major to calculate the circumference of the earth is straight-forward, but ask them to compute and sum the circumference of all the objects in the Milky-way and it gets a bit more challenging.

http://pasteboard.co/7lPE1DwKb.png 

-This is 6 380x Not the 4xx everyone is having problems. Just posting my results.


I have one of 9 down this morning but im not 100 percent sure that i didnt cause the issue. I just forgot to check after making some changes to hardware. Been up over 22 hours still using -i 8 as well.

There might be spikes, but it should be a problem for standard design cards. However, I am also experiencing this problem although I own RX480 ASUS Strix version using 8 pin power cable instead of 6 which is shipped on the standard design, and I also have a decent PSU which is Corsair 1200W (with gold certification) nevertheless, problem persists. Also -nofee 1 option is not helpful.

I have turned back to v8, and my system crashed again after 5 min. however, it was very stable on v8 till this time.  ???

You will have to reduce the frequency of the core. The cores are stressed more. So it is quite unstalbe at present.

How many rx480 card do you use in the rig?

just 2.

It's interesting. Can you attach please a gpuz screenshot (with the max "GPU only power draw")?

same problem.

like a cancer.

Here's today's testing results.
All my cards are RX480. After updating them to V9 couple rigs stopped working. All errors were the same:
except the one:
but first things firs, after playing around with clocks and memory I got the speed ~220h/s but card were overheating, so I resetted wattman settings, now they hash at ~182h/s but fans and temps are stable, but they were still freezing, then I switched to stock bios, after that everything is working fine. Lets see what will happen, I hope stock bios and stable memory/clock will solve the problem.
also wanted to mention 3x RX480 8G - 330watts from the wall on V8 and 345watts from the wall on V9

9 has some serious stability problems.

oh god. 8 was so much better than this. what happened?

I was experiencing same issues with V8 but with less rigs in total. V7 was way stable than v8 and v9

this actually helps

even a broken clock is correct two minutes out of every day :)

referring to the forums signal to noise ratio more rather than Preludes track record of course..

I'd recommend being very cautious when increasing voltage with V9, given the increased power use from spikes.

Same settings V8 vs V9 on my Fury Nitro (1140/550, -12mv), GPU CORE POWER:

V8: AVG 207W; MAX 274W

V9: AVG 210W; MAX 370W

Only a 3W difference in AVG, but a nearly 100W change in MAX. Big difference to begin with, and enormous in an multi-gpu setup.

(As it stands, undervolting to -96mv and reducing clock to 1110mhz saves me 40W on average V9 vs V8, but the MAX is higher in V9 even when significantly undervolting from V8 (283W vs 280W))

Try mining on pools that support low difficult,

You wont have your cards at full usage at once on high difficult shares,
since low dif you grab and found

You wont have that hardcore mining 100% at time

Ex:

High Dif - 3-5 minutes intensive mining for 1 share - so you have all of your Gpu Full of job for 3-5 minute unstopable

Low Dif  5-15 seconds quick mining for 1 share - Grab and Send, you wont have full usage 100% of time

It is just a theory but i think it is valid

 V9 has some MAJOR stability issues for at least some configurations.

 I tried lowering clocks on the 7870 test rig I DOCUMENTED my issues with during fairly extensive testing in the main CLaymore V9 thread - doing so made NO change even when I got the card down to about 200 Mhz UNDERclock on core and 300 on mem (afterburner limits for that card) the ONLY change was that the hashrates reported got lower than what I reported in that thread.

 I can't increase voltages as they're locked on that card (and they're at stock anyway).


 Same card was highly stable back in the days I used it on Scrypt mining, and for all the MooWrapper/Dnet work I did between it's Scrypt days and it's ETH/ZEC days - both of those drove that card a LOT harder than ETH or ZEC do (as measured by the MUCH HIGHER POWER DRAW and higher operating temp at higher fan settings in both cases).


 I've never played with the BIOS on that card in any way, so it's not a "bios issue".

Intensity 2 it is very stable for me on rig that are freezing

has v9 killed anyones card or rendered them temporarily (a few reboots) comatose?

are the spikes in power sync'd with the "donation" portion of the mining?

I'm running 6 cards (r9 380) with zalman 1250w platinum and other 3 cards ( r9 280x, 280 and 7950) with super flower 850w ( golden green)  with no problems.

None of my cards were killed.