Topic: Mining Problems Causing Great Stress Plz Help (Read 1387 times)

I haven't had any BSOD yet (touch wood) and I can run intensity at 20 off the bat, although I didnt at first, just incase it gave me problems.  My original problem is kind of sorted now, i'd like to get 900kh/s but at the minute i'm just trying to get a stable 830 across the board.  On launching the miner i can achieve 830 on each card but it soon drops due to CGWatcher (or something else) dropping my clock speed on the middle mounted card.  I just wondering how i can "lock" in the core clock speeds so they dont change if possible.  So far 933 has worked good for core clock speed and memory at 1500.  Once i get that stable, i'll mess around with it some more, hopefully achieving 900+

I can't speak to Gigabyte R9 290x's, but I originally had an MSI R9 290x and I couldn't get the thing stable above intensity 13.  It kept crashing my drivers, hard hanging, and BSODing my computer to death.  The only way I managed to get it hashing at higher intensities was to increase the intensity slowly and incrementally.  This was happening even with completely stock settings, a proper driver installation, and a very clean computer, and it was frustrating the hell out of me.

I ended up swapping the MSI brand card for a Sapphire, and the Sapphire runs so nice that I bought several more the next day.  Maybe you'd have better luck with a Sapphire, but who knows?

However, I should also mention that one of my R9 290x rigs seems to prefer GUIminer Scrypt while another prefers CGminer.  The nice thing about the latest version of GUIminer Scrypt is that it has an r9 290x high-intensity preset that works very well for me.  I use it for a 3 x R9 290x rig along with --auto-fan and --gpu-memclock=1500 commands running in three separate instances, one for each card.  I'm averaging a solid 900-905 khash/s average between the cards with fans running at about 85% and the cards sitting at 81C, 85C, and 74C respectively.

How do i stop CGWatcher from underclocking my cards?  I have them at 933 core speed, leave it for 10 minutes and i come back to a core speed thats jumping around, i assume its for temperature control reasons however, it doesnt lower the temperature so I dont see the point and I'd like to figure out how to stop it if possible.  I want them locked to 933mhz, any ideas?

Your setup gave me 750 on all 3 cards, what are you getting?  Also setting the power to -37mv, when i click apply it would set it straight to -31mv.  What arguments are you using in your config/batch file?

My R9 290 5GPU rig (non-x) likes following settings in Afterburner:
-37mV, -3% Powertune and stock clocks (948mhz core, 1250 mem).

Don't be afraid to fiddle with the settings. Just make sure you don't press "save" after "apply" unless you're confident it is stable. Otherwise you might have to boot in safe mode or try to switch GPU settings in AB before they're applied on startup.

With my tinkering, I was able to save 3-5% power and increase hashing by 10%.

Ok i set the core clock in afterburner to 933 with mem clock to 1500 and im getting a pretty stable 830kh/s across all 3 cards, its getting better at least, im entering the sweet spot

it really confuses me how the cards temperature goes up when the mining stops, no load on the cards and the temp rises slowly wtf?

Look here: https://litecointalk.org/index.php?topic=12830.0

ok i cleared out the drivers, downloaded and installed cat 13.11 drivers and ran i again, finally im back to where i started.
card 1 (display) 750k (77oc)
card 2 750k (86oC)
card 3 800k (84oC)
gpu clock is 1040, mem 1500 however cgwatcher has underclocked my gpu on 2 cards due to temps im guessing
Card 1 has a core of 1040
Card 2 has a core of 756
Card 3 has a core of 833
All running mem clock at 1500

How do i tweak this further? i mean i might as well got 280x's at this point of performance.

Some people report bad performance with 14.1 drivers. Downgrade to 13.12.

Then, chopping off that CPU down to 1 core is bad idea. Keep 2 and use Cool-And-Quiet and other power saving options.

--temp-target 78 --gpu-fan 60-85

Also, take a look at the core clock speed. Can't be too high. Perhaps --gpu-engine 1000 --gpu-memclock 1500

Wonder if its worth removing catalyst control centre (the software) and just having the drivers?

Its an AMD 8 core (8320 FX), i have underclocked it to 1 core in bios for less power usage.

What is the CPU? Some cheap Intel CPUs don't support PCI-3.0 fully and seem to handicap the GPU or databus.

He used XFX and you use Gigabyte. I know it seems minor but yeah small differences between equipment can effect scrypt results.

Tuning scrypt is a bitch. Follow a tuning guide, nobody else's scrypt settings ever worked for me I always had to tune my own.

The manufacturer drivers can work far better than AMD's generic ones, especially if it's a non-reference model.

Hi, I desperately need help with my mining setup as im pulling my hair out here...

So I had a Rad 6950 that I was using for mining, got it working with no issues so I thought i'd build a dedicated mining rig.

So I bought:
Gigabyte GA-990FXA-UD7 Mobo
Thermaltake 1475w XT Gold PSU
and 3x Gigabyte Radeon R9 290X OC WindForce 4mb (GV-R929XOC-4GD)
16gb RAM
running on win7 x64

Got everything running with a little bit of tweaking and a guide that was very useful here
http://www.thecryptomaniac.com/2013/12/mining-with-radeon-r9-290x-how-to-get.html
But as described in that article, 930mh/s-- I was not getting anywhere near that, infact the best I could get was:
Card 1:- 72oC (display) 740kh/s
Card 2:- 84oC 810kh/s
Card 3:- 84oC 780kh/s
The only real difference was that I am using catalyst drivers 14.1 beta because when I used other drivers, my screen would have vertical coloured lines once I excited mining, i had to clean out the drivers and reinstall to remove the lines which dont reoccur with the 14.1 drivers.

I left it running for 24hrs and had no stability issues so I decided to try and tweak a few settings... BIG MISTAKE.

Now i cannot achieve anything over 500kh/s, i just get a multiple of random problems and its driving me mad because the logic behind it is nowhere to be found.  I cant even go back to my original settings because, i just tried it and the best i could get was around 300kh/s.  It gets worse every time i try something new.

I also have a problem with the gfx cards continuing to heat up once mining has stopped, even restarting the PC, leaving it off for a few minutes and then powering it up, the temp does not idle below 60-80oC when it was idling at 41oC when I first setup the system, there is no load on the cards so I do not understand that at all.

I'm wondering if multiple overclocking programs such as catalyst control centre, msi afterburner, gigabyte oc-guru are somehow causing problems?

My arguments in CGWatcher are:
--scrypt -o -u -p --gpu-platform 0 --worksize 256 --vectors 1 --intensity 20 --gpu-threads 1 --log 3 --thread-concurrency 33792 --lookup-gap 2 --api-port 4028 --expiry 120 --queue 1 --scan-time 60 --temp-hysteresis 3 --temp-cutoff 95 --temp-overheat 90 --temp-target 85 --gpu-powertune 20

Anybody got any ideas how I can fix these problems and get performance around 900kh/s per card like others have reported on here?

Thanks in advance.