Bitcoin Forum

I've 4x 5830's running guiminer -v -w256

I used the AMD utility to set GPU at 900mhz and clock at 900.

These are the highest lowest settings that utility is allowing.

However, it did increase my first GPU to about 285Mh/s... the other three stayed around 244Mhz. How do I setup the 3 that are lagging or is this normal?

Also see the temp at about 86C, most temps I've seen are lower.

Thanks for any help advice...

I would suggest using ATI Tray Tools.

950 core/300 mem w/ default voltage.  Fans at 70-75%.  Ideally want temperature below 85c

Or use cgminer with the core/mem settings and auto-gpu & auto-fan :)

Thanks for the quick reply. I'll try the tray tools and see how that works..

I tried Tray Tools and MSI but still seems only to be overclocking the first GPU...

Am I just not looking in the right place for the "switch"

Thanks

EDIT: Uninstalled AMD Vision Control and tried MSI afterburner again.

4x 5830 = 1.080

900 on the GPUs and 600 Clock. (Still playing with these)

0-0 = 285 Mh/s
0-1 = 243 Mh/s (? - this seems to be the 'default' and does not change with the others)
0-2 = 273 Mh/s
0-3 = 274Mh/s

Maybe CGMiner(https://bitcointalk.org/index.php?topic=28402.0 (https://bitcointalk.org/index.php?topic=28402.0)) is of some help ? It can set the over-/under-clocking and volting from CLI and works like a charm :) You don't need anything else on Windows. (I was using MSI Afterburner combined with AMD's program as well and both have succesfully been replaced by CGMiner)

I would be very uncomfortable running my cards at 85+C 24/7. It wont instantly kill the cards, but electromigration may wear them out quite fast if you are unlucky. Everyone has their personal limit, I know Im anal about temps, but 70C is my feel good limit and Im no where near that. My 5850s are ~60C @ 900 MHz, my 5870 is even sub 50C (though admittedly, because of an absolute overkill Spitfire cooler)

https://i.imgur.com/NLlpX.png

All of them are dead silent. (5850 twin frozr stock cooler @45%, the other 5850 has a 6 euro Battle-axe cooler).

Even more importantly, run GPU-Z and have  a look at your VRM temperatures under load. Depending what kind of VRMs and VRM cooling is used, they can be as high as 100+C or as low at 60C. Not all cards support VRM monitoring, but definitely check if yours do, because overheating VRMs are probably the number 1 cause of dead gpu's.

If the card does not support VRM temp monitoring, that means a non reference design, and Im under the impression most non reference cards have VRMs dont overheat as easily a reference cards, but thats just an impression I have. If you want to be sure, consider buying an IR thermometer.

I give a hearty +1 to cgminer.  I had all kinds of weird glitches and "fun" using AB.  Granted I was using 8 GPU in one system so likely that is outside the norm but still.

Since using cgminer I don't use anything else except maybe GPU-Z to check up on VRM temps occasionally.

I would recommend using dedicated linux distro for your 4 cards. No cpu bug and manual parameters for each card with easy monitoring via ssh. I remember my problems with only two 5850 cards and msi afterburner or trixx. Never was sure about overclocking and fan speeds, but now with linux my rig is working like swiss clock with even more oc and agression parameters.

Intentionally left distro name out, you can try either BAMT or LinuxCoin after reading forums, and donate to author if you're happy with it.

I'm sure you have other pc for your personal needs, dedicated rig deserves dedicated software.

Much good info, thanks for the support!

I did start up cgminer: Its running about 1170 Mh/s for the four cards at arsbitcoin:

cgminer -o http://pool:port -u (X4) -p (X4)  -I 9 --auto-fan --auto-gpu --gpu-engine 965,965,940,965 --gpu-memclock 300

getting about 313,255,283,302 Mh/s respectively...

Best rate thus far... by far.

And it's running about 70-75 temp.

win7 64 bit.