Bitcoin Forum

I have a Win 7 rig with two 5830s. If I mine them at the same time, it works for a random period of time and then reboots. I'm OC'ing with Trixx, heat is not an issue. 600W PSU. One card in a 16x slot one in a 1x with the cablesarus riser cable with the Molex.

I've tried different gui miners to no avail. Upgraded the drivers to 11.6b. Doesn't seem to help.

When you say it reboots do you mean it blue screens then reboots? If thats the case look at the minidump file.

Make sure you selected "Disable ULPS" Also, set your far up higher if you see temps above 74+. Any more help needed, pm me. I have time today.

I had this same issue only difference was I was using a 700W. You might be OCing to high. Some people can go over 1000 on the 5830 with stock voltage some can't. I am stable at 980 at the moment.

It didnt seem to matter how much OC'ing I was doing but I will try it again now at a lower setting.

Found the disable ULPS in the Trixx settings + set my system to not restart on failure. Will try that and then will try lowering the OC.

Thanks for the replies.

I had similar issues with 11.6b and also couldnt OC as high as with 11.6. Try 11.6 if that works better for you.

Well things seem to be stable for now... I had a duplicate worker name configured for one of these cards with my other rig. Would that cause this reboot issue? That would make sense as I don't think this was a BSOD crash.

Spoke too soon. It is not BSOD'ing that much is confirmed. Trying again with slightly lower OC settings.

Have you tried running them without the overclock? maybe it's not the overclock.

You might want to try installing GPU-Z and use the logging to file function. And see if it's overheating or it's some other instability such as malfunctioning PSU, what brand/model is it anyway?

I have been successful at 980mhz for about 48hrs now. Just a heads up.

At 950 and it is good for about 12hrs before it reboots. It is capable of 1010 when only a single card. :(

Sounds like a heat or power issue. Since a 5830 that overclocked would draw close to 200, two of them plus your base system could be pushing the output stability of your system. To verify it isn't a heat issue, try moving the riser mounted card as far away as possible and blow some extra fan at them. :D

Another possibility is, did you manage 1010 on the PCIe riser?

I doubt its a heat issue these 5830's run soooo cool (low 60s) for me.

I did the log file thing overnight with GPU-Z they didn't break 66 and 64.

After I found it rebooted this morning, I dropped the OC another 5 so now at 945/350mem on stock voltage. (I went back to stock voltage & 980 on previous attempts)  Getting 301 Mh/s

If this is good for 24hrs I think I'm going to try to OC only the card in the 1x slot just to cover that base.

What was the brand/model of your PSU again?

I will have to look it up again. I noticed the CPU is getting pegged to 100% by two phoenix process approx 50% each. Found this thread at first glance doesn't seem like there is a solution. http://forum.bitcoin.org/index.php?topic=28392.0

Well the PSU is only 500 not 600 my mistake, Its a SilverStone (silverstonetek.com) ATX 12V 2.3
High Efficiency with 80Plus certification. Model No. SST-ST50F-ES

I think that might be your problem. The ST-50EF-ES is an entry level PSU (Silverstone markets it as such) with only 408W of +12V capacity. The 5830 has similar power consumption as the 5830 and at those clocks, it's likely pushing around 180W ~ 200W per card. Add in the power required by the CPU and other components, it's looking like the PSU is the one crapping out due to the high load and thermal derating.

You could try to verify this by clocking up only ONE card to where it is fails within an hour or so, and not mining with the other card. After you've established a max clock for the card where you can last about 2hr, try it with both card running. If it's the PSU, your system should  reboot again in about half or less amount of time. Don't try for shorter test times because if it's the PSU overloading, you'll don't want the load to overshot too much too quickly in order to be safe.

The difference being, pushing the limits of an OC'ed card locks the system doesn't reboot it like this issue is causing. With no OC'ing It has been working for about 18hrs so far without rebooting itself.

That's why I suspect it's your PSU. Because an excess OC'd like you said, should just result in corrupted display or a system hang. Reboot sounds very much like a PSU issue. So the only way to find out is to reduce system load, by running one card, and clocking as high on the other as possible. If you're able to reach much higher clocks on either of the card alone for much longer than when both are running, this means you know at least it isn't because the GPU core isn't able to handle it.

Since temperatures for both cards are relatively cool even running together, that rules out overheating of the card.

So you're left with the PSU being too weak to do this for extended period.

I've been battling the same problem actually...

It may be a PSU issue for you, however mine seems to be something else entirely:

Swapped out to 750 (80%) PSU.  Random reboots persist.
Replaced both XFX 5830 with 2 Sapphire 5830. still craps out.

OC modification whatever way you want makes no difference.

Replaced the memory stick.  still random reboots.

I have 2 things left to consider:

CPU and mobo.

disconnected the power on one 5830, still reboots.

disconnected power on the other 5830, everything seems fine.

mobo is 870A-G54

Should I bother replacing the CPU (i personally think it's a waste of time) or should I consider replacing the mobo at this point?

The PCIe slot that seems to be isolated as the cause of issues is the closest one to the CPU.

Could someone give me a second opinion on this issue please?  replace mobo?

Download memtest and put your memory through the pace first.

Then download something like prime95(run multiple copies), or OCCT or basically any kind of stress test that comes with verification checks and go through the stress test.

If the CPU stress tests work fine, but craps out on any test that does a lot of communications on the bus e.g. RAM, GPU, then there is a high chance you've got crap board or crap memory.

i had issues like this as well and ended up going to a 900w psu. those 5830s are power hungry especially when over clocked. i know lots og people say 600 or 700 is enough for two or the rule of thumb for btc mining is 200w per card plus 150 for mobo, memory... but for my 6 systems i needed about 250 or more per 5830. just my 2 cents

200+200+150=550 /750 =73.33 % load on a 750W PSU.  if you have an 80% efficiency 750W PSU, it should be more that plenty.

HOWEVER!!!  if you have a 750W PSU and the 12V rail is not delivering enought amps for both cards, then it would be a problem.

I'm not sure how you were drawing 250 /card, sounds like you were overvolting and then, yes you need more juice: 250+250+150=650/750= 86.66% here you'd need a Gold cert. 80% efficiency 750W PSU at least...

Yeah this is kind of where I've been leaning because of the high CPU usage now due to the drivers. The CPU cooler is only a small fan with a moderate heat sink. The 12v rail amp idea is kind of interesting however. With only a bit of OC'ing (300Mh/s each) I can keep it stable for up to a couple of days before a reboot so I've been living with that for now. I plan to change out one of the cards with a lower power / hash one at some point anyway once I get another rig up and running. With the price ATM however it might be later rather than sooner.  :-[

I'm not sure if your 5830's are just power hungry or what but I have a rig of 4 5830's that pull a little over 900 watts and they are all overclocked to go 330 megahash a sec. I know it pulls a little over 900 because I have a kilowatt meter.

hmmm maybe its my cooling system that is pulling more than expected. I have a crapper 1000watt and it will not power power 4 5830s. (i meant 1000w not 900 above) http://www.newegg.com/Product/Product.aspx?Item=N82E16817159101, 4 80mm fans on top, one 120 pushing air from the back.

Yea that doesn't look like the greatest psu. It's also possible the extra fans are draining a lot. I'm using the aluminum case that is stickied in the hardware section and it doesn't require extra fans.

Looking at the price tag on that thing, I think it's almost safe to assume it's nowhere near being capable of 1000W. Which would easily explain why you can't use 4 cards with it.

Unless you have extremely powerful fans, they usually don't pull more than 0.2~0.3A each, so we're talking about 12~24W for 5 fans max, pretty insignificant compared to the GPU.

I'm running 4 5830s o/c to 1020/325 on that same exact psu.  Rest of the rig is minimal.  sempron 130, 1 stick 2gig ram, low power laptop drive.