Personally, I would probably end up keeping the extra motherboard, but I'm a bit of a hoarder. If you really need the cash though, then you're probably better off returning it. The decision will ultimately be yours to make.

Here would by my reasoning to hold onto it:

1) It's never a bad idea to have extra parts on hand. RMAs can take over a month and buying new parts for older hardware only gets increasingly expensive as time passes. Whenever I build a couple of new rigs, I try to keep all of their parts compatible -- ideally they'll all be identical such that if at some point in the future they start to break down I can still cobble their parts together into one good system.

2) If you decide to build a new rig in the future, you'll already have a motherboard. FM1 processors are still cheap (though who knows how much longer they'll be sold at major retailers...) and the FM1 motherboards use standard DDR3 memory (thankfully not old DDR2 -- prices are crazy!).

On the other hand, for $100 you could buy quite a few useful things. I guess this decision really hinges on how badly you need the money.

As for motherboards dying... I would say not very likely. Usually they're either dead on arrival or they're going to live for the life of your system. In some 15 years of building and maintaining systems, I've only ever had two motherboards die and I'm fairly certain that I killed both of them.

As for Corsair, good on them for taking care of you!

Probably about 90% of the systems that I build or own use Corsair power supplies and I've never had one fail. The same goes for Corsair memory and SSDs and now I'm even buying Corsair cases and keyboards... hah.

Whatever decision you make, good luck.

Ohh, good to hear it's winter there! From a mining perspective, that is...

Since your cards are not overclocked and assuming the manufacturer set them to run at a sane voltage, I'd think that you would have been ok (but pushing the limits) on a 750W PSU. Nonetheless, it seems that your particular unit may not be quite up to the task.


My best guess would be that the extra load from primecoin mining was finally enough to push things over the edge (as I understand it, that was the first time you'd run into shutdown issues?). The PSU somehow managed to support the load of primecoin + litecoin mining for a week, but I suppose that was enough to do it in. Continuously supporting a load that meets or exceeds a PSU's limits for extended periods of time will cause it to wear out exponentially faster than normal (if it doesn't outright kill it). PSU manufacturers often take for granted that most people buying their units (especially the cheaper ones) will use less than half the unit's rated capacity. For this reason they'll often build the units just good enough to say that they're rated at XXX watts, but the unit is not truly capable of delivering that kind of power continuously. Of course, on the other hand, sometimes high quality PSUs are capable of performing beyond their rated capacity, but I wouldn't recommend that! In your case I think the issue is partially the high load that CPU+GPU mining placed on the PSU and partially that the PSU itself may not be the greatest to begin with.

A random thought; another thing that may be compounding the issue is that the overcurrent protection circuit in the PSU can wear out too. Particularly if it gets tripped a lot, it can become weaker and more sensitive causing it to trigger sooner and sooner. I honestly don't know if this is playing a part at all or not -- there are really all sorts of things that could be going wrong to cause the odd issues that you're having.

At three months old, your PSU is still fairly new so maybe you can try exchanging it with the manufacturer. I'd expected you to say it was 5 years old or so. Nonetheless, if you still plan to CPU and GPU mine simultaneously, I'd recommend that you upgrade. If you do go this route, definitely find something that's at least 80+ silver rated -- the difference in efficiency will more than make up for the initial added cost within six months of mining.

Anyways, I suppose we still can't be 100% sure that the PSU is the cause yet. I definitely like to be sure before buying new parts. To clarify: when you tested with the 650W PSU, it only shutdown when you tried to run primecoin AND litecoin mining simultaneously, right? I'd guess you were probably just drawing too much from the 650W which caused it to shutdown too. If you have a friendly local computer shop you may be able to get them to bench test your system with a different PSU (say, a strong 850W+ to rule out drawing too much current) for little or no cost. If the problem still persisted then, you'd at least know for sure that the issue is not your power supply. If it wasn't the power supply my next thought would be maybe the motherboard. I would say with near certainty that the issue is not the CPU or memory and probably not the GPUs.

Good luck in getting things worked out!

If you'd rather add those settings to your batch file, append this to the end of your cgminer line:


If you want to control the temperatures for the cards individually, you can use a comma-separated list as illustrated above (i.e. --temp-target 70,76).

As to your problems with the system shutting down, it certainly sounds like your power supply is the issue. The shutdown symptoms you describe are typical of a PSU's protection circuits shutting it down (this is a good thing actually... ultra-cheap PSUs usually don't have these protections and instead simply start on fire!). You mentioned your PSU is a Cooler Master GX-750. Typically, Cooler Master makes decent power supplies... but the GX-750 looks like it may be an exception. Check out this review: http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=188

The unit was unable to continuously deliver the full 720W on the 12V rail that it's rated for. Because the PSU is so inefficient, its over-temperature protections would trigger before the unit could even reach its maximum output capability. This sounds quite like what may be occurring for you which would explain why it might seem to work sometimes and other times not so much. I would guess that the times where it successfully runs primecoin+litecoin for a short while it's not quite hot enough to trigger its over-temperature protection. After a few minutes of the load caused by primecoin+litecoin mining it finally heats up enough and trips the over-temp protection shutting the system down. Additionally, your system may be at the limits of your power supply's overcurrent protection which could be the cause of the immediate shutdowns when starting a miner. This is why you have to use the switch on the back of the power supply... to reset the protection circuit. This usually wouldn't be necessary if the problem were elsewhere (i.e. the electrical circuitry of the building, the motherboard, cpu, gpu, etc).

I suspect that there's a triad of things going wrong here that are coming together to cause your problem.

1.) What's the temperature of the room in which this mining rig is housed? Is it air-conditioned? A high ambient temperature may be contributing to this issue. Make sure that the cooling fan of the PSU is spinning up to 100%... sometimes PSUs have defective fan controllers.

2.) Are your 7970 cards overclocked? Are they undervolted at all? What about your CPU? The AMD FX-8350 can draw quite a bit of power and a lot of this power comes from the +12V rail just like for your GPUs. I would guess that you're already over 50A on the +12V rail and coming up on the 60A rating of your PSU, assuming it can even supply that much.

3.) As mentioned before, your PSU -- while adequately rated -- may not actually be providing the output that you need to run your system. If your PSU is performing like the one in the review (http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=188), it's really only giving you about ~600-650W of real output at best which is inadequate for your system (when mining primecoin+litecoin simultaneously). Add to this the fact that a power supply's output tends to degrade with age (how old is yours?) and we really start to run into issues. That said, even if your PSU is/was capable of delivering the power that your system needs, you're still running up against the red line of it's capabilities which is never a good thing.

I don't believe that the issue lies with your CPU, motherboard, or the electrical circuit to which your system is connected. Your power supply seems to be the culprit but this is just my opinion based on the information provided. The other 650W PSU that you tried testing with probably had the same problem as your 750W. It's not necessarily a bad PSU, just inadequate. Without significant undervolting, 650W (~54A or less on the +12V rail) is really pushing it for two 7970s + a monster CPU at 100%.

I always hate to suggest for someone to run out and spend money on new equipment so here's what I'd try for a solution: If possible, undervolt/downclock your 7970s a bit. I run mine at just 902Mhz core and 998Mhz memory at 980mV and still obtain 640Kh/s each. At these clocks and voltages, each card only uses ~170W-180W -- super efficient. If your cards are factory volted at 1.2V core and 1.6V memory like some... they can use almost 280W a piece with a mild overclock. Knowing this, it's easy to see how you can easily exceed the capabilities of some power supplies.

If you manage to get your cards undervolted/clocked and find that the system can then run stable while mining primecoin+litecoin, I'd look into buying a beefier, higher quality PSU. I would suggest a Corsair, XFX or Seasonic 850W unit with a silver or better efficiency rating (they can be pricey, but worth it). You probably won't need to throw away your current PSU, either; just save it for a less demanding non-mining system.

I have a rig with 3 Sapphire 7950s in it; two are the exact one that you linked: 100352-3L and one is a 100352-2L. I would say that the 3Ls you linked are good cards for the price. That said, I feel that the 2L I have gives me a good frame of reference for comparing the 3L card to others. All around, the 3Ls seem to be a cheaper-made card than the 2L but this is not necessarily a bad thing; they still work well. Compared to the 2L, you cannot adjust the memory voltage on the 3L and there are no VRM temperature sensors. The heatsink on the 3L cards does not seem to be as good (less copper) and so for any given temperature, the fans on the 3L cards run a little bit higher than those on the 2L. Cooling of the VRM components looks to be fine as they have a decent-sized aluminum heatsinks attached to them.

I get exactly the same hash rate for my 2L and 3L cards at any given setting. Their overclock/undervolt capability is nearly identical. If I recall correctly, ~1070Mhz core/ 1250Mhz memory will take them to ~625Kh/s at ~1.1V. I've pushed them above 630Kh/s but I don't mine that high as it kills efficiency. Due to power costs, I tuned my rig for maximum efficiency. Here are my current scrypt mining results:

3x Sapphire 7950 100352-2L/3L
562Kh/s per card, ~1.68Mh/s total
Windows 7 x64, Catalyst 12.8 (I tried many drivers, 12.8 seems to provide me the best results for 7950 cards)
Cgminer settings: --intensity 19 --thread-concurrency 24000 --worksize 256 -g 1 --lookup-gap 2
Core clock: 944Mhz
Memory clock: 900Mhz
Core voltage: 980mV actual (1037mV set in TriXX for 3L, 1018mV for 2L)
Core temperature: ~62-65C, 40-48% fan speed
VRM temperature: ~70-73C (as reported by the 2L, no sensors on 3L)
Room temperature: ~65-70F (~18-21C, air-conditioned)
Configuration: Milk crate rig, 20inch box fan on low setting blowing through the crate.
Total system power usage: 620W at the wall - powered by an 850W silver-rated PSU (XFX PRO850WXXX)

The system uses ~55W without any of the cards plugged in so each card uses only ~188W while mining after PSU efficiency losses. This results in about 2.72Kh/s/W compared to 2.35Kh/s/W when running the cards at 605Kh/s each or ~2.1Kh/s/W when running them at 630Kh/s+ each.

To answer your questions directly, if the cards work as well for you as they do for me I think that you will have no issues with heat, especially if you don't mind undervolting. If you run the cards in a very warm environment, or without airflow, you will definitely need to undervolt them a bit. As to the fan noise, they're quite silent under 50%. Up to about 60% they're not very noticeable. Around 70%, you can tell they're running for sure. At 80%+ they really start to pickup and can get quite loud.

When I first got these cards I had trouble getting them to change their voltage. I tried all kinds of software and did different things to get it to work. I don't remember exactly what did it but here are two things that I remember: On the top of the card is a button to switch BIOSes. I'm running the cards with the second BIOS (the button is depressed and glows blue). I also updated to Sapphire TriXX V4.6.0 and that's what I can finally use to get the core voltages adjusted. I'm not sure if the BIOS button helped or if I just needed the newest TriXX but I figured this might save you some trouble if you have the same issue.

For reference, the cards:
http://www.newegg.com/Product/Product.aspx?Item=N82E16814202026    <---100352-3L
http://www.newegg.com/Product/Product.aspx?Item=N82E16814202006    <---100352-2L

I know you didn't necessarily ask for a comparison but I thought this might be helpful to give a point of reference and maybe for someone else who already has some 2L cards and is looking to pickup some of these 3Ls too. Hopefully this helps you out.

I seem to recall that at some point last year Nvidia released a driver that was causing a pink screen issue on 5xx series cards. The pink screen would occur when starting games or watching Youtube videos. I think it was version 296.10 that caused the problem and it wasn't fixed until 301.42. Since you mentioned that you've reinstalled Windows, I assume you've installed the latest drivers too?

Some people also seem to mention that GPU monitoring/tweaking software could sometimes be the culprit. Some cases of the pink screen issue seem to be solved by uninstalling software like Lucid Virtu. I kind of doubt that's the problem here though.

Pink screens can also be caused by bad monitor cables and connections but that of course seems not to be the case here due to the timing of the issue.

When you mention the video issues, are you speaking of Youtube video? The videos usually use Flash and as a sort of temporary band-aid until you solve this issue you can disable hardware acceleration for Flash. Right-click on a Youtube video (or any other Flash window) and choose "Settings". From there click the left-most tab on the bottom and then un-check "Use hardware acceleration". That should at least allow you to use Youtube until you get to the bottom of the issue.

Wish I could offer more help. Good luck!

No problem.

As for the p2pool -> exchange question...

I believe they say this because p2pool tends pay out in a ton of small transactions. What would happen is you'd be spamming your exchange's wallet with a lot of tiny transactions which would possibly result in higher transaction fees for them when they send out payments to people withdrawing coins (this is my understanding, at least -- I'm no expert). The pools that I use to deposit directly to exchanges are non-p2pool style and I use fairly high payout thresholds.

Good luck

Thanks for the pictures. Nice rig, I like your fan setup.
My 3x 7950 system is a milk crate setup with a single 20 inch box fan blowing across it. The box fan uses 80 Watts even on the lowest setting. I'm thinking of moving to a 3x 120mm fan setup like yours to save some Watts.

That radiator for your CPU is insane! Its like something off a motorcycle or small car! I like it.

Now, if only I could work my way up to 20 rigs like you.

I do this all the time and it works fine.
As you describe, you can just generate a deposit address on Cryptsy and put it into the payout address slot for your pool. Your pool payouts will head directly to your balance on Cryptsy (aside from the lag time of the transaction being confirmed by the blockchain, of course).

Thanks for the kind words. 

The Asus power supply calculator that you linked is nice but it provides very high estimates. This is not an entirely bad thing because of course, as you pointed out, power supplies are typically most efficient around 50% load and it's always better to have a larger power supply than you need.

My motherboard and CPU are quite efficient and I have the CPU underclocked and wayyyyy undervolted. If I unplug my 7950s and plug in an old PCI videocard then boot to desktop, the system idles at only 55 Watts.


Thanks for posting this info, I was just about to ask.
Your setup seems to be running nicely. Have you tried running your memory even lower than 1250 (i.e. ~900)? I was able to drop over 40 watts by taking the memory down from 1250 -> 900 and I didn't lose any hash rate.

It sounds like you're having fun with coins and mining! That's great.
Be careful out there, gambler.

The LTC difficulty is (relatively) nice right now so it's definitely prime time for mining it.

Here's the tool I use to check my power usage: http://www.amazon.com/P3-International-P4400-Electricity-Monitor/dp/B00009MDBU

It's also possible to use an AC line splitter and a clamp-style ammeter to calculate your wattage.

Example clamp meter: http://www.amazon.com/Sinometer-UT202A-Auto-ranging-Clamp-Meter/dp/B005FSSKJA
Example AC line splitter: http://www.amazon.com/Tekpower-M920-AC-Line-Splitter/dp/B003A0MD48

There are so many factors it's really hard to say why that setup didn't work out. It's possible that the 12V rail(s) on the 630W were simply not up to the task (PSU wattage does not tell the whole tale). Antec likes to put multiple +12V rails on their PSUs. Each rail has its own OCP (over current protection) and sometimes on the lower wattage PSUs (650 and under) a single rail simply won't provide enough amperage for mining without tripping. This or any number of other things could have been causing your issue.

I looked into the Antec 850 that you have and it seems decent. It has four +12V rails which are distributed a little oddly but I doubt will cause you problems. According to this review: http://www.hardwaresecrets.com/article/Antec-TruePower-Quattro-850-W-Power-Supply-Review/555 the OCP for each +12V rail doesn't trigger until 32A (the official rating says 18A). If you decide to try three cards, I'd certainly recommend that you undervolt them if possible to reduce the total load on the PSU and also save on your power bills. The stock voltage setting for cards is usually wastefully high and can safely be backed down a bit to lower temperatures and increase efficiency. It's usually not a good idea to run your PSU at 90%+ of its rated capacity as it greatly reduces the expected lifetime of the PSU. Efficiency and voltage regulation capabilities also drop off sharply around this point.

Anyway, I really do believe that your Antec 850 will be enough still, especially if you can undervolt your cards. Also, you never mentioned what you're mining. SHA256 (Bitcoin) mining uses less power than scrypt (Litecoin) mining.

I'm currently running 3 7950s on an 850W PSU (XFX PRO850W) without issue. I did undervolt my cards quite a bit though. At the stock voltages of 1.2V on these Sapphire cards, I was easily running at 90%+ of my PSU's capacity mining Litecoin (~900W measured at the wall). By undervolting the cards significantly and lowering the memory clockspeeds, I was able to get the total system power usage down to 620W at the wall.

Current setup:
3x 7950s
570Kh/s each, 1,710Kh/s total (scrypt)
.985V core
955Mhz core
900Mhz memory EDIT: just double-checked and corrected this value.
620 watts, measured at wall (via Kill-A-Watt)

My hash rate is a bit lower than the numbers often bandied about but since my power costs are fairly high, I've realized that efficiency is more important. My cards will run up to 650Kh/s but the extra power used to get there isn't worth it anyway.

With a bit of tweaking, I think that you can make your setup work out as well. As long as you can lower voltages, your 850W will be more than enough. Good luck!

The site seems to be working again now. It was down for me too.

As to the original question of how the pool had 110% of the network's hash rate, the answer is likely very simple. arg.epools.org is by far the largest ARG pool and often it's running at 80%+ of the network's hash rate. The pool hash rate displayed is calculated by the pool itself based on the rate at which shares are being submitted to it. The global hash rate it displays is retrieved from the Argentum client itself. Due to the way that the network hash rate is calculated by the Argentum client (and this holds true for pretty much any coin based upon Bitcoin), it can at times be wildly inaccurate as it's calculated by measuring the rate at which blocks are found. Due to the element of randomness in block generation, it's possible for the network to at times generate blocks much faster or slower than would be expected for its actual hash rate. This introduces a (sometimes quite large) margin of error into the estimates of the network's hash rate. As the network grows larger, this margin of error shrinks. In the same manner, the pool's hash rate calculations can also become skewed by streaks of good or bad luck. Thus, we end up sometimes with a situation where a large pool may appear to have more than 100% of the network's hash rate. It's really just an estimation error, that's all.

I've been mining ARG and DGC at the *.epools.org pools for the past few weeks and found them to be great. They're definitely not a scam.

That's very good to hear! Congrats on getting things running smoothly.

Hmmmph. Well, I guess maybe the problem lies elsewhere then. Which drivers are you running? I'm currently running my 3x 7950 system on Catalyst 12.8 (best performance) but also had it running fine on version 13.1, 13.3beta and 13.6beta.

The 11.78V on the +12V is technically still within spec (+-5%), but sounds a bit low. It's obvious your PSU is under some major stress.
As you said it seems this indeed might not be your only problem though.

I'm going to go with blockzett on this one. Maybe 1.1V isn't enough core voltage for your cards. You could try again with only two cards, but raise their voltage up a little and see what happens. If the 1.1V that you state is inside of Sapphire Trixx, keep in mind that your true voltage will be lower. My 7950s have a vdroop of around 60mV which means that when I set them to 1100mV in Trixx, they actually run at 1040mV as measured by GPU-Z.

Other than this I'm a bit stumped now, sorry.

Ohhh, interesting!

The Corsair TX750M is a good unit. It's manufactured by CWT (a decent PSU OEM), I believe . I'm running a TX850v1 and a TX850M myself at the moment. That said, it may still be that you're overtaxing the poor thing.

You're probably right at the point before the OCP kicks in but after the +12V rail falls wildly out of spec. This can certainly cause the issue that you're experiencing (but don't take this as a 100% positive diagnosis). Again, if possible try monitoring your +12V when you start cgminer and see if it's going wild. Also, if it's not too much trouble, change nothing except remove 1 card from the system and try firing up cgminer. If it runs perfectly then I'm going to say it's time to upgrade that PSU.

I'm assuming this is your first try mining with 3 cards in this particular system? Have you ever run the system with just one or two cards? I'm kind of wondering if the issue could be power related. Is your PSU a single rail 850W+ from a good manufacturer? Grab a program like GPU-Z (http://www.techpowerup.com/downloads/2244/techpowerup-gpu-z-v0-7-1/) and watch the +12V when you start up cgminer. If it drops down to something like 11.5V or fluctuates wildly in swings of .1V+ then you have some power issues. Normally a power supply would simply shut down if you try to pull too much current from it -- that's not happening in this case. Nonetheless, sometimes PSU manufacturers skimp to save on costs and leave out OCP (over current protection) which can then lead to the kind of symptoms you're experiencing.

It may turn out that your problem is not PSU related at all, but I figured I'd throw this out there since you've tried so many other things without solution.

Good luck

I too have always had good experiences with MSI cards (just haven't mined with any yet). I have two MSI Nvidia GTX460s with the Twin Frozr coolers from my old gaming rigs. They were absolutely excellent cards. They ran super cool and it was possible to tweak voltages too. It's too bad it's not really worthwhile to mine with them.

On each individual GPU, it's only possible to mine one coin at a time. Although I haven't tried it, I believe it would be possible to mine Bitcoin with one GPU and Litecoin with another GPU in a single machine with two GPUs by running two instances of cgminer, but I don't think this is what you're looking for. There are a few coins that can be merge mined (mined simultaneously) with Bitcoin, like Namecoin, but they're not worth much.

As for recommendations on a 7950, I'd throw in a vote for Sapphire. I think the general consensus on the Sapphire cards is that they're good performers (but probably not the best), run cool and provide maximum tweakability (can be undervolted). I have 3 Sapphire 100352-3L 7950s and they've served me very well (620Kh/s each (Litecoin), 1072Mhz core, 1250Mhz memory, 1080mV core - temperatures are 64C core, 75C VRMs in an air-conditioned room). These cards have the "Dual-X" cooler with the big heatsink and dual fans. This keeps the card much cooler than the reference heatsink/fan design. As realtime pointed out, the reference design can still be useful, however, if you're trying to fit many cards directly onto a motherboard. I wouldn't recommend such a setup though -- just use (powered) risers.

Gigabyte and MSI 7950s seem to perform well too, but my understanding is that most Gigabyte cards are voltage-locked. I wish I had some experience with them to be able to tell you more, sorry. XFX cards seem to be all around bad for mining. I see so many complaints about them: too hot, low hash rate, etc. Powercolor cards seem hit or miss.

Whatever you choose, good luck!

The "connecting..." part makes it seem like there may be some connection issues. I'd definitely try another pool briefly just for testing purposes to rule that out. I don't have any experience with GUIMiner but if you decide to try another piece of software I'd certainly recommend CGMiner. One common cause of fluctuations in hash rate is the card's "power limit" setting. If it's set too low for the workload, the card will throttle back resulting in a loss of hash rate. If this is the case you should be able to see the card's clock speeds bouncing around in monitoring software. You can raise the power limit in the Catalyst Control Center under the performance tab (or in another piece of tweaking software).

I can't really offer much help beyond this as I have zero experience with 7990s, sorry.

Good luck.

You know the saying "There's an app for that"?

Well, there is indeed a guide for that!

See here: https://bitcointalk.org/index.php?topic=217636.0