Aries, out of curiosity, what sizes of PSUs were you running before and after on this rig?  Curious because assuming I can get 6 GPUs running this weekend on rig #2 and might switch over to ZEC.  I have 1200W so I doubt it will be an issue as I should still have at least 25-30% headroom, but I know that since v9 there has been power spike that cause issues for some with either low quality PSUs or not enough headroom

Thanks to all for your advice on my issues. I had somewhat of a breakthrough last night in regard to getting a 6th GPU working on my ASRock BTC Pro R2 Mobo.

I realized that I had not tried plugging it in (to that pesky x16 slot) since I ran that 6xGPU registry mod, so I figured I would give it a try.  So, I shut down rig#1 and stole a MSI 470 from it, plugged it into that x16 slot on rig #2 and powered up.  Unfortunately, I got the same problem of the machine not booting up.  However, out of the blue an idea came to my to try to plug the monitor into that x16 slot card and see if that did anything.  Amazingly, after making this switch, I was able to get it to boot with 6 GPUs for the first time.  Unfortunately, 2 problems (maybe the same problem but 2 symptoms) showed up. #1 was when it made it to the Windows login screen, it switched to that weird magnified view... you know, where the screen is showing like 3-4x zoomed in so you can't even see full windows that you open?  To me, this usually occurs when there is a video driver issue.  So, I check device manage and of course, this newest GPU is showing up with that error 43.  Slightly optimistic due to the fact that running the 6xGPU registry mod fixed this problem with GPU#5 on this machine, I ran it again, as Admin, but instead of fixing it, it momentarily showed all 6 GPUs with error 43, but then after reboot, 1-5 are ok again but #6 still shows error 43.  I tried it several times with the same result.

After about an hour of down-time on both rigs, I decide to return the GPU to it's home on rig#1, but encouraged that I could at least get it to boot up with 6 GPUs, I went ahead and ordered another 470 from Amazon to arrive Saturday. This way I can experiment on rig#2 without taking rig #1 down.

I think the first thing I probably need to try is using the GPU driver cleaner that Citronick recommended, them hopefully running 6xGPU mod again will get that 6th GPU working.  I also had a thought that maybe putting a dummy HMDI plug in that x16 GPU could help, meaning keep my monitor plugged into PCIe slot #1 (x1) and using the dummy plug in #2 (x16) just to get it to boot.

Anyway, seems like a little progress and have some things in mind to try, but other suggestions would be welcomed.

hmmm strange... is that screen-shot for 1 RX-480?  not sure what your electricity rates are or your exact build or consumption but I can't get anywhere close to those results on that site with either default values for 5x470 or 5x480, or even my own actual numbers from a 5x470 modded bios rig.

5x480 default @ .11 cents:  XMR = 4.06 profit / ETH = 3.45 profit / ZEC = 3.13 profit
5x470 default @ .11 cents:  XMR = 3.46 profit / ETH = 3.10 profit / ZEC = 2.50 profit
MY 5x470 mod @ 11 cents:  XMR = 3.46 profit / ETH = 3.04 profit / ZEC = 2.41 profit (actually very close to defaults on this site)

I know this is very volotile and rates and difficulties change by the minute but over the past 4-5 days I have not once seen ZEC = or above the other 2 and I have been checking this site as well as CoinWarez.  I did have a bit of an issue with XMR not coming in like the calculators said, but I think that was more of a pool problem. Since changing over to dwarfpool, actual earnings appear to be in line with these estimates.  Don't get me wrong, I hope people keep mining ZEC like crazy so as to hopefully keep the diff lower on XMR and ETH, but I had just assumed that most of those people were the ones with older GPUs that do better on ZEC.  I thought most people with 470/480 had already moved back to XMR or ETH.

how exactly do you figure XMR is useless with RX cards lol?  In reality, XMR is actually the most profitable (currently 20% better than ETH and nearly 50% better than ZEC) for the last several days.  And that is on my 5-470 un-modded rig!  After I strap-mod those puppies, the profitability margin of XMR over ZEC will only improve for me (all else being equal of course).  Not only is XMR currently the most profitable to mine with XMR, it's probably the only profitable CPU mineable coin at present.  I have 3 i7's getting close to 900 h/s on XMR... so basically the equivalent of 1.5 more 470's. 

ah ok... sounds like default -h of 1024 is the way to go then.  and 735 for a modded 480 seems about right compared to the 690 i got from my modded 470... in the ballpark at least.  What I can't figure out is how 800 is possible on a 470 as Wolf mentioned in this or another thread.  But I dont remember seeing anyone else claiming anywhere close to that high so who knows.  Will be interesting to see what kind of bump the 4xx series gets from Claymore's next rev of this miner.  Unfortunately, I think it will be delayed a bit as he mentioned on the ZEC thread that he finally caved and will be doing a linux miner for ZEC after all. Not sure why the demand to ZEC miners is still so high given that both XMR and ETH are currently considerably more profitable to mine. I guess a lot of people still hoping ZEC prices will go through the roof.

Has anyone using MoneroPool.com seen any productivity issues the last few days?  I was going along fine at pretty close to their own  profit calculator as well as the major external ones, then about 3 days ago it dropped dramatically. For instance, my last 2 payouts of .5 XMR took 30 hours each (.4 XMR/day).  My previous 5 payouts were averaging .67 XMR/day, which is very close to the calculators but now it has dropped almost 40%.  Yes, I understand that difficulty went up a bit the last few days but now it has actually dropped back down again.  Has anyone else had this issue on this pool?  I'm highly considering trying dwarfoool since this really seems to be a pool issue.

Yes, I would appreciate that as well.  In addition, would you mind sharing your "-a" and "-h" settings within Claymore .bat file?  Since he made this before the 470/480's came out, they are not mentioned anywhere in the readme file.  I think that -a 2 is probably best but anything I do with the -h value seems to reduce the hash-rate of my 470's. I assume that -h value will at least be similar for 480s and 470s.

Hotmetal is absolutely correct. Just watching the miner results window can be deceiving so ignore that. If you read the readme file, you will see that enabling the nofee option will reduce your speed by 5%, so what you have done is not only cut Claymore out of the fee he deserves and you are netting 2.5% less overall, so a lose-lose situation.

FYI: Claymore ZEC 9.2 was just released. Looks like 470's capable of close to 220-230 now, which, coupled with ZEC getting close to $40 again, may push some (myself included) off of ETH and back to ZEC.  XMR still showing more profitable than both for me but I may consider switching my ETH rig over to ZEC if I can really get close to 225 per GPU.

Would someone be so kind as to post recommended optimal "-h" settings for RX470 4GB GPUs?  I am running a 5-GPU (not bios modded yet) rig of them with "-a 2".  Default -h setting of 1024 gets me about 2860 total hash-rate (~570 per GPU).  However, when I tried increasing that to 1280 last night, it actually resulted in a lower hash-rate of about 2500.  Is there a sweet spot for these cards?  I also have 1 8GB 470 on this rig. should I specify a higher -h value for it vs the 4GB cards?

Also, I think it was Wolf that mentioned 470's are capable of close to 800 H/s on Monero.  Would you mind sharing how you were able to achieve this?  On my other rig with 4x470 4GB (bios strap-1500 modded), I am able to get 690 H/s per card with under-volting, and core/clock settings of 1150/1950.  These cards cannot really be pushed past 2000 so there must be something else to do to get that extra ~100 per card.  Maybe it's the question I asked above about -h value?  If so, would the recommended -h value for strap-modded GPUs be any different that un-modded ones?

glad it worked   My 4790k is on default, which is equal to -t 4, and I am getting about 305 or so, so same ballpark as you.  At least on all of my PCs, you will gain slightly some hash-rate over several hours, so I would think you will also end up over 300 if you leave it on for quite a while.  I have not experimented with CPU over-clocking but I would think at best it might only improve h/r modestly.  I am already heavily taxing my systems elsewhere so don't want to push the CPU too much.

Well yeah, obviously I could pull that orphan Sapphire as you mention, but I would rather avoid that if possible for a couple of reasons.  Firstly, I had a hell of a time getting 5 GPUs working on this rig so aside from eventually trying to get a 6th working in that x16 slot, I'm not eager to start unplugging and re-plugging cards in case it screws things up again.

The other reason is this... I am even unsure how pulling that card would affect my device numbers? For instance, will windows re-number the GPUs after reboot only based on the now 4 that are plugged in?  meaning they would then be 0, 1, 2 & 3? 

Is windows really so lame that it doesn't have a utility or easy way to ID which GPU it assigned which device number?  It seems more likely that there is a way, I just can't figure it out.

So, I am getting ready to strap-mod the bios on my 2nd mining rig. However, I realized that I need to know the device numbers of each of the 5 GPUs in order to do this, since the actual bios flash command line specifies the device to flash.

This was not a concern on rig#1 because this has 4 identical MSI 470s in it.  However, on rig #2 I have 1 sapphire nitro 470 8GB in the mix, so I need to make sure I am flashing the correct cards.  It may actually be that that sapphire takes the same strap-mod (more on that later) but i really need to know how to identify which device number windows as assigned to each of my 5 GPUs.

If someone could point me to a resource of education on this or suggest a way to easily identify them on Windows 7 I would appreciate it.

For instance, what determines which device # is assigned? Is it the order you install the cards? Or the PCIe slot they are plugged into? Or, is it something else?

The PCIe slots on my ASRock BTC R2 Mobo are numbered 1-6 from right to left, with the lone x16 slot (currently unoccupied) being slot #2.  However, when I go into device manager and check the properties for each GPU, it shows only PCI Bus 2-6 for the 5 GPUs, meaning it shows nothing as PCI Bus 1, but I really don't know how that correlates anyway since I know the device #s I am looking for begin with 0, so am assuming i should be looking for device #s 0-5, not 1-6.  FYI, my lone Sapphire is plugged into PCIe slot #4 (3rd installed slot from the right) and I am pretty sure (but not 100% confident) that it was also the 3rd GPU I installed.  Should this mean that this is device #2 (after 0 and 1)?  or, is it not that straight forward. Thanks in advance for any suggestions.

To Phil and anyone else about to or considering putting Claymore's CPU miner to work, here are some findings from my 3 PCs over the past few days:

- i7 6800K (6 cores, 12 threads, 15MB Cache):  default settings (6 threads) results in 313 H/s.  adding a "-t 7" to use 7 threads instead of 6 results in 347 H/s.  -t 7 appears to be the optimal setting for this CPU

- i7 4790K (4 cores, 8 threads, 8MB Cache): default setting of 4 threads results in about 307 H/r, which appears to be the optimal setting for this CPU

-i7 2600K ((4 cores, 8 threads, 8MB Cache): default setting of 4 threads results in about 190 H/r but note that this is in slow mode. I need to tweak a few things but since this is my main PC and I have other things running on it, I haven't had a chance to tweak and reboot yet. But, I think in fast mode this will get in the mid 200's, so about 230-240 maybe. I will update in a few days when I have more data

The main thing I wanted to point out is that my tests confirmed the statements I have seen on other threads about this CPU miner being very reliant on Cache... meaning, each thread you have mining required 2MB of cache to work effectively. Case in point on the top 2 above, but especially on the 6800k.  Increasing from 6 to 7 threads (requiring 14MB of Cache vs 12MB) resulted in an increase of about 35 H/s.  BUT... when I tried to use 8 threads, the H/r actually dipped back down around the 300 mark. Basically, it not only doesn't increase the rate over 7 threads but it reduces it.

Another thing to check when running this miner is the mode (slow vs fast). To enable fast mode, you need to run the EXE as Admin.  The readme says you only have to do this once, then after reboot it will run in fast mode without running as Admin, but I have had mixed results with this, so just always run it as Admin with no ill effects.

The one thing I do find interesting are the results between the 6800 & 4790.  The 6800 gets max 347 using 7 threads (about 50 per thread) but the 4790, which is 2 generations older, gets about 75 hashes per thread. Not sure why this is or whether to be happy with the 4790 or disappointed with the 6800 lol.

I concur with those 470 speeds as my modded rig does about 690 per RX 470 and my new rig that I have nodded modded yet is getting about 575 per GPU.  Mode "-a 2" seems to seems to the best choice on both of my rigs. I can get maybe 2/3 hashes more per card using "a -4" but it also consumes a bit more wattage so not worth it for the nominal hash gain.

What I am very curious about is the -h command.  I have only used 1024 as an experiment but it didn't really change any visible hashrates within the miner.  I do not fully understand the technical aspect of this command so would appreciate some insight from those that do.  I am trying to figure out the optimal -h command for my GPUs, which are:

-4x MSI RX 470 Gaming 4GB
-1x Sapphire Nitro RX 470 8GB

Also, am I right in assuming that the nitro could be pushed farther since it has twice the memory of the MSI's?




 

Thanks Claymore! Good to know this is on your radar.

btw, I believe Claymore meant updating ZEC (stand-alone) GPU miner and was probably referring to his ETH + DCR, LBRY, etc dual miner since there is currently no ZEC dual miner.  Looking forward to the update of all 3. 

You need to set the NsCpuCNMiner64.exe file to Run As Admin (right click-->properties-->compatibility, then check the Run as Admin box and Apply). The readme says this only needs to be done once but I have had issues with 1 of my 3 PCs so i just leave it always on run as admin with no issues.  This should get you into fast mode with scfg: 1 right away but if not, try rebooting then do it again.  I have that exact same CPU in my 2nd mining rig and am getting about 310 H/s on XMR, so you should see a bump of 50-60 hashes once in fast mode.

Yeah, I just ran my actual numbers through that calculator and was shocked to see ZEC only about 15-20% less profitable to mine than both ETH & XMR, both of which show exact profit of $2.39/day on my 4x470 strop-modded over-clocked under-volted rig.  Either that site is whack or something major changed with difficulty, because that gap in profitability actually narrowed vs the last few days despite ZEC falling another ~$4 in the last ~24 hours.  Yes, ETH and XMR are down a bit but not to the extent ZEC is, so must be something with difficulty, probably driven by more and more people changing from mining ZEC to 1 or both of those 2 coins.

Personally, I still have my 4-card rig on ETH and 5-card rig on XMR, but spent some time experimenting with both CPU & GPU XMR mining today.  This is probably known already but what i found is that the strap-mod for the MSI 470s is definately much more important for XMR than it is for ZEC.  I can only get my non-modded 470s to about 580 H/s on XMR, while my modded ones did 690 straight away on a test run over from ETH for about 2 hours.  So tomorrrow I will likely take a crack at getting those other 5 470s bios modded as that will likely get me about the equivalent of another GPU's hash-rate on XMR.

One of the cool things about XMR is that it can still be mined effectively with CPUs.  I have 3 CPUs on it now and am getting close to 900 H/s total, which amazingly pencils out to a better hash per watt rate than even the modded 470's.  This is actually a good thing for those of you (I think there were a few on this thread) that bought/build some higher-end CPU-based rigs when ZEC was profitable to CPU mine in the first couple of weeks.  My experience on all 3 PC has been excellent running Claymore's XMR CPU miner as in most cases, you will only use half your available threads, leaving plenty of power to do other things.  In fact, my 6800k-based rig is capable of running 4x470's on ETH (110 H/s), 7 CPU threads on XMR (350 H/s), 10GB HDD mining Burst, & 4 threads dedicated to writing new Burst plots... all simultaneously.  I guess that's the benefit of having 12 threads with 15MB of cache.

Monitoring them, and a certain degree of managing them as well, is very easy via the Chrome Remote Desktop app.  You simply need to install the app on your main (monitoring) PC, enable remote management on your target PCs, and have Chrome installed on all. They you set a pin for security purposes and you are all set!  I have been using this for about 2 months and it works great. You can essentially log on and control the PC 99% like you were in front of it. The only thing you cannot do of course is a hard reboot if your target machine crashes and hangs but other than that, it is an amazing and very easy to install and use tool.  I tried getting Windows Remote to work prior to finding this and it was a huge pain to try to configure and I never could get it to work outside of my own home network.

300 for 480s?  aren't they at about 200-220 now?  Everything i have seen and heard indicates that there are future improvements possible but certainly not anything in the range of 50% or more.  Unless Claymore has had a major breakthrough in the last few days, I thought he was talking maybe in the 10% range over 9.1.  Once again, unless ZEC starts going up instead of down, it seems fruitless to try to suck 5-10% hashrate more out of a coin that is already about half as profitable as both XMR and ETH.  I think only Claymore himself probably knows which one of his miners has more room for improvement between XMR 9.5 and ZEC 9.1.  But most certainly, if ZEC continues its decline, it won't matter if he can even get 20-25% more hashrate out of it since it will barely be (if at all) profitable to mine.